Dual image sensor image processing system and method

[0001] The present disclosure relates generally to digital imaging devices and, more particularly, to systems and method for processing image data obtained using an image sensor of a digital imaging device.

[0002] This section is intended to introduce the reader to various aspects of art that may be related to various aspects of the present techniques, which are described and/or claimed below. This discussion is believed to be helpful in providing the reader with background information to facilitate a better understanding of the various aspects of the present disclosure. Accordingly, it should be understood that these statements are to be read in this light, and not as admissions of prior art.

[0003] In recent years, digital imaging devices have become increasing popular due, at least in part, to such devices becoming more and more affordable for the average consumer. Further, in addition to a number of stand-alone digital cameras currently available on the market, it is not uncommon for digital imaging devices to be integrated as part of another electronic device, such as a desktop or notebook computer, a cellular phone, or a portable media player.

[0004] To acquire image data, most digital imaging devices include an image sensor that provides a number of light-detecting elements (e.g., photodetectors) configured to convert light detected by the image sensor into an electrical signal. An image sensor may also include a color filter array that filters light captured by the image sensor to capture color information. The image data captured by the image sensor may then be processed by an image processing pipeline, which may apply a number of various image processing operations to the image data to generate a full color image that may be displayed for viewing on a display device, such as a monitor.

[0005] While conventional image processing techniques generally aim to produce a viewable image that is both objectively and subjectively pleasing to a viewer, such conventional techniques may not adequately address errors and/or distortions in the image data introduced by the imaging device and/or the image sensor. For instance, defective pixels on the image sensor, which may be due to manufacturing defects or operational failure, may fail to sense light levels accurately and, if not corrected, may manifest as artifacts appearing in the resulting processed image. Additionally, light intensity fall-off at the edges of the image sensor, which may be due to imperfections in the manufacture of the lens, may adversely affect characterization measurements and may result in an image in which the overall light intensity is non-uniform. The image processing pipeline may also perform one or more processes to sharpen the image. Conventional sharpening techniques, however, may not adequately account for existing noise in the image signal, or may be unable to distinguish the noise from edges and textured areas in the image. In such instances, conventional sharpening techniques may actually increase the appearance of noise in the image, which is generally undesirable.

[0006] Another image processing operation that may be applied to the image data captured by the image sensor is a demosaicing operation. Because the color filter array generally provides color data at one wavelength per sensor pixel, a full set of color data is generally interpolated for each color channel in order to reproduce a full color image (e.g., RGB image). Conventional demosaicing techniques generally interpolate values for the missing color data in a horizontal or a vertical direction, generally depending on some type of fixed threshold. However, such conventional demosaicing techniques may not adequately account for the locations and direction of edges within the image, which may result in edge artifacts, such as aliasing, checkerboard artifacts, or rainbow artifacts, being introduced into the full color image, particularly along diagonal edges within the image.

[0007] Accordingly, various considerations should be addressed when processing a digital image obtained with a digital camera or other imaging device in order to improve the appearance of the resulting image. In particular, certain aspects of the disclosure below may address one or more of the drawbacks briefly mentioned above.

[0008] A summary of certain embodiments disclosed herein is set forth below. It should be understood that these aspects are presented merely to provide the reader with a brief summary of these certain embodiments and that these aspects are not intended to limit the scope of this disclosure. Indeed, this disclosure may encompass a variety of aspects that may not be set forth below.

[0009] The present disclosure provides various techniques for processing image data acquired using a digital image sensor. In accordance with aspects of the present disclosure, one such technique may relate to the processing of image data in a system that supports multiple image sensors. In one embodiment, the image processing system may include control circuitry configured to determine whether a device is operating in a single sensor mode (one active sensor) or a dual sensor mode (two active sensors). When operating in the single sensor mode, data may be provided directly to a front-end pixel processing unit from the sensor interface of the active sensor. When operating in a dual sensor mode, the image frames from the first and second sensors are provided to the front-end pixel processing unit in an interleaved manner. For instance, in one embodiment, the image frames from the first and second sensors are written to a memory, and then read out to the front-end pixel processing unit in an interleaved manner.

[0010] Various refinements of the features noted above may exist in relation to various aspects of the present disclosure. Further features may also be incorporated in these various aspects as well. These refinements and additional features may exist individually or in any combination. For instance, various features discussed below in relation to one or more of the illustrated embodiments may be incorporated into any of the above-described aspects of the present disclosure alone or in any combination. Again, the brief summary presented above is intended only to familiarize the reader with certain aspects and contexts of embodiments of the present disclosure without limitation to the claimed subject matter.

[0085] One or more specific embodiments of the present disclosure will be described below. These described embodiments are only examples of the presently disclosed techniques. Additionally, in an effort to provide a concise description of these embodiments, all features of an actual implementation may not be described in the specification. It should be appreciated that in the development of any such actual implementation, as in any engineering or design project, numerous implementation-specific decisions must be made to achieve the developers' specific goals, such as compliance with system-related and business-related constraints, which may vary from one implementation to another. Moreover, it should be appreciated that such a development effort might be complex and time consuming, but would nevertheless be a routine undertaking of design, fabrication, and manufacture for those of ordinary skill having the benefit of this disclosure.

[0086] When introducing elements of various embodiments of the present disclosure, the articles “a,” “an,” and “the” are intended to mean that there are one or more of the elements. The terms “comprising,” “including,” and “having” are intended to be inclusive and mean that there may be additional elements other than the listed elements. Additionally, it should be understood that references to “one embodiment” or “an embodiment” of the present disclosure are not intended to be interpreted as excluding the existence of additional embodiments that also incorporate the recited features.

[0087] As will be discussed below, the present disclosure relates generally to techniques for processing image data acquired via one or more image sensing devices. In particular, certain aspects of the present disclosure may relate to techniques for detecting and correcting defective pixels, techniques for demosaicing a raw image pattern, techniques for sharpening a luminance image using a multi-scale unsharp mask, and techniques for applying lens shading gains to correct for lens shading irregularities. Further, it should be understood that the presently disclosed techniques may be applied to both still images and moving images (e.g., video), and may be utilized in any suitable type of imaging application, such as a digital camera, an electronic device having an integrated digital camera, a security or video surveillance system, a medical imaging system, and so forth.

[0088] Keeping the above points in mind, FIG. 1 is a block diagram illustrating an example of an electronic device 10 that may provide for the processing of image data using one or more of the image processing techniques briefly mentioned above. The electronic device 10 may be any type of electronic device, such as a laptop or desktop computer, a mobile phone, a digital media player, or the like, that is configured to receive and process image data, such as data acquired using one or more image sensing components. By way of example only, the electronic device 10 may be a portable electronic device, such as a model of an iPod® or iPhone®, available from Apple Inc. of Cupertino, Calif. Additionally, the electronic device 10 may be a desktop or laptop computer, such as a model of a MacBook®, MacBook® Pro, MacBook Air®, iMac®, Mac® Mini, or Mac Pro®, available from Apple Inc. In other embodiments, electronic device 10 may also be a model of an electronic device from another manufacturer that is capable of acquiring and processing image data.

[0089] Regardless of its form (e.g., portable or non-portable), it should be understood that the electronic device 10 may provide for the processing of image data using one or more of the image processing techniques briefly discussed above, which may include defective pixel correction and/or detection techniques, lens shading correction techniques, demosaicing techniques, or image sharpening techniques, among others. In some embodiments, the electronic device 10 may apply such image processing techniques to image data stored in a memory of the electronic device 10. In further embodiments, the electronic device 10 may include one or more imaging devices, such as an integrated or external digital camera, configured to acquire image data, which may then be processed by the electronic device 10 using one or more of the above-mentioned image processing techniques. Embodiments showing both portable and non-portable embodiments of electronic device 10 will be further discussed below in FIGS. 3-6.

[0090] As shown in FIG. 1, the electronic device 10 may include various internal and/or external components which contribute to the function of the device 10. Those of ordinary skill in the art will appreciate that the various functional blocks shown in FIG. 1 may comprise hardware elements (including circuitry), software elements (including computer code stored on a computer-readable medium) or a combination of both hardware and software elements. For example, in the presently illustrated embodiment, the electronic device 10 may include input/output (I/O) ports 12, input structures 14, one or more processors 16, memory device 18, non-volatile storage 20, expansion card(s) 22, networking device 24, power source 26, and display 28. Additionally, the electronic device 10 may include one or more imaging devices 30, such as a digital camera, and image processing circuitry 32. As will be discussed further below, the image processing circuitry 32 may be configured implement one or more of the above-discussed image processing techniques when processing image data. As can be appreciated, image data processed by image processing circuitry 32 may be retrieved from the memory 18 and/or the non-volatile storage device(s) 20, or may be acquired using the imaging device 30.

[0091] Before continuing, it should be understood that the system block diagram of the device 10 shown in FIG. 1 is intended to be a high-level control diagram depicting various components that may be included in such a device 10. That is, the connection lines between each individual component shown in FIG. 1 may not necessarily represent paths or directions through which data flows or is transmitted between various components of the device 10. Indeed, as discussed below, the depicted processor(s) 16 may, in some embodiments, include multiple processors, such as a main processor (e.g., CPU), and dedicated image and/or video processors. In such embodiments, the processing of image data may be primarily handled by these dedicated processors, thus effectively offloading such tasks from a main processor (CPU).

[0092] With regard to each of the illustrated components in FIG. 1, the I/O ports 12 may include ports configured to connect to a variety of external devices, such as a power source, an audio output device (e.g., headset or headphones), or other electronic devices (such as handheld devices and/or computers, printers, projectors, external displays, modems, docking stations, and so forth). In one embodiment, the I/O ports 12 may be configured to connect to an external imaging device, such as a digital camera, for the acquisition of image data that may be processed using the image processing circuitry 32. The I/O ports 12 may support any suitable interface type, such as a universal serial bus (USB) port, a serial connection port, an IEEE-1394 (FireWire) port, an Ethernet or modem port, and/or an AC/DC power connection port.

[0093] In some embodiments, certain I/O ports 12 may be configured to provide for more than one function. For instance, in one embodiment, the I/O ports 12 may include a proprietary port from Apple Inc. that may function not only to facilitate the transfer of data between the electronic device 10 and an external source, but also to couple the device 10 to a power charging interface such as an power adapter designed to provide power from a electrical wall outlet, or an interface cable configured to draw power from another electrical device, such as a desktop or laptop computer, for charging the power source 26 (which may include one or more rechargeable batteries). Thus, the I/O port 12 may be configured to function dually as both a data transfer port and an AC/DC power connection port depending, for example, on the external component being coupled to the device 10 via the I/O port 12.

[0094] The input structures 14 may provide user input or feedback to the processor(s) 16. For instance, input structures 14 may be configured to control one or more functions of electronic device 10, such as applications running on electronic device 10. By way of example only, input structures 14 may include buttons, sliders, switches, control pads, keys, knobs, scroll wheels, keyboards, mice, touchpads, and so forth, or some combination thereof. In one embodiment, input structures 14 may allow a user to navigate a graphical user interface (GUI) displayed on device 10. Additionally, input structures 14 may include a touch sensitive mechanism provided in conjunction with display 28. In such embodiments, a user may select or interact with displayed interface elements via the touch sensitive mechanism.

[0095] The input structures 14 may include the various devices, circuitry, and pathways by which user input or feedback is provided to one or more processors 16. Such input structures 14 may be configured to control a function of the device 10, applications running on the device 10, and/or any interfaces or devices connected to or used by the electronic device 10. For example, the input structures 14 may allow a user to navigate a displayed user interface or application interface. Examples of the input structures 14 may include buttons, sliders, switches, control pads, keys, knobs, scroll wheels, keyboards, mice, touchpads, and so forth.

[0096] In certain embodiments, an input structure 14 and the display device 28 may be provided together, such as in the case of a “touchscreen,” whereby a touch-sensitive mechanism is provided in conjunction with the display 28. In such embodiments, the user may select or interact with displayed interface elements via the touch-sensitive mechanism. In this way, the displayed interface may provide interactive functionality, allowing a user to navigate the displayed interface by touching the display 28. For example, user interaction with the input structures 14, such as to interact with a user or application interface displayed on the display 26, may generate electrical signals indicative of the user input. These input signals may be routed via suitable pathways, such as an input hub or data bus, to the one or more processors 16 for further processing.

[0097] In addition to processing various input signals received via the input structure(s) 14, the processor(s) 16 may control the general operation of the device 10. For instance, the processor(s) 16 may provide the processing capability to execute an operating system, programs, user and application interfaces, and any other functions of the electronic device 10. The processor(s) 16 may include one or more microprocessors, such as one or more “general-purpose” microprocessors, one or more special-purpose microprocessors and/or application-specific microprocessors (ASICs), or a combination of such processing components. For example, the processor(s) 16 may include one or more instruction set (e.g., RISC) processors, as well as graphics processors (GPU), video processors, audio processors and/or related chip sets. As will be appreciated, the processor(s) 16 may be coupled to one or more data buses for transferring data and instructions between various components of the device 10. In certain embodiments, the processor(s) 16 may provide the processing capability to execute an imaging applications on the electronic device 10, such as Photo Booth®, Aperture®, iPhoto®, or Preview®, available from Apple Inc., or the “Camera” and/or “Photo” applications provided by Apple Inc. and available on models of the iPhone®.

[0098] The instructions or data to be processed by the processor(s) 16 may be stored in a computer-readable medium, such as a memory device 18. The memory device 18 may be provided as a volatile memory, such as random access memory (RAM) or as a non-volatile memory, such as read-only memory (ROM), or as a combination of one or more RAM and ROM devices. The memory 18 may store a variety of information and may be used for various purposes. For example, the memory 18 may store firmware for the electronic device 10, such as a basic input/output system (BIOS), an operating system, various programs, applications, or any other routines that may be executed on the electronic device 10, including user interface functions, processor functions, and so forth. In addition, the memory 18 may be used for buffering or caching during operation of the electronic device 10. For instance, in one embodiment, the memory 18 include one or more frame buffers for buffering video data as it is being output to the display 28.

[0099] In addition to the memory device 18, the electronic device 10 may further include a non-volatile storage 20 for persistent storage of data and/or instructions. The non-volatile storage 20 may include flash memory, a hard drive, or any other optical, magnetic, and/or solid-state storage media, or some combination thereof. Thus, although depicted as a single device in FIG. 1 for purposes of clarity, it should understood that the non-volatile storage device(s) 20 may include a combination of one or more of the above-listed storage devices operating in conjunction with the processor(s) 16. The non-volatile storage 20 may be used to store firmware, data files, image data, software programs and applications, wireless connection information, personal information, user preferences, and any other suitable data. In accordance with aspects of the present disclosure, image data stored in the non-volatile storage 20 and/or the memory device 18 may be processed by the image processing circuitry 32 prior to being output on a display.

[0100] The embodiment illustrated in FIG. 1 may also include one or more card or expansion slots. The card slots may be configured to receive an expansion card 22 that may be used to add functionality, such as additional memory, I/O functionality, or networking capability, to the electronic device 10. Such an expansion card 22 may connect to the device through any type of suitable connector, and may be accessed internally or external with respect to a housing of the electronic device 10. For example, in one embodiment, the expansion card 24 may be flash memory card, such as a SecureDigital (SD) card, mini- or microSD, CompactFlash card, or the like, or may be a PCMCIA device. Additionally, the expansion card 24 may be a Subscriber Identity Module (SIM) card, for use with an embodiment of the electronic device 10 that provides mobile phone capability.

[0101] The electronic device 10 also includes the network device 24, which may be a network controller or a network interface card (NIC) that may provide for network connectivity over a wireless 802.11 standard or any other suitable networking standard, such as a local area network (LAN), a wide area network (WAN), such as an Enhanced Data Rates for GSM Evolution (EDGE) network, a 3G data network, or the Internet. In certain embodiments, the network device 24 may provide for a connection to an online digital media content provider, such as the iTunes® music service, available from Apple Inc.

[0102] The power source 26 of the device 10 may include the capability to power the device 10 in both non-portable and portable settings. For example, in a portable setting, the device 10 may include one or more batteries, such as a Li-Ion battery, for powering the device 10. The battery may be re-charged by connecting the device 10 to an external power source, such as to an electrical wall outlet. In a non-portable setting, the power source 26 may include a power supply unit (PSU) configured to draw power from an electrical wall outlet, and to distribute the power to various components of a non-portable electronic device, such as a desktop computing system.

[0103] The display 28 may be used to display various images generated by device 10, such as a GUI for an operating system, or image data (including still images and video data) processed by the image processing circuitry 32, as will be discussed further below. As mentioned above, the image data may include image data acquired using the imaging device 30 or image data retrieved from the memory 18 and/or non-volatile storage 20. The display 28 may be any suitable type of display, such as a liquid crystal display (LCD), plasma display, or an organic light emitting diode (OLED) display, for example. Additionally, as discussed above, the display 28 may be provided in conjunction with the above-discussed touch-sensitive mechanism (e.g., a touch screen) that may function as part of a control interface for the electronic device 10.

[0104] The illustrated imaging device(s) 30 may be provided as a digital camera configured to acquire both still images and moving images (e.g., video). The camera 30 may include a lens and one or more image sensors configured to capturing and converting light into electrical signals. By way of example only, the image sensor may include a CMOS image sensor (e.g., a CMOS active-pixel sensor (APS)) or a CCD (charge-coupled device) sensor. Generally, the image sensor in the camera 30 includes an integrated circuit having an array of pixels, wherein each pixel includes a photodetector for sensing light. As those skilled in the art will appreciate, the photodetectors in the imaging pixels generally detect the intensity of light captured via the camera lenses. However, photodetectors, by themselves, are generally unable to detect the wavelength of the captured light and, thus, are unable to determine color information.

[0105] Accordingly, the image sensor may further include a color filter array (CFA) that may overlay or be disposed over the pixel array of the image sensor to capture color information. The color filter array may include an array of small color filters, each of which may overlap a respective pixel of the image sensor and filter the captured light by wavelength. Thus, when used in conjunction, the color filter array and the photodetectors may provide both wavelength and intensity information with regard to light captured through the camera, which may be representative of a captured image.

[0106] In one embodiment, the color filter array may include a Bayer color filter array, which provides a filter pattern that is 50% green elements, 25% red elements, and 25% blue elements. For instance, FIG. 2 shows a 2×2 pixel block of a Bayer CFA includes 2 green elements (Gr and Gb), 1 red element (R), and 1 blue element (B). Thus, an image sensor that utilizes a Bayer color filter array may provide information regarding the intensity of the light received by the camera 30 at the green, red, and blue wavelengths, whereby each image pixel records only one of the three colors (RGB). This information, which may be referred to as “raw image data” or data in the “raw domain,” may then be processed using one or more demosaicing techniques to convert the raw image data into a full color image, generally by interpolating a set of red, green, and blue values for each pixel. As will be discussed further below, such demosaicing techniques may be performed by the image processing circuitry 32.

[0107] As mentioned above, the image processing circuitry 32 may provide for various image processing steps, such as defective pixel detection/correction, lens shading correction, demosaicing, and image sharpening, noise reduction, gamma correction, image enhancement, color-space conversion, image compression, chroma sub-sampling, and image scaling operations, and so forth. In some embodiments, the image processing circuitry 32 may include various subcomponents and/or discrete units of logic that collectively form an image processing “pipeline” for performing each of the various image processing steps. These subcomponents may be implemented using hardware (e.g., digital signal processors or ASICs) or software, or via a combination of hardware and software components. The various image processing operations that may be provided by the image processing circuitry 32 and, particularly those processing operations relating to defective pixel detection/correction, lens shading correction, demosaicing, and image sharpening, will be discussed in greater detail below.

[0108] Before continuing, it should be noted that while various embodiments of the various image processing techniques discussed below may utilize a Bayer CFA, the presently disclosed techniques are not intended to be limited in this regard. Indeed, those skilled in the art will appreciate that the image processing techniques provided herein may be applicable to any suitable type of color filter array, including RGBW filters, CYGM filters, and so forth.

[0109] Referring again to the electronic device 10, FIGS. 3-6 illustrate various forms that the electronic device 10 may take. As mentioned above, the electronic device 10 may take the form of a computer, including computers that are generally portable (such as laptop, notebook, and tablet computers) as well as computers that are generally non-portable (such as desktop computers, workstations and/or servers), or other type of electronic device, such as handheld portable electronic devices (e.g., digital media player or mobile phone). In particular, FIGS. 3 and 4 depict the electronic device 10 in the form of a laptop computer 40 and a desktop computer 50, respectively. FIGS. 5 and 6 show front and rear views, respectively, of the electronic device 10 in the form of a handheld portable device 60.

[0110] As shown in FIG. 3, the depicted laptop computer 40 includes a housing 42, the display 28, the I/O ports 12, and the input structures 14. The input structures 14 may include a keyboard and a touchpad mouse that are integrated with the housing 42. Additionally, the input structure 14 may include various other buttons and/or switches which may be used to interact with the computer 40, such as to power on or start the computer, to operate a GUI or an application running on the computer 40, as well as adjust various other aspects relating to operation of the computer 40 (e.g., sound volume, display brightness, etc.). The computer 40 may also include various I/O ports 12 that provide for connectivity to additional devices, as discussed above, such as a FireWire® or USB port, a high definition multimedia interface (HDMI) port, or any other type of port that is suitable for connecting to an external device. Additionally, the computer 40 may include network connectivity (e.g., network device 26), memory (e.g., memory 20), and storage capabilities (e.g., storage device 22), as described above with respect to FIG. 1.

[0111] Further, the laptop computer 40, in the illustrated embodiment, may include an integrated imaging device 30 (e.g., camera). In other embodiments, the laptop computer 40 may utilize an external camera (e.g., an external USB camera or a “webcam”) connected to one or more of the I/O ports 12 instead of or in addition to the integrated camera 30. For instance, an external camera may be an iSight® camera available from Apple Inc. The camera 30, whether integrated or external, may provide for the capture and recording of images. Such images may then be viewed by a user using an image viewing application, or may be utilized by other applications, including video-conferencing applications, such as iChat®, and image editing/viewing applications, such as Photo Booth®, Aperture®, iPhoto®, or Preview®, which are available from Apple Inc. In certain embodiments, the depicted laptop computer 40 may be a model of a MacBook®, MacBook® Pro, MacBook Air®, or PowerBook® available from Apple Inc. Additionally, the computer 40, in one embodiment, may be a portable tablet computing device, such as a model of an iPad® tablet computer, also available from Apple Inc.

[0112] FIG. 4 further illustrates an embodiment in which the electronic device 10 is provided as a desktop computer 50. As will be appreciated, the desktop computer 50 may include a number of features that may be generally similar to those provided by the laptop computer 40 shown in FIG. 4, but may have a generally larger overall form factor. As shown, the desktop computer 50 may be housed in an enclosure 42 that includes the display 28, as well as various other components discussed above with regard to the block diagram shown in FIG. 1. Further, the desktop computer 50 may include an external keyboard and mouse (input structures 14) that may be coupled to the computer 50 via one or more I/O ports 12 (e.g., USB) or may communicate with the computer 50 wirelessly (e.g., RF, Bluetooth, etc.). The desktop computer 50 also includes an imaging device 30, which may be an integrated or external camera, as discussed above. In certain embodiments, the depicted desktop computer 50 may be a model of an iMac®, Mac® mini, or Mac Pro®, available from Apple Inc.

[0113] As further shown, the display 28 may be configured to generate various images that may be viewed by a user. For example, during operation of the computer 50, the display 28 may display a graphical user interface (“GUI”) 52 that allows the user to interact with an operating system and/or application running on the computer 50. The GUI 52 may include various layers, windows, screens, templates, or other graphical elements that may be displayed in all, or a portion, of the display device 28. For instance, in the depicted embodiment, an operating system GUI 52 may include various graphical icons 54, each of which may correspond to various applications that may be opened or executed upon detecting a user selection (e.g., via keyboard/mouse or touchscreen input). The icons 54 may be displayed in a dock 56 or within one or more graphical window elements 58 displayed on the screen. In some embodiments, the selection of an icon 54 may lead to a hierarchical navigation process, such that selection of an icon 54 leads to a screen or opens another graphical window that includes one or more additional icons or other GUI elements. By way of example only, the operating system GUI 52 displayed in FIG. 4 may be from a version of the Mac OS® operating system, available from Apple Inc.

[0114] Continuing to FIGS. 5 and 6, the electronic device 10 is further illustrated in the form of portable handheld electronic device 60, which may be a model of an iPod® or iPhone® available from Apple Inc. In the depicted embodiment, the handheld device 60 includes an enclosure 42, which may function to protect the interior components from physical damage and to shield them from electromagnetic interference. The enclosure 42 may be formed from any suitable material or combination of materials, such as plastic, metal, or a composite material, and may allow certain frequencies of electromagnetic radiation, such as wireless networking signals, to pass through to wireless communication circuitry (e.g., network device 24), which may be disposed within the enclosure 42, as shown in FIG. 5.

[0115] The enclosure 42 also includes various user input structures 14 through which a user may interface with the handheld device 60. For instance, each input structure 14 may be configured to control one or more respective device functions when pressed or actuated. By way of example, one or more of the input structures 14 may be configured to invoke a “home” screen 42 or menu to be displayed, to toggle between a sleep, wake, or powered on/off mode, to silence a ringer for a cellular phone application, to increase or decrease a volume output, and so forth. It should be understood that the illustrated input structures 14 are merely exemplary, and that the handheld device 60 may include any number of suitable user input structures existing in various forms including buttons, switches, keys, knobs, scroll wheels, and so forth.

[0116] As shown in FIG. 5, the handheld device 60 may include various I/O ports 12. For instance, the depicted I/O ports 12 may include a proprietary connection port 12a for transmitting and receiving data files or for charging a power source 26 and an audio connection port 12b for connecting the device 60 to an audio output device (e.g., headphones or speakers). Further, in embodiments where the handheld device 60 provides mobile phone functionality, the device 60 may include an I/O port 12c for receiving a subscriber identify module (SIM) card (e.g., an expansion card 22).

[0117] The display device 28, which may be an LCD, OLED, or any suitable type of display, may display various images generated by the handheld device 60. For example, the display 28 may display various system indicators 64 providing feedback to a user with regard to one or more states of handheld device 60, such as power status, signal strength, external device connections, and so forth. The display may also display a GUI 52 that allows a user to interact with the device 60, as discussed above with reference to FIG. 4. The GUI 52 may include graphical elements, such as the icons 54 which may correspond to various applications that may be opened or executed upon detecting a user selection of a respective icon 54. By way of example, one of the icons 54 may represent a camera application 66 that may be used in conjunction with a camera 30 (shown in phantom lines in FIG. 5) for acquiring images. Referring briefly to FIG. 6, a rear view of the handheld electronic device 60 depicted in FIG. 5 is illustrated, which shows the camera 30 as being integrated with the housing 42 and positioned on the rear of the handheld device 60.

[0118] As mentioned above, image data acquired using the camera 30 may be processed using the image processing circuitry 32, which my include hardware (e.g., disposed within the enclosure 42) and/or software stored on one or more storage devices (e.g., memory 18 or non-volatile storage 20) of the device 60. Images acquired using the camera application 66 and the camera 30 may be stored on the device 60 (e.g., in storage device 20) and may be viewed at a later time using a photo viewing application 68.

[0119] The handheld device 60 may also include various audio input and output elements. For example, the audio input/output elements, depicted generally by reference numeral 70, may include an input receiver, such as one or more microphones. For instance, where the handheld device 60 includes cell phone functionality, the input receivers may be configured to receive user audio input, such as a user's voice. Additionally, the audio input/output elements 70 may include one or more output transmitters. Such output transmitters may include one or more speakers which may function to transmit audio signals to a user, such as during the playback of music data using a media player application 72. Further, in embodiments where the handheld device 60 includes a cell phone application, an additional audio output transmitter 74 may be provided, as shown in FIG. 5. Like the output transmitters of the audio input/output elements 70, the output transmitter 74 may also include one or more speakers configured to transmit audio signals to a user, such as voice data received during a telephone call. Thus, the audio input/output elements 70 and 74 may operate in conjunction to function as the audio receiving and transmitting elements of a telephone.

[0120] Having now provided some context with regard to various forms that the electronic device 10 may take, the present discussion will now focus on the image processing circuitry 32 depicted in FIG. 1. As mentioned above, the image processing circuitry 32 may be implemented using hardware and/or software components, and may include various processing units that define an image signal processing (ISP) pipeline. In particular, the following discussion may focus on aspects of the image processing techniques set forth in the present disclosure, particularly those relating to defective pixel detection/correction techniques, lens shading correction techniques, demosaicing techniques, and image sharpening techniques.

[0121] Referring now to FIG. 7, a simplified top-level block diagram depicting several functional components that may be implemented as part of the image processing circuitry 32 is illustrated, in accordance with one embodiment of the presently disclosed techniques. Particularly, FIG. 7 is intended to illustrate how image data may flow through the image processing circuitry 32, in accordance with at least one embodiment. In order to provide a general overview of the image processing circuitry 32, a general description of how these functional components operate to process image data is provided here with reference to FIG. 7, while a more specific description of each of the illustrated functional components, as well as their respective sub-components, will be further provided below.

[0122] Referring to the illustrated embodiment, the image processing circuitry 32 may include image signal processing (ISP) front-end processing logic 80, ISP pipe processing logic 82, and control logic 84. Image data captured by the imaging device 30 may first be processed by the ISP front-end logic 80 and analyzed to capture image statistics that may be used to determine one or more control parameters for the ISP pipe logic 82 and/or the imaging device 30. The ISP front-end logic 80 may be configured to capture image data from an image sensor input signal. For instance, as shown in FIG. 7, the imaging device 30 may include a camera having one or more lenses 88 and image sensor(s) 90. As discussed above, the image sensor(s) 90 may include a color filter array (e.g., a Bayer filter) and may thus provide both light intensity and wavelength information captured by each imaging pixel of the image sensors 90 to provide for a set of raw image data that may be processed by the ISP front-end logic 80. For instance, the output 92 from the imaging device 30 may be received by a sensor interface 94, which may then provide the raw image data 96 to the ISP front-end logic 80 based, for example, on the sensor interface type. By way of example, the sensor interface 94 may utilize a Standard Mobile Imaging Architecture (SMIA) interface or other serial or parallel camera interfaces, or some combination thereof. In certain embodiments, the ISP front-end logic 80 may operate within its own clock domain and may provide an asynchronous interface to the sensor interface 94 to support image sensors of different sizes and timing requirements.

[0123] The raw image data 96 may be provided to the ISP front-end logic 80 and processed on a pixel-by-pixel basis in a number of formats. For instance, each image pixel may have a bit-depth of 8, 10, 12, or 14 bits. The ISP front-end logic 80 may perform one or more image processing operations on the raw image data 96, as well as collect statistics about the image data 96. The image processing operations, as well as the collection of statistical data, may be performed at the same or at different bit-depth precisions. For example, in one embodiment, processing of the raw image pixel data 96 may be performed at a precision of 14-bits. In such embodiments, raw pixel data received by the ISP front-end logic 80 that has a bit-depth of less than 14 bits (e.g., 8-bit, 10-bit, 12-bit) may be up-sampled to 14-bits for image processing purposes. In another embodiment, statistical processing may occur at a precision of 8-bits and, thus, raw pixel data having a higher bit-depth may be down-sampled to an 8-bit format for statistics purposes. As will be appreciated, down-sampling to 8-bits may reduce hardware size (e.g., area) and also reduce processing/computational complexity for the statistics data. Additionally, the raw image data may be averaged spatially to allow for the statistics data to be more robust to noise.

[0124] Further, as shown in FIG. 7, the ISP front-end logic 80 may also receive pixel data from the memory 108. For instance, as shown by reference number 98, the raw pixel data may be sent to the memory 108 from the sensor interface 94. The raw pixel data residing in the memory 108 may then be provided to the ISP front-end logic 80 for processing, as indicated by reference number 100. The memory 108 may be part of the memory device 18, the storage device 20, or may be a separate dedicated memory within the electronic device 10 and may include direct memory access (DMA) features. Further, in certain embodiments, the ISP front-end logic 80 may operate within its own clock domain and provide an asynchronous interface to the sensor interface 94 to support sensors of different sizes and having different timing requirements.

[0125] Upon receiving the raw image data 96 (from sensor interface 94) or 100 (from memory 108), the ISP front-end logic 80 may perform one or more image processing operations, such as temporal filtering and/or binning compensation filtering. The processed image data may then be provided to the ISP pipe logic 82 (output signal 109) for additional processing prior to being displayed (e.g., on display device 28), or may be sent to the memory (output signal 110). The ISP pipe logic 82 receives the “front-end” processed data, either directly form the ISP front-end logic 80 or from the memory 108 (input signal 112), and may provide for additional processing of the image data in the raw domain, as well as in the RGB and YCbCr color spaces. Image data processed by the ISP pipe logic 82 may then be output (signal 114) to the display 28 for viewing by a user and/or may be further processed by a graphics engine or GPU. Additionally, output from the ISP pipe logic 82 may be sent to memory 108 (signal 115) and the display 28 may read the image data from memory 108 (signal 116), which may, in certain embodiments, be configured to implement one or more frame buffers. Further, in some implementations, the output of the ISP pipe logic 82 may also be provided to a compression/decompression engine 118 (signal 117) for encoding/decoding the image data. The encoded image data may be stored and then later decompressed prior to being displayed on the display 28 device (signal 119). By way of example, the compression engine or “encoder” 118 may be a JPEG compression engine for encoding still images, or an H.264 compression engine for encoding video images, or some combination thereof, as well as a corresponding decompression engine for decoding the image data. Additional information with regard to image processing operations that may be provided in the ISP pipe logic 82 will be discussed in greater detail below with regard to FIGS. 49-79. Also, it should be noted that the ISP pipe logic 82 may also receive raw image data from the memory 108, as depicted by input signal 112.

[0126] Statistical data 102 determined by the ISP front-end logic 80 may be provided to a control logic unit 84. The statistical data 102 may include, for example, image sensor statistics relating to auto-exposure, auto-white balance, auto-focus, flicker detection, black level compensation (BLC), lens shading correction, and so forth. The control logic 84 may include a processor and/or microcontroller configured to execute one or more routines (e.g., firmware) that may be configured to determine, based upon the received statistical data 102, control parameters 104 for the imaging device 30, as well as control parameters 106 for the ISP pipe processing logic 82. By way of example only, the control parameters 104 may include sensor control parameters (e.g., gains, integration time for exposure control), camera flash control parameters, lens control parameters (e.g., focal length for focusing or zoom), or a combination of such parameters. The ISP control parameters 106 may include gain levels and color correction matrix (CCM) coefficients for auto-white balance and color adjustment (e.g., during RGB processing), as well as lens shading correction parameters which, as discussed below, may be determined based upon white point balance parameters. In some embodiments, the control logic 84 may, in addition to analyzing statistics data 102, also analyze historical statistics, which may be stored on the electronic device 10 (e.g., in memory 18 or storage 20).

[0127] Due to the generally complex design of the image processing circuitry 32 shown herein, it may be beneficial to separate the discussion of the ISP front-end logic 80 and the ISP pipe processing logic 82 into separate sections, as shown below. Particularly, FIGS. 8 to 48 of the present application may relate to the discussion of various embodiments and aspects of the ISP front-end logic 80, while FIGS. 49 to 79 of the present application may relate to the discussion of various embodiments and aspects of the ISP pipe processing logic 82.

[0128] FIG. 8 is a more detailed block diagram showing functional logic blocks that may be implemented in the ISP front-end logic 80, in accordance with one embodiment. Depending on the configuration of the imaging device 30 and/or sensor interface 94, as discussed above in FIG. 7, raw image data may be provided to the ISP front-end logic 80 by one or more image sensors 90. In the depicted embodiment, raw image data may be provided to the ISP front-end logic 80 by a first image sensor 90a (Sensor0) and a second image sensor 90b (Sensor1). As will be discussed further below, each image sensor 90a and 90b may be configured to apply binning to full resolution image data in order to increase signal-to-noise ratio of the image signal. For instance, a binning technique, such as 2×2 binning, may be applied which may interpolate a “binned” raw image pixel based upon four full-resolution image pixels of the same color. In one embodiment, this may result in there being four accumulated signal components associated with the binned pixel versus a single noise component, thus improving signal-to-noise of the image data, but reducing overall resolution. Additionally, binning may also result in an uneven or non-uniform spatial sampling of the image data, which may be corrected using binning compensation filtering, as will be discussed in more detail below.

[0129] As shown, the image sensors 90a and 90b may provide the raw image data as signals Sif0, and Sif1, respectively. Each of the image sensors 90a and 90b may be generally associated with the respective statistics processing units 120 (StatsPipe0) and 122 (StatsPipe1), which may be configured to process image data for the determination of one or more sets of statistics (as indicated by signals Stats0 and Stats1), including statistics relating to auto-exposure, auto-white balance, auto-focus, flicker detection, black level compensation, and lens shading correction, and so forth. In certain embodiments, when only one of the sensors 90a or 90b is actively acquiring image, the image data may be sent to both StatsPipe0 and StatsPipe1 if additional statistics are desired. For instance, to provide one example, if StatsPipe0 and StatsPipe1 are both available, StatsPipe0 may be utilized to collect statistics for one color space (e.g., RGB), and StatsPipe1 may be utilized to collect statistics for another color space (e.g., YUV or YCbCr). That is, the statistics process units 120 and 122 may operate in parallel to collect multiple sets of statistics for each frame of the image data acquired by the active sensor.

[0130] In the present embodiment, five asynchronous sources of data are provided in the ISP front-end 80. These include: (1) a direct input from a sensor interface corresponding to Sensor0 (90a) (referred to as Sif0 or Sens0), (2) a direct input from a sensor interface corresponding to Sensor1 (90b) (referred to as Sif1 or Sens1), (3) Sensor0 data input from the memory 108 (referred to as SifIn0 or Sens0DMA), which may include a DMA interface, (4) Sensor1 data input from the memory 108 (referred to as SifIn1 or Sens1DMA), and (5) a set of image data with frames from Sensor0 and Sensor1 data input retrieved from the memory 108 (referred to as FeProcIn or ProcInDMA). The ISP front-end 80 may also include multiple destinations to which image data from the sources may be routed, wherein each destination may be either a storage location in memory (e.g., in 108), or a processing unit. For instance, in the present embodiment, the ISP front-end 80 includes six destinations: (1) Sif0DMA for receiving Sensor0 data in the memory 108, (2) Sif1DMA for receiving Sensor1 data in the memory 108, (3) the first statistics processing unit 120 (StatsPipe0), (4) the second statistics processing unit 122 (StatsPipe1), (5) the front-end pixel processing unit (FEProc) 130, and (6) FeOut (or FEProcOut) to memory 108 or the ISP pipeline 82 (discussed in further detail below). In one embodiment, the ISP front-end 80 may be configured such that only certain destinations are valid for a particular source, as shown in Table 1 below.

[0131]

TABLE 1
Example of ISP Front-end valid destinations for each source
	SIf0DMA	SIf1DMA	StatsPipe0	StatsPipe1	FEProc	FEOut
Sens0	X		X	X	X	X
Sens1		X	X	X	X	X
Sens0DMA			X
Sens1DMA				X
ProcInDMA					X	X

[0132] For instance, in accordance with Table 1, source Sens0 (sensor interface of Sensor0) may be configured to provide data to destinations SIf0DMA (signal 134), StatsPipe0 (signal 136), StatsPipe1 (signal 138), FEProc (signal 140), or FEOut (signal 142). With regard to FEOut, source data may, in some instances, be provided to FEOut to bypass pixel processing by FEProc, such as for debugging or test purposes. Additionally, source Sens1 (sensor interface of Sensor1) may be configured to provide data to destinations SIf1DMA (signal 144), StatsPipe0 (signal 146), StatsPipe1 (signal 148), FEProc (signal 150), or FEOut (signal 152), source Sens0DMA (Sensor0 data from memory 108) may be configured to provide data to StatsPipe0 (signal 154), source Sens1DMA (Sensor1 data from memory 108) may be configured to provide data to StatsPipe1 (signal 156), and source ProcInDMA (Sensor0 and Sensor1 data from memory 108) may be configured to provide data to FEProc (signal 158) and FEOut (signal 160).

[0133] It should be noted that the presently illustrated embodiment is configured such that Sens0DMA (Sensor0 frames from memory 108) and Sens1DMA (Sensor1 frames from memory 108) are only provided to StatsPipe0 and StatesPipe1, respectively. This configuration allows the ISP front-end 80 to retain a certain number of previous frames (e.g., 5 frames) in memory. For example, due to a delay or lag between the time a user initiates a capture event (e.g., transitioning the image system from a preview mode to a capture or a recording mode, or even by just turning on or initializing the image sensor) using the image sensor to when an image scene is captured, not every frame that the user intended to capture may be captured and processed in substantially real-time. Thus, by retaining a certain number of previous frames in memory 108 (e.g., from a preview phase), these previous frames may be processed later or alongside the frames actually captured in response to the capture event, thus compensating for any such lag and providing a more complete set of image data.

[0134] With regard to the illustrated configuration of FIG. 8, it should be noted that the StatsPipe0 120 is configured to receive one of the inputs 136 (from Sens0), 146 (from Sens1), and 154 (from Sens0DMA), as determined by a selection logic 124, such as a multiplexer. Similarly, selection logic 126 may select an input from the signals 138, 156, and 148 to provide to StatsPipe1, and selection logic 132 may select an input from the signals 140, 150, and 158 to provide to FEProc. As mentioned above, the statistical data (Stats0 and Stats1) may be provided to the control logic 84 for the determination of various control parameters that may be used to operate the imaging device 30 and/or the ISP pipe processing logic 82. As can be appreciated, the selection logic blocks (120, 122, and 132) shown in FIG. 8 may be provided by any suitable type of logic, such as a multiplexer that selects one of multiple input signals in response to a control signal.

[0135] The pixel processing unit (FEProc) 130 may be configured to perform various image processing operations on the raw image data on a pixel-by-pixel basis. As shown, FEProc 130, as a destination processing unit, may receive image data from sources Sens0 (signal 140), Sens1 (signal 150), or ProcInDMA (signal 158) by way of the selection logic 132. FEProc 130 may also receive and output various signals (e.g., Rin, Hin, Hout, and Yout—which may represent motion history and luma data used during temporal filtering) when performing the pixel processing operations, which may include temporal filtering and binning compensation filtering, as will be discussed further below. The output 109 (FEProcOut) of the pixel processing unit 130 may then be forwarded to the ISP pipe logic 82, such as via one or more first-in-first-out (FIFO) queues, or may be sent to the memory 108.

[0136] Further, as shown in FIG. 8, the selection logic 132, in addition to receiving the signals 140, 150, and 158, may further receive the signals 159 and 161. The signal 159 may represented “pre-processed” raw image data from StatsPipe0, and the signal 161 may represent “pre-processed” raw image data from StatsPipe1. As will be discussed below, each of the statistics processing units may apply one or more pre-processing operations to the raw image data before collecting statistics. In one embodiment, each of the statistics processing units may perform a degree of defective pixel detection/correction, lens shading correction, black level compensation, and inverse black level compensation. Thus, the signals 159 and 161 may represent raw image data that has been processed using the aforementioned pre-processing operations (as will be discussed in further detail below in FIG. 37). Thus, the selection logic 132 gives the ISP front-end processing logic 80 the flexibility of providing either un-pre-processed raw image data from the Sensor0 (signal 140) and Sensor1 (signal 150) or pre-processed raw image data from StatsPipe0 (signal 159) and StatsPipe1 (signal 161). Additionally, as shown by selection logic units 162 and 163, the ISP front-end processing logic 80 also has the flexibility of writing either un-pre-processed raw image data from Sensor0 (signal 134) or Sensor1 (signal 144) to the memory 108, or writing pre-processed raw image data from StatsPipe0 (signal 159) or StatsPipe1 (signal 161) to the memory 108.

[0137] To control the operation of the ISP front-end logic 80, a front-end control unit 164 is provided. The control unit 164 may be configured to initialize and program control registers (referred to herein as “go registers”) for configuring and starting the processing of an image frame and to select an appropriate register bank(s) for updating double-buffered data registers. In some embodiments, the control unit 164 may also provide performance monitoring logic to log clock cycles, memory latency, and quality of service (QOS) information. Further, the control unit 164 may also control dynamic clock gating, which may be used to disable clocks to one or more portions of the ISP front-end 0 when there is not enough data in the input queue from an active sensor.

[0138] Using the “go registers” mentioned above, the control unit 164 may be able to control the updating of various parameters for each of the processing units (e.g., StatsPipe0, StatsPipe1, and FEProc) and may interface with the sensor interfaces to control the starting and stopping of the processing units. Generally each of the front-end processing units operates on a frame-by-frame basis. As discussed above (Table 1), the input to the processing units may be from the sensor interface (Sens0 or Sens1) or from memory 108. Further, the processing units may utilize various parameters and configuration data, which may be stored in corresponding data registers. In one embodiment, the data registers associated with each processing unit or destination may be grouped into blocks forming a register bank group. In the embodiment of FIG. 8, seven register bank groups may be defined in ISP Front-end: SIf0, SIf1, StatsPipe0, StatsPipe1, ProcPipe, FEOut and ProcIn. Each register block address space is duplicated to provide two banks of registers. Only the registers that are double buffered are instantiated in the second bank. If a register is not double buffered, the address in the second bank may be mapped to the address of the same register in the first bank.

[0139] For registers that are double buffered, registers from one bank are active and used by the processing units while the registers from the other bank are shadowed. The shadowed register may be updated by the control unit 164 during the current frame interval while hardware is using the active registers. The determination of which bank to use for a particular processing unit at a particular frame may be specified by a “NextBk” (next bank) field in a go register corresponding to the source providing the image data to the processing unit. Essentially, NextBk is a field that allows the control unit 164 to control which register bank becomes active on a triggering event for the subsequent frame.

[0140] Before discussing the operation of the go registers in detail, FIG. 9 provides a general method 166 for processing image data on a frame-by-frame basis in accordance with the present techniques. Beginning at step 168, the destination processing units targeted by a data source (e.g., Sens0, Sens1, Sens0DMA, Sens1DMA, or ProcInDMA) enter an idle state. This may indicate that processing for the current frame is completed and, therefore, the control unit 164 may prepare for processing the next frame. For instance, at step 170, programmable parameters for each destination processing unit are updated. This may include, for example, updating the NextBk field in the go register corresponding to the source, as well as updating any parameters in the data registers corresponding to the destination units. Thereafter, at step 172, a triggering event may place the destination units into a run state. Further, as shown at step 174, each destination unit targeted by the source completes its processing operations for the current frame, and the method 166 may subsequently return to step 168 for the processing of the next frame.

[0141] FIG. 10 depicts a block diagram view showing two banks of data registers 176 and 178 that may be used by the various destination units of the ISP-front end. For instance, Bank 0 (176) may include the data registers 1-n (176a-176d), and Bank 1 (178) may include the data registers 1-n (178a-178d). As discussed above, the embodiment shown in FIG. 8 may utilize a register bank (Bank 0) having seven register bank groups (e.g., SIf0, SIf1, StatsPipe0, StatsPipe1, ProcPipe, FEOut and ProcIn). Thus, in such an embodiment, the register block address space of each register is duplicated to provide a second register bank (Bank 1).

[0142] FIG. 10 also illustrates go register 180 that may correspond to one of the sources. As shown, the go register 180 includes a “NextVld” field 182 and the above-mentioned “NextBk” field 184. These fields may be programmed prior to starting the processing of the current frame. Particularly, NextVld may indicate the destination(s) to where data from the source is to be sent. As discussed above, NextBk may select a corresponding data register from either Bank® or Bank1 for each destination targeted, as indicated by NextVld. Though not shown in FIG. 10, the go register 180 may also include an arming bit, referred to herein as a “go bit,” which may be set to arm the go register. When a triggering event 192 for a current frame is detected, NextVld and NextBk may be copied into a CurrVld field 188 and a CurrBk field 190 of a corresponding current or “active” register 186. In one embodiment, the current register(s) 186 may be read-only registers that may set by hardware, while remaining inaccessible to software commands within the ISP front-end 80.

[0143] As will be appreciated, for each ISP front-end source, a corresponding go register may be provided. For the purposes of this disclosure, the go registers corresponding to the above-discussed sources Sens0, Sens1, Sens0DMA, Sens1DMA, and ProcInDMA may be referred to as Sens0Go, Sens1Go, Sens0DMAGo, Sens1DMAGo and ProcInDMAGo, respectively. As mentioned above, the control unit may utilize the go registers to control the sequencing of frame processing within the ISP front end 80. Each go register contains a NextVld field and a NextBk field to indicate what destinations will be valid, and which register bank (0 or 1) will be used, respectively, for the next frame. When the next frame's triggering event 192 occurs, the NextVld and NextBk fields are copied to a corresponding active read-only register 186 that indicates the current valid destinations and bank numbers, as shown above in FIG. 10. Each source may be configured to operate asynchronously and can send data to any of its valid destinations. Further, it should be understood that for each destination, generally only one source may be active during a current frame.

[0144] With regard to the arming and triggering of the go register 180, asserting an arming bit or “go bit” in the go register 180 arms the corresponding source with the associated NextVld and NextBk fields. For triggering, various modes are available depending on whether the source input data is read from memory (e.g., Sens0DMA, Sens1DMA or ProcInDMA), or whether the source input data is from a sensor interface (e.g., Sens0 or Sens1). For instance, if the input is from memory 108, the arming of the go bit itself may serve as the triggering event, since the control unit 164 has control over when data is read from the memory 108. If the image frames are being input by the sensor interface, then triggering event may depend on the timing at which the corresponding go register is armed relative to when data from the sensor interface is received. In accordance with the present embodiment, three different techniques for triggering timing from a sensor interface input are shown in FIGS. 11-13.

[0145] Referring first to FIG. 11, a first scenario is illustrated in which triggering occurs once all destinations targeted by the source transition from a busy or run state to an idle state. Here, a data signal VVALID (196) represents an image data signal from a source. The pulse 198 represents a current frame of image data, the pulse 202 represents the next frame of image data, and the interval 200 represents a vertical blanking interval (VBLANK) 200 (e.g., represents the time differential between the last line of the current frame 198 and the next frame 202). The time differential between the rising edge and falling edge of the pulse 198 represents a frame interval 201. Thus, in FIG. 11, the source may be configured to trigger when all targeted destinations have finished processing operations on the current frame 198 and transition to an idle state. In this scenario, the source is armed (e.g., by setting the arming or “go” bit) before the destinations complete processing so that the source can trigger and initiate processing of the next frame 202 as soon as the targeted destinations go idle. During the vertical blanking interval 200 the processing units may be set up and configured for the next frame 202 using the register banks specified by the go register corresponding to the source before the sensor input data arrives. By way of example only, read buffers used by FEProc 130 may be filled before the next frame 202 arrives. In this case, shadowed registers corresponding to the active register banks may be updated after the triggering event, thus allowing for a full frame interval to setup the double-buffered registers for the next frame (e.g., after frame 202).

[0146] FIG. 12 illustrates a second scenario in which the source is triggered by arming the go bit in the go register corresponding to the source. Under this “trigger-on-go” configuration, the destination units targeted by the source are already idle and the arming of the go bit is the triggering event. This triggering mode may be utilized for registers that are not double-buffered and, therefore, are updated during vertical blanking (e.g., as opposed to updating a double-buffered shadow register during the frame interval 201).

[0147] FIG. 13 illustrates a third triggering mode in which the source is triggered upon detecting the start of the next frame, i.e., a rising VSYNC. However, it should be noted that in this mode, if the go register is armed (by setting the go bit) after the next frame 202 has already started processing, the source will use the target destinations and register banks corresponding to the previous frame, since the CurrVld and CurrBk fields are not updated before the destination start processing. This leaves no vertical blanking interval for setting up the destination processing units and may potentially result in dropped frames, particularly when operating in a dual sensor mode. It should be noted, however, that this mode may nonetheless result in accurate operation if the image processing circuitry 32 is operating in a single sensor mode that uses the same register banks for each frame (e.g., the destination (NextVld) and register banks (NextBk) do not change).

[0148] Referring now to FIG. 14, a control register (or “go register”) 180 is illustrated in more detail. The go register 180 includes the arming “go” bit 204, as well as the NextVld field 182 and the NextBk field 184. As discussed above, each source (e.g., Sens0, Sens1, Sens0DMA, Sens1DMA, or ProcInDMA) of the ISP front-end 80 may have a corresponding go register 180. In one embodiment, the go bit 204 may be a single-bit field, and the go register 180 may be armed by setting the go bit 204 to 1. The NextVld field 182 may contain a number of bits corresponding to the number of destinations in the ISP front-end 80. For instance, in the embodiment shown in FIG. 8, the ISP front-end includes six destinations: Sif0DMA, Sif1DMA, StatsPipe0, StatsPipe1, FEProc, and FEOut. Thus, the go register 180 may include six bits in the NextVld field 182, with one bit corresponding to each destination, and wherein targeted destinations are set to 1. Similarly, the NextBk field 182 may contain a number of bits corresponding to the number of data registers in the ISP front-end 80. For instance, as discussed above, the embodiment of the ISP front-end 80 shown in FIG. 8 may include seven data registers: SIf0, SIf1, StatsPipe0, StatsPipe1, ProcPipe, FEOut and ProcIn. Accordingly, the NextBk field 184 may include seven bits, with one bit corresponding to each data register, and wherein data registers corresponding to Bank 0 and 1 are selected by setting their respective bit values to 0 or 1, respectively. Thus, using the go register 180, the source, upon triggering, knows precisely which destination units are to receive frame data, and which register banks are to be used for configuring the targeted destination units.

[0149] Additionally, due to the dual sensor configuration supported by the ISP circuitry 32, the ISP front-end may operate in a single sensor configuration mode (e.g., only one sensor is acquiring data) and a dual sensor configuration mode (e.g., both sensors are acquiring data). In a typical single sensor configuration, input data from a sensor interface, such as Sens0, is sent to StatsPipe0 (for statistics processing) and FEProc (for pixel processing). In addition, sensor frames may also be sent to memory (SIf0DMA) for future processing, as discussed above.

[0150] An example of how the NextVld fields corresponding to each source of the ISP front-end 80 may be configured when operating in a single sensor mode is depicted below in Table 2.

[0151]

TABLE 2
NextVld per source example: Single sensor mode
	SIf0DMA	SIf1DMA	StatsPipe0	StatsPipe1	FEProc	FEOut
Sens0Go	1	X	1	0	1	0
Sens1Go	X	0	0	0	0	0
Sens0DMAGo	X	X	0	X	X	X
Sens1DMAGo	X	X	X	0	X	X
ProcInDMAGo	X	X	X	X	0	0

As discussed above with reference to Table 1, the ISP front-end 80 may be configured such that only certain destinations are valid for a particular source. Thus, the destinations in Table 2 marked with “X” are intended to indicate that the ISP front-end 80 is not configured to allow a particular source to send frame data to that destination. For such destinations, the bits of the NextVld field of the particular source corresponding to that destination may always be 0. It should be understood, however, that this is merely one embodiment and, indeed, in other embodiments, the ISP front-end 80 may be configured such that each source is capable of targeting each available destination unit.

[0152] The configuration shown above in Table 2 represents a single sensor mode in which only Sensor0 is providing frame data. For instance, the Sens0Go register indicates destinations as being SIf0DMA, StatsPipe0, and FEProc. Thus, when triggered, each frame of the Sensor0 image data, is sent to these three destinations. As discussed above, SIf0DMA may store frames in memory 108 for later processing, StatsPipe0 applies statistics processing to determine various statistic data points, and FEProc processes the frame using, for example, temporal filtering and binning compensation filtering. Further, in some configurations where additional statistics are desired (e.g., statistics in different color spaces), StatsPipe1 may also be enabled (corresponding NextVld set to 1) during the single sensor mode. In such embodiments, the Sensor0 frame data is sent to both StatsPipe0 and StatsPipe1. Further, as shown in the present embodiment, only a single sensor interface (e.g., Sens0 or alternatively Sen0) is the only active source during the single sensor mode.

[0153] With this in mind, FIG. 15 provides a flow chart depicting a method 206 for processing frame data in the ISP front-end 80 when only a single sensor is active (e.g., Sensor 0). While the method 206 illustrates in particular the processing of Sensor0 frame data by FEProc 130 as an example, it should be understood that this process may be applied to any other source and corresponding destination unit in the ISP front-end 80. Beginning at step 208, Sensor0 begins acquiring image data and sending the captured frames to the ISP front-end 80. The control unit 164 may initialize programming of the go register corresponding to Sens0 (the Sensor0 interface) to determine target destinations (including FEProc) and what bank registers to use, as shown at step 210. Thereafter, decision logic 212 determines whether a source triggering event has occurred. As discussed above, frame data input from a sensor interface may utilize different triggering modes (FIGS. 11-13). If a trigger event is not detected, the process 206 continues to wait for the trigger. Once triggering occurs, the next frame becomes the current frame and is sent to FEProc (and other target destinations) for processing at step 214. FEProc may be configured using data parameters based on a corresponding data register (ProcPipe) specified in the NextBk field of the Sens0Go register. After processing of the current frame is completed at step 216, the method 206 may return to step 210, at which the Sens0Go register is programmed for the next frame.

[0154] When both Sensor0 and Sensor1 of the ISP front-end 80 are both active, statistics processing remains generally straightforward, since each sensor input may be processed by a respective statistics block, StatsPipe0 and StatsPipe1. However, because the illustrated embodiment of the ISP front-end 80 provides only a single pixel processing unit (FEProc), FEProc may be configured to alternate between processing frames corresponding to Sensor0 input data and frames corresponding to Sensor1 input data. As will be appreciated, the image frames are read from FEProc in the illustrated embodiment to avoid a condition in which image data from one sensor is processed in real-time while image data from the other sensor is not processed in real-time. For instance, as shown in Table 3 below, which depicts one possible configuration of NextVld fields in the go registers for each source when the ISP-front end 80 is operating in a dual sensor mode, input data from each sensor is sent to memory (SIf0DMA and SIf1DMA) and to the corresponding statistics processing unit (StatsPipe0 and StatsPipe1).

[0155]

TABLE 3
NextVld per source example: Dual sensor mode
	SIf0DMA	SIf1DMA	StatsPipe0	StatsPipe1	FEProc	FEOut
Sens0Go	1	X	1	0	0	0
Sens1Go	X	1	0	1	0	0
Sens0DMAGo	X	X	0	X	X	X
Sens1DMAGo	X	X	X	0	X	X
ProcInDMAGo	X	X	X	X	1	0

[0156] The sensor frames in memory are sent to FEProc from the ProcInDMA source, such that they alternate between Sensor0 and Sensor1 at a rate based on their corresponding frame rates. For instance, if Sensor0 and Sensor1 are both acquiring image data at a rate of 30 frames per second (fps), then their sensor frames may be interleaved in a 1-to-1 manner. If Sensor0 (30 fps) is acquiring image data at a rate twice that of Sensor1 (15 fps), then the interleaving may be 2-to-1, for example. That is, two frames of Sensor0 data are read out of memory for every one frame of Sensor1 data.

[0157] With this in mind, FIG. 16 depicts a method 220 for processing frame data in the ISP front-end 80 having two sensors acquiring image data simultaneously. At step 222, both Sensor0 and Sensor1 begin acquiring image frames. As will be appreciated, Sensor0 and Sensor1 may acquire the image frames using different frame rates, resolutions, and so forth. At step 224, the acquired frames from Sensor0 and Sensor1 written to memory 108 (e.g., using SIf0DMA and SIf1DMA destinations). Next, source ProcInDMA reads the frame data from the memory 108 in an alternating manner, as indicated at step 226. As discussed, frames may alternate between Sensor0 data and Sensor1 data depending on frame rate at which the data is acquired. At step 228, the next frame from ProcInDMA is acquired. Thereafter, at step 230, the NextVld and NextBk fields of the go register corresponding to the source, here ProcInDMA, is programmed depending on whether the next frame is Sensor0 or Sensor1 data. Thereafter, decision logic 232 determines whether a source triggering event has occurred. As discussed above, data input from memory may be triggered by arming the go bit (e.g., “trigger-on-go” mode). Thus, triggering may occur once the go bit of the go register is set to 1. Once triggering occurs, the next frame becomes the current frame and is sent to FEProc for processing at step 234. As discussed above, FEProc may be configured using data parameters based on a corresponding data register (ProcPipe) specified in the NextBk field of the ProcInDMAGo register. After processing of the current frame is completed at step 236, the method 220 may return to step 228 and continue.

[0158] A further operational event that the ISP front-end 80 is configured to handle is a configuration change during image processing. For instance, such an event may occur when the ISP front-end 80 transitions from a single sensor configuration to a dual sensor configuration, or vice-versa. As discussed above, the NextVld fields for certain sources may be different depending on whether one or both image sensors are active. Thus, when the sensor configuration is changed, the ISP front-end control unit 164 may release all destination units before they are targeted by a new source. This may avoid invalid configurations (e.g., assigning multiple sources to one destination). In one embodiment, the release of the destination units may be accomplished by setting the NextVld fields of all the go registers to 0, thus disabling all destinations, and arming the go bit. After the destination units are released, the go registers may be reconfigured depending on the current sensor mode, and image processing may continue.

[0159] A method 240 for switching between single and dual sensor configurations is shown in FIG. 17, in accordance with one embodiment. Beginning at step 242, a next frame of image data from a particular source of the ISP front-end 80 is identified. At step 244, the target destinations (NextVld) are programmed into the go register corresponding to the source. Next, at step 246, depending on the target destinations, NextBk is programmed to point to the correct data registers associated with the target destinations. Thereafter, decision logic 248 determines whether a source triggering event has occurred. Once triggering occurs, the next frame is sent to the destination units specified by NextVld and processed by the destination units using the corresponding data registers specified by NextBk, as shown at step 250. The processing continues until step 252, at which the processing of the current frame is completed.

[0160] Subsequently, decision logic 254 determines whether there is a change in the target destinations for the source. As discussed above, NextVld settings of the go registers corresponding to Sens0 and Sens1 may vary depending on whether one sensor or two sensors are active. For instance, referring to Table 2, if only Sensor0 is active, Sensor0 data is sent to SIf0DMA, StatsPipe0, and FEProc. However, referring to Table 3, if both Sensor0 and Sensor1 are active, then Sensor0 data is not sent directly FEProc. Instead, as mentioned above, Sensor0 and Sensor1 data is written to memory 108 and is read out to FEProc in an alternating manner by source ProcInDMA. Thus, if no target destination change is detected at decision logic 254, the control unit 164 deduces that the sensor configuration has not changed, and the method 240 returns to step 246, whereat the NextBk field of the source go register is programmed to point to the correct data registers for the next frame, and continues.

[0161] If, however, at decision logic 254, a destination change is detected, then the control unit 164 determines that a sensor configuration change has occurred. For instance, this could represent switching from single sensor mode to dual sensor mode, or shutting off the sensors altogether. Accordingly, the method 240 continues to step 256, at which all bits of the NextVld fields for all go registers are set to 0, thus effectively disabling the sending of frames to any destination on the next trigger. Then, at decision logic 258, a determination is made as to whether all destination units have transition to an idle state. If not, the method 240 waits at decision logic 258 until all destinations units have completed their current operations. Next, at decision logic 260, a determination is made as to whether image processing is to continue. For instance, if the destination change represented the deactivation of both Sensor0 and Sensor1, then image processing ends at step 262. However, if it is determined that image processing is to continue, then the method 240 returns to step 244 and the NextVld fields of the go registers are programmed in accordance with the current operation mode (e.g., single sensor or dual sensor). As shown here, the steps 254-262 for clearing the go registers and destination fields may collectively be referred to by reference number 264.

[0162] Next, FIG. 18 shows a further embodiment by way of the flow chart (method 265) that provides for another dual sensor mode of operation. The method 265 depicts a condition in which one sensor (e.g., Sensor0) is actively acquiring image data and sending the image frames to FEProc 130 for processing, while also sending the image frames to StatsPipe0 and/or memory 108 (Sif0DMA), while the other sensor (e.g., Sensor1) is inactive (e.g., turned off), as shown at step 266. Decision logic 268 then detects for a condition in which Sensor1 will become active on the next frame to send image data to FEProc. If this condition is not met, then the method 265 returns to step 266. However, if this condition is met, then the method 265 proceeds by performing action 264 (collectively steps 254-262 of FIG. 17), whereby the destination fields of the sources are cleared and reconfigured at step 264. For instance, at step 264, the NextVld field of the go register associated with Sensor1 may be programmed to specify FEProc as a destination, as well as StatsPipe1 and/or memory (Sif1DMA), while the NextVld field of the go register associated with Sensor0 may be programmed to clear FEProc as a destination. In this embodiment, although frames captured by Sensor0 are not sent to FEProc on the next frame, Sensor0 may remain active and continue to send its image frames to StatsPipe0, as shown at step 270, while Sensor1 captures and sends data to FEProc for processing at step 272. Thus, both sensors, Sensor0 and Sensor1 may continue to operate in this “dual sensor” mode, although only image frames from one sensor are sent to FEProc for processing. For the purposes of this example, a sensor sending frames to FEProc for processing may be referred to as an “active sensor,” a sensor that is not sending frame FEProc but is still sending data to the statistics processing units may be referred to as a “semi-active sensor,” and a sensor that is not acquiring data at all may be referred to as an “inactive sensor.”

[0163] One benefit of the foregoing technique is that the because statistics continue to be acquired for the semi-active sensor (Sensor0), the next time the semi-active sensor transitions to an active state and the current active sensor (Sensor 1) transitions to a semi-active or inactive state, the semi-active sensor may begin acquiring data within one frame, since color balance and exposure parameters may already be available due to the continued collection of image statistics. This technique may be referred to as “hot switching” of the image sensors, and avoids drawbacks associated with “cold starts” of the image sensors (e.g., starting with no statistics information available). Further, to save power, since each source is asynchronous (as mentioned above), the semi-active sensor may operate at a reduced clock and/or frame rate during the semi-active period.

[0164] Before continuing with a more detailed description of the statistics processing and pixel processing operations depicted in the ISP front-end logic 80 of FIG. 8, it is believed that a brief introduction regarding the definitions of various ISP frame regions will help to facilitate a better understanding of the present subject matter. With this in mind, various frame regions that may be defined within an image source frame are illustrated in FIG. 19. The format for a source frame provided to the image processing circuitry 32 may use either the tiled or linear addressing modes discussed above, as may utilize pixel formats in 8, 10, 12, or 14-bit precision. The image source frame 274, as shown in FIG. 19, may include a sensor frame region 276, a raw frame region 276, and an active region 278. The sensor frame 276 is generally the maximum frame size that the image sensor 90 can provide to the image processing circuitry 32. The raw frame region 278 may be defined as the region of the sensor frame 276 that is sent to the ISP front-end processing logic 80. The active region 280 may be defined as a portion of the source frame 274, typically within the raw frame region 278, on which processing is performed for a particular image processing operation. In accordance with embodiments of the present technique, that active region 280 may be the same or may be different for different image processing operations.

[0165] In accordance with aspects of the present technique, the ISP front-end logic 80 only receives the raw frame 278. Thus, for the purposes of the present discussion, the global frame size for the ISP front-end processing logic 80 may be assumed as the raw frame size, as determined by the width 282 and height 284. In some embodiments, the offset from the boundaries of the sensor frame 276 to the raw frame 278 may be determined and/or maintained by the control logic 84. For instance, the control logic 84 may be include firmware that may determine the raw frame region 278 based upon input parameters, such as the x-offset 286 and the y-offset 288, that are specified relative to the sensor frame 276. Further, in some cases, a processing unit within the ISP front-end logic 80 or the ISP pipe logic 82 may have a defined active region, such that pixels in the raw frame but outside the active region 280 will not be processed, i.e., left unchanged. For instance, an active region 280 for a particular processing unit having a width 290 and height 292 may be defined based upon an x-offset 294 and y-offset 296 relative to the raw frame 278. Further, where an active region is not specifically defined, one embodiment of the image processing circuitry 32 may assume that the active region 280 is the same as the raw frame 278 (e.g., x-offset 294 and y-offset 296 are both equal to 0). Thus, for the purposes of image processing operations performed on the image data, boundary conditions may be defined with respect to the boundaries of the raw frame 278 or active region 280.

[0166] Keeping these points in mind and referring to FIG. 20, a more detailed view of the ISP front-end pixel processing logic 130 (previously discussed in FIG. 8) is illustrated, in accordance with an embodiment of the present technique. As shown, the ISP front-end pixel processing logic 130 includes a temporal filter 298 and a binning compensation filter 300. The temporal filter 298 may receive one of the input image signals Sif0, Sif1, FEProcIn, or pre-processed image signals (e.g., 159, 161) and may operate on the raw pixel data before any additional processing is performed. For example, the temporal filter 298 may initially process the image data to reduce noise by averaging image frames in the temporal direction. The binning compensation filter 300, which is discussed in more detail below, may apply scaling and re-sampling on binned raw image data from an image sensor (e.g., 90a, 90b) to maintain an even spatial distribution of the image pixels.

[0167] The temporal filter 298 may be pixel-adaptive based upon motion and brightness characteristics. For instance, when pixel motion is high, the filtering strength may be reduced in order to avoid the appearance of “trailing” or “ghosting artifacts” in the resulting processed image, whereas the filtering strength may be increased when little or no motion is detected. Additionally, the filtering strength may also be adjusted based upon brightness data (e.g., “luma”). For instance, as image brightness increases, filtering artifacts may become more noticeable to the human eye. Thus, the filtering strength may be further reduced when a pixel has a high level of brightness.

[0168] In applying temporal filtering, the temporal filter 298 may receive reference pixel data (Rin) and motion history input data (Hin), which may be from a previous filtered or original frame. Using these parameters, the temporal filter 298 may provide motion history output data (Hout) and filtered pixel output (Yout). The filtered pixel output Yout is then passed to the binning compensation filter 300, which may be configured to perform one or more scaling operations on the filtered pixel output data Yout to produce the output signal FEProcOut. The processed pixel data FEProcOut may then be forwarded to the ISP pipe processing logic 82, as discussed above.

[0169] Referring to FIG. 21, a process diagram depicting a temporal filtering process 302 that may be performed by the temporal filter shown in FIG. 20 is illustrated, in accordance with a first embodiment. The temporal filter 298 may include a 2-tap filter, wherein the filter coefficients are adjusted adaptively on a per pixel basis based at least partially upon motion and brightness data. For instance, input pixels x(t), with the variable “t” denoting a temporal value, may be compared to reference pixels r(t−1) in a previously filtered frame or a previous original frame to generate a motion index lookup in a motion history table (M) 304 that may contain filter coefficients. Additionally, based upon motion history input data h(t−1), a motion history output h(t) corresponding to the current input pixel x(t) may be determined.

[0170] The motion history output h(t) and a filter coefficient, K, may be determined based upon a motion delta d(j,i,t), wherein (j,i) represent coordinates of the spatial location of a current pixel x(j,i,t). The motion delta d(j,i,t) may be computed by determining the maximum of three absolute deltas between original and reference pixels for three horizontally collocated pixels of the same color. For instance, referring briefly to FIG. 22, the spatial locations of three collocated reference pixels 308, 309, and 310 that corresponding to original input pixels 312, 313, and 314 are illustrated. In one embodiment, the motion delta may be calculated based on these original and reference pixels using formula below:
d(j,i,t)=max3[abs(x(j,i−2,t)−r(j,i−2,t−1)),
(abs(x(j,i,t)−r(j,i,t−1)),
(abs(x(j,i+2,t)−r(j,i+2,t−1))]  (1a)
A flow chart depicting this technique for determining the motion delta value is illustrated further below in FIG. 24. Further, it should be understood that the technique for calculating the motion delta value, as shown above in Equation 1a (and below in FIG. 24), is only intended to provide one embodiment for determining a motion delta value.

[0171] In other embodiments, an array of same-colored pixels could be evaluated to determine a motion delta value. For instance, in addition to the three pixels referenced in Equation 1a, one embodiment for determining motion delta values may include also evaluating the absolute deltas between same colored pixels from two rows above (e.g., j−2; assuming a Bayer pattern) the reference pixels 312, 313, and 314 and their corresponding collocated pixels, and two rows below (e.g., j+2; assuming a Bayer pattern) the reference pixels 312, 313, and 314 and their corresponding collocated pixels. For instance, in one embodiment, the motion delta value may be expressed as follows:
d(j,i,t)=max9[abs(x(j,i−2,t)−r(j,i−2,t−1)),
(abs(x(j,i,t)−r(j,i,t−1)),
(abs(x(j,i+2,t)−r(j,i+2,t−1)),
(abs(x(j−2,i−2,t)−r(j−2,i−2,t−1)),
(abs(x(j−2,i,t)−r(j−2,i,t−1)),
(abs(x(j−2,i+2,t)−r(j−2,i+2,t−1)),
(abs(x(j+2,i−2,t)−r(j+2,i−2,t−1))
(abs(x(j+2,i,t)−r(j+2,i,t−1)),
(abs(x(j+2,i+2,t)−r(j+2,i+2,t−1))]  (1b)
Thus, in the embodiment depicted by Equation 1b, the motion delta value may be determined by comparing the absolute delta between a 3×3 array of same-colored pixels, with the current pixel (313) being located at the center of the 3×3 array (e.g., really a 5×5 array for Bayer color patterns if pixels of different colors are counted). It should be appreciated, that any suitable two-dimensional array of same-colored pixels (e.g., including arrays having all pixels in the same row (e.g., Equation 1a) or arrays having all pixels in the same column) with the current pixel (e.g., 313) being located at the center of the array could be analyzed to determine a motion delta value. Further, while the motion delta value could be determined as the maximum of the absolute deltas (e.g., as shown in Equations 1a and 1b), in other embodiments, the motion delta value could also be selected as the mean or median of the absolute deltas. Additionally, the foregoing techniques may also be applied to other types of color filter arrays (e.g., RGBW, CYGM, etc.), and is not intended to be exclusive to Bayer patterns.

[0172] Referring back to FIG. 21, once the motion delta value is determined, a motion index lookup that may be used to selected the filter coefficient K from the motion table (M) 304 may be calculated by summing the motion delta d(t) for the current pixel (e.g., at spatial location (j,i)) with the motion history input h(t−1). For instance, the filter coefficient K may be determined as follows:
K=M[d(j,i,t)+h(j,i,t−1)]  (2a)
Additionally, the motion history output h(t) may be determined using the following formula:
h(j,i,t)=d(j,i,t)+(1−K)×h(j,i,t−1)  (3a)

[0173] Next, the brightness of the current input pixel x(t) may be used to generate a luma index lookup in a luma table (L) 306. In one embodiment, the luma table may contain attenuation factors that may be between 0 and 1, and may be selected based upon the luma index. A second filter coefficient, K′, may be calculated by multiplying the first filter coefficient K by the luma attenuation factor, as shown in the following equation:
K′=K×L[x(j,i,t)]  (4a)

[0174] The determined value for K′ may then be used as the filtering coefficient for the temporal filter 298. As discussed above, the temporal filter 298 may be a 2-tap filter. Additionally, the temporal filter 298 may be configured as an infinite impulse response (IIR) filter using previous filtered frame or as a finite impulse response (FIR) filter using previous original frame. The temporal filter 298 may compute the filtered output pixel y(t) (Yout) using the current input pixel x(t), the reference pixel r(t−1), and the filter coefficient K′ using the following formula:
y(j,i,t)=r(j,i,t−1)+K′(x(j,i,t)−r(j,i,t−1))  (5a)
As discussed above, the temporal filtering process 302 shown in FIG. 21 may be performed on a pixel-by-pixel basis. In one embodiment, the same motion table M and luma table L may be used for all color components (e.g., R, G, and B). Additionally, some embodiments may provide a bypass mechanism, in which temporal filtering may be bypassed, such as in response to a control signal from the control logic 84. Further, as will be discussed below with respect to FIGS. 26 and 27, one embodiment of the temporal filter 298 may utilize separate motion and luma tables for each color component of the image data.

[0175] The embodiment of the temporal filtering technique described with reference to FIGS. 21 and 22 may be better understood in view of FIG. 23, which depicts a flow chart illustrating a method 315, in accordance with the above-described embodiment. The method 315 begins at step 316, at which a current pixel x(t) located at spatial location (j,i) of a current frame of image data is received by the temporal filtering system 302. At step 317, a motion delta value d(t) is determined for the current pixel x(t) based at least partially upon one or more collocated reference pixels (e.g., r(t−1)) from a previous frame of the image data (e.g., the image frame immediately preceding the current frame). A technique for determining a motion delta value d(t) at step 317 is further explained below with reference to FIG. 24, and may be performed in accordance with Equation 1a, as shown above.

[0176] Once the motion delta value d(t) from step 317 is obtained, a motion table lookup index may be determined using the motion delta value d(t) and a motion history input value h(t−1) corresponding to the spatial location (j,i) from the previous frame, as shown in step 318. Additionally, though not shown, a motion history value h(t) corresponding to the current pixel x(t) may also be determined at step 318 once the motion delta value d(t) is known, for example, by using Equation 3a shown above. Thereafter, at step 319, a first filter coefficient K may be selected from a motion table 304 using the motion table lookup index from step 318. The determination of the motion table lookup index and the selection of the first filter coefficient K from the motion table may be performed in accordance with Equation 2a, as shown above.

[0177] Next, at step 320, an attenuation factor may be selected from a luma table 306. For instance, the luma table 306 may contain attenuation factors ranging from between approximately 0 and 1, and the attenuation factor may be selected from the luma table 306 using the value of the current pixel x(t) as a lookup index. Once the attenuation factor is selected, a second filter coefficient K′ may be determined at step 321 using the selected attenuation factor and the first filter coefficient K (from step 319), as shown in Equation 4a above. Then, at step 322, a temporally filtered output value y(t) corresponding to the current input pixel x(t) is determined based upon the second filter coefficient K′ (from step 320), the value of the collocated reference pixel r(t−1), and the value of the input pixel x(t). For instance, in one embodiment, the output value y(t) may be determined in accordance with Equation 5a, as shown above.

[0178] Referring to FIG. 24, the step 317 for determining the motion delta value d(t) from the method 315 is illustrated in more detail in accordance with one embodiment. In particular, the determination of the motion delta value d(t) may generally correspond to the operation depicted above in accordance with Equation 1a. As shown, the step 317 may include the sub-steps 324-327. Beginning at sub-step 324, a set of three horizontally adjacent pixels having the same color value as the current input pixel x(t) are identified. By way of example, in accordance with the embodiment shown in FIG. 22 the image data may include Bayer image data, and the three horizontally adjacent pixels may include the current input pixel x(t) (313), a second pixel 312 of the same color to the left of the current input pixel 313, and a third pixel of the same color to the right of the current input pixel 313.

[0179] Next, at sub-step 325, three collocated reference pixels 308, 309, and 310 from the previous frame corresponding to the selected set of three horizontally adjacent pixels 312, 313, and 314 are identified. Using the selected pixels 312, 313, and 314 and the three collocated reference pixels 308, 309, and 310, the absolute values of the differences between each of the three selected pixels 312, 313, and 314 and their corresponding collocated reference pixels 308, 309, and 310, respectively, are determined at sub-step 326. Subsequently, at sub-step 327, the maximum of the three differences from sub-step 326 is selected as the motion delta value d(t) for the current input pixel x(t). As discussed above, FIG. 24, which illustrates the motion delta value calculation technique shown in Equation 1a, is only intended to provide one embodiment. Indeed, as discussed above, any suitable two-dimensional array of same-colored pixels with the current pixel being centered in the array may be used to determine a motion delta value (e.g., Equation 1b).

[0180] Another embodiment of a technique for applying temporal filtering to image data is further depicted in FIG. 25. For instance, since signal to noise ratios for different color components of the image data may be different, a gain may be applied to the current pixel, such that the current pixel is gained before selecting motion and luma values from the motion table 304 and luma table 306. By applying a respective gain that is color dependent, signal to noise ratio may be more consistent among the different color components. By way of example only, in an implementation that uses raw Bayer image data, the red and blue color channels may generally be more sensitive compared to the green (Gr and Gb) color channels. Thus, by applying an appropriate color-dependent gain to each processed pixel, the signal to noise variation between each color component may be generally reduced, thereby reducing, among other things, ghosting artifacts, as well as consistency across different colors after auto-white balance gains.

[0181] With this in mind, FIG. 25 provides a flow chart depicting a method 328 for applying temporal filtering to image data received by the front-end processing unit 130 in accordance with such an embodiment. Beginning at step 329, a current pixel x(t) located at spatial location (j,i) of a current frame of image data is received by the temporal filtering system 302. At step 330, a motion delta value d(t) is determined for the current pixel x(t) based at least partially upon one or more collocated reference pixels (e.g., r(t−1)) from a previous frame of the image data (e.g., the image frame immediately preceding the current frame). The step 330 may be similar to the step 317 of FIG. 23, and may utilize the operation represented in Equation 1 above.

[0182] Next, at step 331, a motion table lookup index may be determined using the motion delta value d(t), a motion history input value h(t−1) corresponding to the spatial location (j,i) from the previous frame (e.g., corresponding to the collocated reference pixel r(t−1)), and a gain associated with the color of the current pixel. Thereafter, at step 332, a first filter coefficient K may be selected from the motion table 304 using the motion table lookup index determined at step 331. By way of example only, in one embodiment, the filter coefficient K and the motion table lookup index may be determined as follows:
K=M[gain[c]×(d(j,i,t)+h(j,i,t−1))],  (2b)
wherein M represents the motion table, and wherein the gain[c] corresponds to a gain associated with the color of the current pixel. Additionally, though not shown in FIG. 25, it should be understood that a motion history output value h(t) for the current pixel may also be determined and may be used to apply temporal filtering to a collocated pixel of a subsequent image frame (e.g., the next frame). In the present embodiment, the motion history output h(t) for the current pixel x(t) may be determined using the following formula:
h(j,i,t)=d(j,i,t)+K[h(j,i,t−1)−d(j,i,t)]  (3b)

[0183] Next, at step 333, an attenuation factor may be selected from the luma table 306 using a luma table lookup index determined based upon the gain (gain[c]) associated with the color of the current pixel x(t). As discussed above, the attenuation factors stored in the luma table may have a range from approximately 0 to 1. Thereafter, at step 334, a second filter coefficient K′ may be calculated based upon the attenuation factor (from step 333) and the first filter coefficient K (from step 332). By way of example only, in one embodiment, the second filter coefficient K′ and the luma table lookup index may be determined as follows:
K′=K×L[gain[c]×x(j,i,t)]  (4b)

[0184] Next, at step 335, a temporally filtered output value y(t) corresponding to the current input pixel x(t) is determined based upon the second filter coefficient K′ (from step 334), the value of the collocated reference pixel r(t−1), and the value of the input pixel x(t). For instance, in one embodiment, the output value y(t) may be determined as follows:
y(j,i,t)=x(j,i,t)+K′(r(j,i,t−1)−x(j,i,t))  (5b)

[0185] Continuing to FIG. 26, a further embodiment of the temporal filtering process 336 is depicted. Here, the temporal filtering process 336 may be accomplished in a manner similar to the embodiment discussed in FIG. 25, except that instead of applying a color-dependent gain (e.g., gain[c]) to each input pixel and using shared motion and luma tables, separate motion and luma tables are provided for each color components. For instance, as shown in FIG. 26, the motion tables 304 may include a motion table 304a corresponding to a first color, a motion table 304b corresponding to a second color, and a motion table 304c corresponding to an nth color, wherein n depends on the number of colors present in the raw image data. Similarly, the luma tables 306 may include a luma table 306a corresponding to the first color, a luma table 306b corresponding to the second color, and the motion table 304c corresponding to the nth color. Thus, in an embodiment where the raw image data is Bayer image data, three motion and luma tables may be provided for each of the red, blue, and green color components. As discussed below, the selection of filtering coefficients K and attenuation factors may depend on the motion and luma table selected for the current color (e.g., the color of the current input pixel).

[0186] A method 338 illustrating a further embodiment for temporal filtering using color-dependent motion and luma tables is shown in FIG. 27. As will be appreciated, the various calculations and formulas that may be employed by the method 338 may be similar to the embodiment shown in FIG. 23, but with a particular motion and luma table being selected for each color, or similar to the embodiment shown in FIG. 25, but replacing the use of the color dependent gain[c] with the selection of a color-dependent motion and luma table.

[0187] Beginning at step 339, a current pixel x(t) located at spatial location (j,i) of a current frame of image data is received by the temporal filtering system 336 (FIG. 26). At step 340, a motion delta value d(t) is determined for the current pixel x(t) based at least partially upon one or more collocated reference pixels (e.g., r(t−1)) from a previous frame of the image data (e.g., the image frame immediately preceding the current frame). Step 340 may be similar to the step 317 of FIG. 23, and may utilize the operation shown in Equation 1 above.

[0188] Next, at step 341, a motion table lookup index may be determined using the motion delta value d(t) and a motion history input value h(t−1) corresponding to the spatial location (j,i) from the previous frame (e.g., corresponding to the collocated reference pixel r(t−1)). Thereafter, at step 342, a first filter coefficient K may be selected from one of the available motion tables (e.g., 304a, 304b, 304c) based upon the color of the current input pixel. For instance, one the appropriate motion table is identified, the first filter coefficient K may be selected using the motion table lookup index determined in step 341.

[0189] After selecting the first filter coefficient K, a luma table corresponding to the current color is selected and an attenuation factor is selected from the selected luma table based upon the value of the current pixel x(t), as shown at step 343. Thereafter, at step 344, a second filter coefficient K′ is determined based upon the attenuation factor (from step 343) and the first filter coefficient K (step 342). Next, at step 345, a temporally filtered output value y(t) corresponding to the current input pixel x(t) is determined based upon the second filter coefficient K′ (from step 344), the value of the collocated reference pixel r(t−1), and the value of the input pixel x(t). While the technique shown in FIG. 27 may be more costly to implement (e.g., due to the memory needed for storing additional motion and luma tables), it may, in some instances, offer further improvements with regard to ghosting artifacts and consistency across different colors after auto-white balance gains.

[0190] In accordance with further embodiments, the temporal filtering process provided by the temporal filter 298 may utilize a combination of color-dependent gains and color-specific motion and/or luma tables for applying temporal filtering to the input pixels. For instance, in one such embodiment, a single motion table may be provided for all color components, and the motion table lookup index for selecting the first filtering coefficient (K) from the motion table may be determined based upon a color dependent gain (e.g., as shown in FIG. 25, steps 331-332), while the luma table lookup index may not have a color dependent gain applied thereto, but may be used to select the brightness attenuation factor from one of multiple luma tables depending upon the color of the current input pixel (e.g., as shown in FIG. 27, step 343). Alternatively, in another embodiment, multiple motion tables may be provided and a motion table lookup index (without a color dependent gain applied) may be used to select the first filtering coefficient (K) from a motion table corresponding to the color of the current input pixel (e.g., as shown in FIG. 27, step 342), while a single luma table may be provided for all color components, and wherein the luma table lookup index for selecting the brightness attenuation factor may be determined based upon a color dependent gain (e.g., as shown in FIG. 25, steps 333-334). Further, in one embodiment where a Bayer color filter array is utilized, one motion table and/or luma table may be provided for each of the red (R) and blue (B) color components, while a common motion table and/or luma table may be provided for both green color components (Gr and Gb).

[0191] The output of the temporal filter 298 may subsequently be sent to the binning compensation filter (BCF) 300, which may be configured to process the image pixels to compensate for non-linear placement (e.g., uneven spatial distribution) of the color samples due to binning by the image sensor(s) 90a or 90b, such that subsequent image processing operations in the ISP pipe logic 82 (e.g., demosaicing, etc.) that depend on linear placement of the color samples can operate correctly. For example, referring now to FIG. 28, a full resolution sample 346 of Bayer image data is depicted. This may represent a full resolution sample raw image data captured by the image sensor 90a (or 90b) coupled to the ISP front-end processing logic 80.

[0192] As will be appreciated, under certain image capture conditions, it may be not be practical to send the full resolution image data captured by the image sensor 90a to the ISP circuitry 32 for processing. For instance, when capturing video data, in order to preserve the appearance of a fluid moving image from the perspective of the human eye, a frame rate of at least approximately 30 frames per second may be desired. However, if the amount of pixel data contained in each frame of a full resolution sample exceeds the processing capabilities of the ISP circuitry 32 when sampled at 30 frames per second, binning compensation filtering may be applied in conjunction with binning by the image sensor 90a to reduce the resolution of the image signal while also improving signal-to-noise ratio. For instance, as discussed above, various binning techniques, such as 2×2 binning, may be applied to produce a “binned” raw image pixel by averaging the values of surrounding pixels in the active region 280 of the raw frame 278.

[0193] Referring to FIG. 29, an embodiment of the image sensor 90a that may be configured to bin the full resolution image data 346 of FIG. 28 to produce corresponding binned raw image data 358 shown in FIG. 30 is illustrated in accordance with one embodiment. As shown, the image sensor 90a may capture the full resolution raw image data 346. Binning logic 357 may be configured to apply binning to the full resolution raw image data 346 to produce the binned raw image data 358, which may be provided to the ISP front-end processing logic 80 using the sensor interface 94a which, as discussed above, may be an SMIA interface or any other suitable parallel or serial camera interfaces.

[0194] As illustrated in FIG. 30, the binning logic 357 may apply 2×2 binning to the full resolution raw image data 346. For example, with regard to the binned image data 358, the pixels 350, 352, 354, and 356 may form a Bayer pattern and may be determined by averaging the values of the pixels from the full resolution raw image data 346. For instance, referring to both FIGS. 28 and 30, the binned Gr pixel 350 may be determined as the average or mean of the full resolution Gr pixels 350a-350d. Similarly, the binned R pixel 352 may be determined as the average of the full resolution R pixels 352a-352d, the binned B pixel 354 may be determined as the average of the full resolution B pixels 354a-354d, and the binned Gb pixel 356 may be determined as the average of the full resolution Gb pixels 356a-356d. Thus, in the present embodiment, 2×2 binning may provide a set of four full resolution pixels including an upper left (e.g., 350a), upper right (e.g., 350b), lower left (e.g., 350c), and lower right (e.g., 350d) pixel that are averaged to derive a binned pixel located at the center of a square formed by the set of four full resolution pixels. Accordingly, the binned Bayer block 348 shown in FIG. 30 contains four “superpixels” that represent the 16 pixels contained in the Bayer blocks 348a-348d of FIG. 28.

[0195] In addition to reducing spatial resolution, binning also offers the added advantage of reducing noise in the image signal. For instance, whenever an image sensor (e.g., 90a) is exposed to a light signal, there may be a certain amount of noise, such as photon noise, associated with the image. This noise may be random or systematic and it also may come from multiple sources. Thus, the amount of information contained in an image captured by the image sensor may be expressed in terms of a signal-to-noise ratio. For example, every time an image is captured by an image sensor 90a and transferred to a processing circuit, such as the ISP circuitry 32, there may be some degree of noise in the pixels values because the process of reading and transferring the image data inherently introduces “read noise” into the image signal. This “read noise” may be random and is generally unavoidable. By using the average of four pixels, noise, (e.g., photon noise) may generally be reduced irrespective of the source of the noise.

[0196] Thus, when considering the full resolution image data 346 of FIG. 28, each Bayer pattern (2×2 block) 348a-348d contains 4 pixels, each of which contains a signal and noise component. If each pixel in, for example, the Bayer block 348a, is read separately, then four signal components and four noise components are present. However, by applying binning, as shown in FIGS. 28 and 30, such that four pixels (e.g., 350a, 350b, 350c, 350d) may be represented by a single pixel (e.g., 350) in the binned image data, the same area occupied by the four pixels in the full resolution image data 346 may be read as a single pixel with only one instance of a noise component, thus improving signal-to-noise ratio.

[0197] Further, while the present embodiment depicts the binning logic 357 of FIG. 29 as being configured to apply a 2×2 binning process, it should be appreciated that the binning logic 357 may be configured to apply any suitable type of binning process, such as 3×3 binning, vertical binning, horizontal binning, and so forth. In some embodiments, the image sensor 90a may be configured to select between different binning modes during the image capture process. Additionally, in further embodiments, the image sensor 90a may also be configured to apply a technique that may be referred to as “skipping,” wherein instead of average pixel samples, the logic 357 selects only certain pixels from the full resolution data 346 (e.g., every other pixel, every 3 pixels, etc.) to output to the ISP front-end 80 for processing. Further, while only the image sensor 90a is shown in FIG. 29, it should be appreciated that the image sensor 90b may be implemented in a similar manner.

[0198] As also depicted in FIG. 30, one effect of the binning process is that the spatial sampling of the binned pixels may not be equally spaced. This spatial distortion may, in some systems, result in aliasing (e.g., jagged edges), which is generally not desirable. Further, because certain image processing steps in the ISP pipe logic 82 may depend upon on the linear placement of the color samples in order to operate correctly, the binning compensation filter (BCF) 300 may be applied to perform re-sampling and re-positioning of the binned pixels such that the binned pixels are spatially evenly distributed. That is, the BCF 300 essentially compensates for the uneven spatial distribution (e.g., shown in FIG. 30) by re-sampling the position of the samples (e.g., pixels). For instance, FIG. 31 illustrates a re-sampled portion of binned image data 360 after being processed by the BCF 300, wherein the Bayer block 361 containing the evenly distributed re-sampled pixels 362, 363, 364, and 365 correspond to the binned pixels 350, 352, 354, and 356, respectively, of the binned image data 358 from FIG. 30. Additionally, in an embodiment that utilizes skipping (e.g., instead of binning), as mentioned above, the spatial distortion shown in FIG. 30 may not be present. In this case, the BCF 300 may function as a low pass filter to reduce artifacts (e.g., aliasing) that may result when skipping is employed by the image sensor 90a.

[0199] FIG. 32 shows a block diagram of the binning compensation filter 300 in accordance with one embodiment. The BCF 300 may include binning compensation logic 366 that may process binned pixels 358 to apply horizontal and vertical scaling using horizontal scaling logic 368 and vertical scaling logic 370, respectively, to re-sample and re-position the binned pixels 358 so that they are arranged in a spatially even distribution, as shown in FIG. 31. In one embodiment, the scaling operation(s) performed by the BCF 300 may be performed using horizontal and vertical multi-tap polyphase filtering. For instance, the filtering process may include selecting the appropriate pixels from the input source image data (e.g., the binned image data 358 provided by the image sensor 90a), multiplying each of the selected pixels by a filtering coefficient, and summing up the resulting values to form an output pixel at a desired destination.

[0200] The selection of the pixels used in the scaling operations, which may include a center pixel and surrounding neighbor pixels of the same color, may be determined using separate differential analyzers 372, one for vertical scaling and one for horizontal scaling. In the depicted embodiment, the differential analyzers 372 may be digital differential analyzers (DDAs) and may be configured to control the current output pixel position during the scaling operations in the vertical and horizontal directions. In the present embodiment, a first DDA (referred to as 372a) is used for all color components during horizontal scaling, and a second DDA (referred to as 372b) is used for all color components during vertical scaling. By way of example only, the DDA 372 may be provided as a 32-bit data register that contains a 2's-complement fixed-point number having 16 bits in the integer portion and 16 bits in the fraction. The 16-bit integer portion may be used to determine the current position for an output pixel. The fractional portion of the DDA 372 may be used to determine a current index or phase, which may be based the between-pixel fractional position of the current DDA position (e.g., corresponding to the spatial location of the output pixel). The index or phase may be used to select an appropriate set of coefficients from a set of filter coefficient tables 374. Additionally, the filtering may be done per color component using same colored pixels. Thus, the filtering coefficients may be selected based not only on the phase of the current DDA position, but also the color of the current pixel. In one embodiment, 8 phases may be present between each input pixel and, thus, the vertical and horizontal scaling components may utilize 8-deep coefficient tables, such that the high-order 3 bits of the 16-bit fraction portion are used to express the current phase or index. Thus, as used herein, the term “raw image” data or the like shall be understood to refer to multi-color image data that is acquired by a single sensor with a color filter array pattern (e.g., Bayer) overlaying it, those providing multiple color components in one plane. In another embodiment, separate DDAs may be used for each color component. For instance, in such embodiments, the BCF 300 may extract the R, B, Gr, and Gb components from the raw image data and process each component as a separate plane.

[0201] In operation, horizontal and vertical scaling may include initializing the DDA 372 and performing the multi-tap polyphase filtering using the integer and fractional portions of the DDA 372. While performed separately and with separate DDAs, the horizontal and vertical scaling operations are carried out in a similar manner. A step value or step size (DDAStepX for horizontal scaling and DDAStepY for vertical scaling) determines how much the DDA value (currDDA) is incremented after each output pixel is determined, and multi-tap polyphase filtering is repeated using the next currDDA value. For instance, if the step value is less than 1, then the image is up-scaled, and if the step value is greater than 1, the image is downscaled. If the step value is equal to 1, then no scaling occurs. Further, it should be noted that same or different step sizes may be used for horizontal and vertical scaling.

[0202] Output pixels are generated by the BCF 300 in the same order as input pixels (e.g., using the Bayer pattern). In the present embodiment, the input pixels may be classified as being even or odd based on their ordering. For instance, referring to FIG. 33, a graphical depiction of input pixel locations (row 375) and corresponding output pixel locations based on various DDAStep values (rows 376-380) are illustrated. In this example, the depicted row represents a row of red (R) and green (Gr) pixels in the raw Bayer image data. For horizontal filtering purposes, the red pixel at position 0.0 in the row 375 may be considered an even pixel, the green pixel at position 1.0 in the row 375 may be considered an odd pixel, and so forth. For the output pixel locations, even and odd pixels may be determined based on the least significant bit in the fraction portion (lower 16 bits) of the DDA 372. For instance, assuming a DDAStep of 1.25, as shown in row 377, the least significant bit corresponds to the bit 14 of the DDA, as this bit gives a resolution of 0.25. Thus, the red output pixel at the DDA position (currDDA) 0.0 may be considered an even pixel (the least significant bit, bit 14, is 0), the green output pixel at currDDA 1.0 (bit 14 is 1), and so forth. Further, while FIG. 33 is discussed with respect to filtering in the horizontal direction (using DDAStepX), it should be understood that the determination of even and odd input and output pixels may be applied in the same manner with respect to vertical filtering (using DDAStepY). In other embodiments, the DDAs 372 may also be used to track locations of the input pixels (e.g., rather than track the desired output pixel locations). Further, it should be appreciated that DDAStepX and DDAStepY may be set to the same or different values. Further, assuming a Bayer pattern is used, it should be noted that the starting pixel used by the BCF 300 could be any one of a Gr, Gb, R, or B pixel depending, for instance, on which pixel is located at a corner within the active region 280.

[0203] With this in mind, the even/odd input pixels are used to generate the even/odd output pixels, respectively. Given an output pixel location alternating between even and odd position, a center source input pixel location (referred to herein as “currPixel”) for filtering purposes is determined by the rounding the DDA to the closest even or odd input pixel location for even or odd output pixel locations (based on DDAStepX), respectively. In an embodiment where the DDA 372a is configured to use 16 bits to represent an integer and 16 bits to represent a fraction, currPixel may be determined for even and odd currDDA positions using Equations 6a and 6b below:

[0204] Even output pixel locations may be determined based on bits [31:16] of:
(currDDA+1.0)&0xFFFE.0000  (6a)

[0205] Odd output pixel locations may be determined based on bits [31:16] of:
(currDDA)|0x0001.0000  (6b)
Essentially, the above equations present a rounding operation, whereby the even and odd output pixel positions, as determined by currDDA, are rounded to the nearest even and odd input pixel positions, respectively, for the selection of currPixel.

[0206] Additionally, a current index or phase (currIndex) may also be determined at each currDDA position. As discussed above, the index or phase values represent the fractional between-pixel position of the output pixel position relative to the input pixel positions. For instance, in one embodiment, 8 phases may be defined between each input pixel position. For instance, referring again to FIG. 33, 8 index values 0-7 are provided between the first red input pixel at position 0.0 and the next red input pixel at position 2.0. Similarly, 8 index values 0-7 are provided between the first green input pixel at position 1.0 and the next green input pixel at position 3.0. In one embodiment, the currIndex values may be determined in accordance with Equations 7a and 7b below for even and odd output pixel locations, respectively:

[0207] Even output pixel locations may be determined based on bits [16:14] of:
(currDDA+0.125)  (7a)

[0208] Odd output pixel locations may be determined based on bits [16:14] of:
(currDDA+1.125)  (7b)
For the odd positions, the additional 1 pixel shift is equivalent to adding an offset of four to the coefficient index for odd output pixel locations to account for the index offset between different color components with respect to the DDA 372.

[0209] Once currPixel and currIndex have been determined at a particular currDDA location, the filtering process may select one or more neighboring same-colored pixels based on currPixel (the selected center input pixel). By way of example, in an embodiment where the horizontal scaling logic 368 includes a 5-tap polyphase filter and the vertical scaling logic 370 includes a 3-tap polyphase filter, two same-colored pixels on each side of currPixel in the horizontal direction may be selected for horizontal filtering (e.g., −2, −1, 0, +1, +2), and one same-colored pixel on each side of currPixel in the vertical direction may be selected for vertical filtering (e.g., −1, 0, +1). Further, currIndex may be used as a selection index to select the appropriate filtering coefficients from the filter coefficients table 374 to apply to the selected pixels. For instance, using the 5-tap horizontal/3-tap vertical filtering embodiment, five 8-deep tables may be provided for horizontal filtering, and three 8-deep tables may be provided for vertical filtering. Though illustrated as part of the BCF 300, it should be appreciated that the filter coefficient tables 374 may, in certain embodiments, be stored in a memory that is physically separate from the BCF 300, such as the memory 108.

[0210] Before discussing the horizontal and vertical scaling operations in further detail, Table 4 below shows examples of how currPixel and currIndex values, as determined based on various DDA positions using different DDAStep values (e.g., could apply to DDAStepX or DDAStepY).

[0211]

TABLE 4
Binning Compensation Filter - DDA Examples of currPixel and currIndex calculation
Output	DDA		DDA		DDA		DDA
Pixel	Step	1.25	Step	1.5	Step	1.75	Step	2.0
(Even or	curr	curr	curr	curr	curr	curr	curr	curr	curr	curr	curr	curr
Odd)	DDA	Index	Pixel	DDA	Index	Pixel	DDA	Index	Pixel	DDA	Index	Pixel
0	0.0	0	0	0.0	0	0	0.0	0	0	0.0	0	0
1	1.25	1	1	1.5	2	1	1.75	3	1	2	4	3
0	2.5	2	2	3	4	4	3.5	6	4	4	0	4
1	3.75	3	3	4.5	6	5	5.25	1	5	6	4	7
0	5	4	6	6	0	6	7	4	8	8	0	8
1	6.25	5	7	7.5	2	7	8.75	7	9	10	4	11
0	7.5	6	8	9	4	10	10.5	2	10	12	0	12
1	8.75	7	9	10.5	6	11	12.25	5	13	14	4	15
0	10	0	10	12	0	12	14	0	14	16	0	16
1	11.25	1	11	13.5	2	13	15.75	3	15	18	4	19
0	12.5	2	12	15	4	16	17.5	6	18	20	0	20
1	13.75	3	13	16.5	6	17	19.25	1	19	22	4	23
0	15	4	16	18	0	18	21	4	22	24	0	24
1	16.25	5	17	19.5	2	19	22.75	7	23	26	4	27
0	17.5	6	18	21	4	22	24.5	2	24	28	0	28
1	18.75	7	19	22.5	6	23	26.25	5	27	30	4	31
0	20	0	20	24	0	24	28	0	28	32	0	32

[0212] To provide an example, let us assume that a DDA step size (DDAStep) of 1.5 is selected (row 378 of FIG. 33), with the current DDA position (currDDA) beginning at 0, indicating an even output pixel position. To determine currPixel, Equation 6a may be applied, as shown below:

[0213] $\mathrm{currDDA}=0.0\left(\mathrm{even}\right)$ $\begin{array}{c}\begin{array}{c}0000000000000001.0000000000000000\\ \left(\mathrm{AND}\right)1111111111111110.0000000000000000\end{array}\begin{array}{c}\left(\mathrm{currDDA}+1.0\right)\\ \left(0\times \mathrm{FFFE}\mathrm{.0000}\right)\end{array}\\ =\underset{\_}{0000000000000000}\mathrm{.0000}000000000000\end{array}$ $\mathrm{currPixel}\left(\mathrm{determined}as\mathrm{bits}\left[31\text{:}16\right]\mathrm{of}\mathrm{the}\mathrm{result}\right)=0;$
Thus, at the currDDA position 0.0 (row 378), the source input center pixel for filtering corresponds to the red input pixel at position 0.0 of row 375.

[0214] To determine currIndex at the even currDDA 0.0, Equation 7a may be applied, as shown below:

[0215] $\mathrm{currDDA}=0.0\left(\mathrm{even}\right)$ $\begin{array}{c}\\ +\\ =\end{array}\begin{array}{cc}0000000000000000.0000000000000000& \left(\mathrm{currDDA}\right)\\ 0000000000000000.0010000000000000& \left(0.125\right)\\ 000000000000000\underset{\_}{0.00}10000000000000& \end{array}$ $\mathrm{currIndex}\left(\mathrm{determined}\mathrm{as}\mathrm{bits}\left[16\text{:}14\right]\mathrm{of}\mathrm{the}\mathrm{result}\right)=\left[000\right]=0;$
Thus, at the currDDA position 0.0 (row 378), a currIndex value of 0 may be used to select filtering coefficients from the filter coefficients table 374.

[0216] Accordingly, filtering (which may be vertical or horizontal depending on whether DDAStep is in the X (horizontal) or Y (vertical) direction) may applied based on the determined currPixel and currIndex values at currDDA 0.0, and the DDA 372 is incremented by DDAStep (1.5), and the next currPixel and currIndex values are determined. For instance, at the next currDDA position 1.5 (an odd position), currPixel may be determined using Equation 6b as follows:

[0217] $\mathrm{currDDA}=0.0\left(\mathrm{odd}\right)$ $\left(\mathrm{OR}\right)\begin{array}{c}\begin{array}{c}\\ \end{array}\\ =\end{array}\begin{array}{cc}0000000000000001.1000000000000000& \left(\mathrm{currDDA}\right)\\ 0000000000000001.0000000000000000& \left(0\times 0001.0000\right)\\ \underset{\_}{0000000000000001}\mathrm{.1000}000000000000& \end{array}$ $\mathrm{currPixel}\left(\mathrm{determined}\mathrm{as}\mathrm{bits}\left[31\text{:}16\right]\mathrm{of}\mathrm{the}\mathrm{result}\right)=1;$
Thus, at the currDDA position 1.5 (row 378), the source input center pixel for filtering corresponds to the green input pixel at position 1.0 of row 375.

[0218] Further, currIndex at the odd currDDA 1.5 may be determined using Equation 7b, as shown below:

[0219] $\mathrm{currDDA}=1.5\left(\mathrm{odd}\right)$ $\begin{array}{ccc}& 0000000000000001.1000000000000000& \left(\mathrm{currDDA}\right)\\ +& 0000000000000001.0010000000000000& \left(1.125\right)\\ =& 000000000000001\underset{\_}{0.10}10000000000000& \end{array}$ $\mathrm{currIndex}\left(\mathrm{determined}\mathrm{as}\mathrm{bits}\left[16\text{:}14\right]\mathrm{of}\mathrm{the}\mathrm{result}\right)=\left[010\right]=2;$
Thus, at the currDDA position 1.5 (row 378), a currIndex value of 2 may be used to select the appropriate filtering coefficients from the filter coefficients table 374. Filtering (which may be vertical or horizontal depending on whether DDAStep is in the X (horizontal) or Y (vertical) direction) may thus be applied using these currPixel and currIndex values.

[0220] Next, the DDA 372 is incremented again by DDAStep (1.5), resulting in a currDDA value of 3.0. The currPixel corresponding to currDDA 3.0 may be determined using Equation 6a, as shown below:

[0221] $\mathrm{currDDA}=3.0\left(\mathrm{even}\right)$ $\begin{array}{c}\begin{array}{c}0000000000000100.0000000000000000\\ \left(\mathrm{AND}\right)1111111111111110.0000000000000000\end{array}\begin{array}{c}\left(\mathrm{currDDA}+1.0\right)\\ \left(0\times \mathrm{FFFE}\mathrm{.0000}\right)\end{array}\\ =\underset{\_}{0000000000000100}\mathrm{.0000}000000000000\end{array}$ $\mathrm{currPixel}\left(\mathrm{determined}\mathrm{as}\mathrm{bits}\left[31\text{:}16\right]\mathrm{of}\mathrm{the}\mathrm{result}\right)=4;$
Thus, at the currDDA position 3.0 (row 378), the source input center pixel for filtering corresponds to the red input pixel at position 4.0 of row 375.

[0222] Next, currIndex at the even currDDA 3.0 may be determined using Equation 7a, as shown below:

[0223] $\mathrm{currDDA}=3.0\left(\mathrm{even}\right)$ $\begin{array}{ccc}& 0000000000000011.0000000000000000& \left(\mathrm{currDDA}\right)\\ +& 0000000000000000.0010000000000000& \left(0.125\right)\\ =& 000000000000001\underset{\_}{1.00}10000000000000& \end{array}$ $\mathrm{currIndex}\left(\mathrm{determined}\mathrm{as}\mathrm{bits}\left[16:14\right]\mathrm{of}\mathrm{the}\mathrm{result}\right)=\left[100\right]=4;$
Thus, at the currDDA position 3.0 (row 378), a currIndex value of 4 may be used to select the appropriate filtering coefficients from the filter coefficients table 374. As will be appreciated, the DDA 372 may continue to be incremented by DDAStep for each output pixel, and filtering (which may be vertical or horizontal depending on whether DDAStep is in the X (horizontal) or Y (vertical) direction) may be applied using the currPixel and currIndex determined for each currDDA value.

[0224] As discussed above, currIndex may be used as a selection index to select the appropriate filtering coefficients from the filter coefficients table 374 to apply to the selected pixels. The filtering process may include obtaining the source pixel values around the center pixel (currPixel), multiplying each of the selected pixels by the appropriate filtering coefficients selected from the filter coefficients table 374 based on currIndex, and summing the results to obtain a value of the output pixel at the location corresponding to currDDA. Further, because the present embodiment utilizes 8 phases between same colored pixels, using the 5-tap horizontal/3-tap vertical filtering embodiment, five 8-deep tables may be provided for horizontal filtering, and three 8-deep tables may be provided for vertical filtering. In one embodiment, each of the coefficient table entries may include a 16-bit 2's complement fixed point number with 3 integer bits and 13 fraction bits.

[0225] Further, assuming a Bayer image pattern, in one embodiment, the vertical scaling component may include four separate 3-tap polyphase filters, one for each color component: Gr, R, B, and Gb. Each of the 3-tap filters may use the DDA 372 to control the stepping of the current center pixel and the index for the coefficients, as described above. Similarly, the horizontal scaling components may include four separate 5-tap polyphase filters, one for each color component: Gr, R, B, and Gb. Each of the 5-tap filters may use the DDA 372 to control the stepping (e.g., via DDAStep) of the current center pixel and the index for the coefficients. It should be understood however, that fewer or more taps could be utilized by the horizontal and vertical scalars in other embodiments.

[0226] For boundary cases, the pixels used in the horizontal and vertical filtering process may depend upon the relationship of the current DDA position (currDDA) relative to a frame border (e.g., border defined by the active region 280 in FIG. 19). For instance, in horizontal filtering, if the currDDA position, when compared to the position of the center input pixel (SrcX) and the width (SrcWidth) of the frame (e.g., width 290 of the active region 280 of FIG. 19) indicates that the DDA 372 is close to the border such that there are not enough pixels to perform the 5-tap filtering, then the same-colored input border pixels may be repeated. For instance, if the selected center input pixel is at the left edge of the frame, then the center pixel may be replicated twice for horizontal filtering. If the center input pixel is near the left edge of the frame such that only one pixel is available between the center input pixel and the left edge, then, for horizontal filtering purposes, the one available pixel is replicated in order to provide two pixel values to the left of the center input pixel. Further, the horizontal scaling logic 368 may be configured such that the number of input pixels (including original and replicated pixels) cannot exceed the input width. This may be expressed as follows:
StartX=(((DDAInitX+0x0001.0000)&0xFFFE.0000)>>16)
EndX=(((DDAInitX+DDAStepX*(BCFOutWidth−1))|0x0001.0000)>>16)
EndX−StartX<=SrcWidth−1
wherein, DDAInitX represents the initial position of the DDA 372, DDAStepX represents the DDA step value in the horizontal direction, and BCFOutWidth represents the width of the frame output by the BCF 300.

[0227] For vertical filtering, if the currDDA position, when compared to the position of the center input pixel (SrcY) and the width (SrcHeight) of the frame (e.g., width 290 of the active region 280 of FIG. 19) indicates that the DDA 372 is close to the border such that there are not enough pixels to perform the 3-tap filtering, then the input border pixels may be repeated. Further, the vertical scaling logic 370 may be configured such that the number of input pixels (including original and replicated pixels) cannot exceed the input height. This may be expressed as follows:
StartY=(((DDAInitY+0x0001.0000)&0xFFFE.0000)>>16)
EndY=(((DDAInitY+DDAStepY*(BCFOutHeight−1))|0x0001.0000)>>16)
EndY−StartY<=SrcHeight−1
wherein, DDAInitY represents the initial position of the DDA 372, DDAStepY represents the DDA step value in the vertical direction, and BCFOutHeight represents the width of the frame output by the BCF 300.

[0228] Referring now to FIG. 34, a flow chart depicting a method 382 for applying binning compensation filtering to image data received by the front-end pixel processing unit 130 in accordance with an embodiment. It will be appreciated that the method 382 illustrated in FIG. 34 may apply to both vertical and horizontal scaling. Beginning at step 383 the DDA 372 is initialized and a DDA step value (which may correspond to DDAStepX for horizontal scaling and DDAStepY for vertical scaling) is determined. Next, at step 384, a current DDA position (currDDA), based on DDAStep, is determined. As discussed above, currDDA may correspond to an output pixel location. Using currDDA, the method 382 may determine a center pixel (currPixel) from the input pixel data that may be used for binning compensation filtering to determine a corresponding output value at currDDA, as indicated at step 385. Subsequently, at step 386, an index corresponding to currDDA (currIndex) may be determined based on the fractional between-pixel position of currDDA relative to the input pixels (e.g., row 375 of FIG. 33). By way of example, in an embodiment where the DDA includes 16 integer bits and 16 fraction bits, currPixel may be determined in accordance with Equations 6a and 6b, and currIndex may be determined in accordance with Equations 7a and 7b, as shown above. While the 16 bit integer/16 bit fraction configuration is described herein as one example, it should be appreciated that other configurations of the DDA 372 may be utilized in accordance with the present technique. By way of example, other embodiments of the DDA 372 may be configured to include a 12 bit integer portion and 20 bit fraction portion, a 14 bit integer portion and 18 bit fraction portion, and so forth.

[0229] Once currPixel and currIndex are determined, same-colored source pixels around currPixel may be selected for multi-tap filtering, as indicated by step 387. For instance, as discussed above, one embodiment may utilize 5-tap polyphase filtering in the horizontal direction (e.g., selecting 2 same-colored pixels on each side of currPixel) and may utilize 3-tap polyphase filtering in the vertical direction (e.g., selecting 1 same-colored pixel on each side of currPixel). Next, at step 388, once the source pixels are selected, filtering coefficients may be selected from the filter coefficients table 374 of the BCF 300 based upon currIndex.

[0230] Thereafter, at step 389, filtering may be applied to the source pixels to determine the value of an output pixel corresponding to the position represented by currDDA. For instance, in one embodiment, the source pixels may be multiplied by their respective filtering coefficients, and the results may be summed to obtain the output pixel value. The direction in which filtering is applied at step 389 may be vertical or horizontal depending on whether DDAStep is in the X (horizontal) or Y (vertical) direction. Finally, at step 263, the DDA 372 is incremented by DDAStep at step 390, and the method 382 returns to step 384, whereby the next output pixel value is determined using the binning compensation filtering techniques discussed herein.

[0231] Referring to FIG. 35, the step 385 for determining currPixel from the method 382 is illustrated in more detail in accordance with one embodiment. For instance, step 385 may include the sub-step 392 of determining whether the output pixel location corresponding to currDDA (from step 384) is even or odd. As discussed above, an even or odd output pixel may be determined based on the least significant bit of currDDA based on DDAStep. For instance, given a DDAStep of 1.25, a currDDA value of 1.25 may be determined as odd, since the least significant bit (corresponding to bit 14 of the fractional portion of the DDA 372) has a value of 1. For a currDDA value of 2.5, bit 14 is 0, thus indicating an even output pixel location.

[0232] At decision logic 393, a determination is made as to whether the output pixel location corresponding to currDDA is even or odd. If the output pixel is even, decision logic 393 continues to sub-step 394, wherein currPixel is determined by incrementing the currDDA value by 1 and rounding the result to the nearest even input pixel location, as represented by Equation 6a above. If the output pixel is odd, then decision logic 393 continues to sub-step 395, wherein currPixel is determined by rounding the currDDA value to the nearest odd input pixel location, as represented by Equation 6b above. The currPixel value may then be applied to step 387 of the method 382 to select source pixels for filtering, as discussed above.

[0233] Referring also to FIG. 36, the step 386 for determining currIndex from the method 382 is illustrated in more detail in accordance with one embodiment. For instance, step 386 may include the sub-step 396 of determining whether the output pixel location corresponding to currDDA (from step 384) is even or odd. This determination may be performed in a similar manner as step 392 of FIG. 35. At decision logic 397, a determination is made as to whether the output pixel location corresponding to currDDA is even or odd. If the output pixel is even, decision logic 397 continues to sub-step 398, wherein currIndex is determined by incrementing the currDDA value by one index step determining currIndex based on the lowest order integer bit and the two highest order fraction bits of the DDA 372. For instance, in an embodiment wherein 8 phases are provided between each same-colored pixel, and wherein the DDA includes 16 integer bits and 16 fraction bits, one index step may correspond to 0.125, and currIndex may be determined based on bits [16:14] of the currDDA value incremented by 0.125 (e.g., Equation 7a). If the output pixel is odd, decision logic 397 continues to sub-step 399, wherein currIndex is determined by incrementing the currDDA value by one index step and one pixel shift, and determining currIndex based on the lowest order integer bit and the two highest order fraction bits of the DDA 372. Thus, in an embodiment wherein 8 phases are provided between each same-colored pixel, and wherein the DDA includes 16 integer bits and 16 fraction bits, one index step may correspond to 0.125, one pixel shift may correspond to 1.0 (a shift of 8 index steps to the next same colored pixel), and currIndex may be determined based on bits [16:14] of the currDDA value incremented by 1.125 (e.g., Equation 7b).

[0234] While the presently illustrated embodiment provides the BCF 300 as a component of the front-end pixel processing unit 130, other embodiments may incorporate the BCF 300 into a raw image data processing pipeline of the ISP pipe 82 which, as discussed further below, may include defective pixel detection/correction logic, gain/offset/compensation blocks, noise reduction logic, lens shading correction logic, and demosaicing logic. Further, in embodiments where the aforementioned defective pixel detection/correction logic, gain/offset/compensation blocks, noise reduction logic, lens shading correction logic do not rely upon the linear placement of the pixels, the BCF 300 may be incorporated with the demosaicing logic to perform binning compensation filtering and reposition the pixels prior to demoasicing, as demosaicing generally does rely upon the even spatial positioning of the pixels. For instance, in one embodiment, the BCF 300 may be incorporated anywhere between the sensor input and the demosaicing logic, with temporal filtering and/or defective pixel detection/correction being applied to the raw image data prior to binning compensation.

[0235] As discussed above the output of the BCF 300, which may be the output FEProcOut (109) having spatially evenly distributed image data (e.g., sample 360 of FIG. 31), may be forwarded to the ISP pipe processing logic 82 for additional processing. However, before shifting the focus of this discussion to the ISP pipe processing logic 82, a more detailed description of various functionalities that may be provided by the statistics processing units (e.g., 122 and 124) that may be implemented in the ISP front-end logic 80 will first be provided.

[0236] Referring back to the general description of the statistics processing units 120 and 122, these units may be configured to collect various statistics about the image sensors that capture and provide the raw image signals (Sif0 and Sif1), such as statistics relating to auto-exposure, auto-white balance, auto-focus, flicker detection, black level compensation, and lens shading correction, and so forth. In doing so, the statistics processing units 120 and 122 may first apply one or more image processing operations to their respective input signals, Sif0 (from Sensor0) and Sif1 (from Sensor1).

[0237] For example, referring to FIG. 37, a more detailed block diagram view of the statistics processing unit 120 associated with Sensor0 (90a) is illustrated in accordance with one embodiment. As shown, the statistics processing unit 120 may include the following functional blocks: defective pixel detection and correction logic 460, black level compensation (BLC) logic 462, lens shading correction logic 464, inverse BLC logic 466, and statistics collection logic 468. Each of these functional blocks will be discussed below. Further, it should be understood that the statistics processing unit 122 associated with Sensor 1 (90b) may be implemented in a similar manner.

[0238] Initially, the output of selection logic 124 (e.g., Sif0 or SifIn0) is received by the front-end defective pixel correction logic 460. As will be appreciated, “defective pixels” may be understood to refer to imaging pixels within the image sensor(s) 90 that fail to sense light levels accurately. Defective pixels may attributable to a number of factors, and may include “hot” (or leaky) pixels, “stuck” pixels, and “dead pixels.” A “hot” pixel generally appears as being brighter than a non-defective pixel given the same amount of light at the same spatial location. Hot pixels may result due to reset failures and/or high leakage. For example, a hot pixel may exhibit a higher than normal charge leakage relative to non-defective pixels, and thus may appear brighter than non-defective pixels. Additionally, “dead” and “stuck” pixels may be the result of impurities, such as dust or other trace materials, contaminating the image sensor during the fabrication and/or assembly process, which may cause certain defective pixels to be darker or brighter than a non-defective pixel, or may cause a defective pixel to be fixed at a particular value regardless of the amount of light to which it is actually exposed. Additionally, dead and stuck pixels may also result from circuit failures that occur during operation of the image sensor. By way of example, a stuck pixel may appear as always being on (e.g., fully charged) and thus appears brighter, whereas a dead pixel appears as always being off.

[0239] The defective pixel detection and correction (DPDC) logic 460 in the ISP front-end logic 80 may correct (e.g., replace defective pixel values) defective pixels before they are considered in statistics collection (e.g., 468). In one embodiment, defective pixel correction is performed independently for each color component (e.g., R, B, Gr, and Gb for a Bayer pattern). Generally, the front-end DPDC logic 460 may provide for dynamic defect correction, wherein the locations of defective pixels are determined automatically based upon directional gradients computed using neighboring pixels of the same color. As will be understand, the defects may be “dynamic” in the sense that the characterization of a pixel as being defective at a given time may depend on the image data in the neighboring pixels. By way of example, a stuck pixel that is always on maximum brightness may not be regarded as a defective pixel if the location of the stuck pixel is in an area of the current image that is dominate by brighter or white colors. Conversely, if the stuck pixel is in a region of the current image that is dominated by black or darker colors, then the stuck pixel may be identified as a defective pixel during processing by the DPDC logic 460 and corrected accordingly.

[0240] The DPDC logic 460 may utilize one or more horizontal neighboring pixels of the same color on each side of a current pixel to determine if the current pixel is defective using pixel-to-pixel directional gradients. If a current pixel is identified as being defective, the value of the defective pixel may be replaced with the value of a horizontal neighboring pixel. For instance, in one embodiment, five horizontal neighboring pixels of the same color that are inside the raw frame 278 (FIG. 19) boundary are used, wherein the five horizontal neighboring pixels include the current pixel and two neighboring pixels on either side. Thus, as illustrated in FIG. 38, for a given color component c and for the current pixel P, horizontal neighbor pixels P0, P1, P2, and P3 may be considered by the DPDC logic 460. It should be noted, however, that depending on the location of the current pixel P, pixels outside the raw frame 278 are not considered when calculating pixel-to-pixel gradients.

[0241] For instance, as shown in FIG. 38, in a “left edge” case 470, the current pixel P is at the leftmost edge of the raw frame 278 and, thus, the neighboring pixels P0 and P1 outside of the raw frame 278 are not considered, leaving only the pixels P, P2, and P3 (N=3). In a “left edge+1” case 472, the current pixel P is one unit pixel away from the leftmost edge of the raw frame 278 and, thus, the pixel P0 is not considered. This leaves only the pixels P1, P, P2, and P3 (N=4). Further, in a “centered” case 474, pixels P0 and P1 on the left side of the current pixel P and pixels P2 and P3 on the right side of the current pixel P are within the raw frame 278 boundary and, therefore, all of the neighboring pixels P0, P1, P2, and P3 (N=5) are considered in calculating pixel-to-pixel gradients. Additionally, similar cases 476 and 478 may be encountered as the rightmost edge of the raw frame 278 is approached. For instance, given the “right edge −1” case 476, the current pixel P is one unit pixel away the rightmost edge of the raw frame 278 and, thus, the pixel P3 is not considered (N=4). Similarly, in the “right edge” case 478, the current pixel P is at the rightmost edge of the raw frame 278 and, thus, both of the neighboring pixels P2 and P3 are not considered (N=3).

[0242] In the illustrated embodiment, for each neighboring pixel (k=0 to 3) within the picture boundary (e.g., raw frame 278), the pixel-to-pixel gradients may be calculated as follows:
G_k=abs(P−P_k), for 0≦k≦3(only for k within the raw frame)  (8)
Once the pixel-to-pixel gradients have been determined, defective pixel detection may be performed by the DPDC logic 460 as follows. First, it is assumed that a pixel is defective if a certain number of its gradients G_k are at or below a particular threshold, denoted by the variable dprTh. Thus, for each pixel, a count (C) of the number of gradients for neighboring pixels inside the picture boundaries that are at or below the threshold dprTh is accumulated. By way of example, for each neighbor pixel inside the raw frame 278, the accumulated count C of the gradients G_k that are at or below the threshold dprTh may be computed as follows:

[0243] $\begin{array}{cc}C=\stackrel{N}{\sum _k}\left(G_k\le \mathrm{dprTh}\right),& \left(9\right)\end{array}$

[0244] for 0≦k≦3 (only for k within the raw frame)

[0245] As will be appreciated, depending on the color components, the threshold value dprTh may vary. Next, if the accumulated count C is determined to be less than or equal to a maximum count, denoted by the variable dprMaxC, then the pixel may be considered defective. This logic is expressed below:
if(C≦dprMaxC), then the pixel is defective.  (10)

[0246] Defective pixels are replaced using a number of replacement conventions. For instance, in one embodiment, a defective pixel may be replaced with the pixel to its immediate left, P1. At a boundary condition (e.g., P1 is outside of the raw frame 278), a defective pixel may replaced with the pixel to its immediate right, P2. Further, it should be understood that replacement values may be retained or propagated for successive defective pixel detection operations. For instance, referring to the set of horizontal pixels shown in FIG. 38, if P0 or P1 were previously identified by the DPDC logic 460 as being defective pixels, their corresponding replacement values may be used for the defective pixel detection and replacement of the current pixel P.

[0247] To summarize the above-discussed defective pixel detection and correction techniques, a flow chart depicting such a process is provided in FIG. 39 and referred to by reference number 480. As shown, process 480 begins at step 482, at which a current pixel (P) is received and a set of neighbor pixels is identified. In accordance with the embodiment described above, the neighbor pixels may include two horizontal pixels of the same color component from opposite sides of the current pixel (e.g., P0, P1, P2, and P3). Next, at step 484, horizontal pixel-to-pixel gradients are calculated with respect to each neighboring pixel within the raw frame 278, as described in Equation 8 above. Thereafter, at step 486, a count C of the number of gradients that are less than or equal to a particular threshold dprTh is determined. As shown at decision logic 488, if C is less than or equal to dprMaxC, then the process 480 continues to step 490, and the current pixel is identified as being defective. The defective pixel is then corrected at step 492 using a replacement value. Additionally, referring back to decision logic 488, if C is greater than dprMaxC, then the process continues to step 494, and the current pixel is identified as not being defective, and its value is not changed.

[0248] It should be noted that the defective pixel detection/correction techniques applied during the ISP front-end statistics processing may be less robust than defective pixel detection/correction that is performed in the ISP pipe logic 82. For instance, as will be discussed in further detail below, defective pixel detection/correction performed in the ISP pipe logic 82 may, in addition to dynamic defect correction, further provide for fixed defect correction, wherein the locations of defective pixels are known a priori and loaded in one or more defect tables. Further, dynamic defect correction may in the ISP pipe logic 82 may also consider pixel gradients in both horizontal and vertical directions, and may also provide for the detection/correction of speckling, as will be discussed below.

[0249] Returning to FIG. 37, the output of the DPDC logic 460 is then passed to the black level compensation (BLC) logic 462. The BLC logic 462 may provide for digital gain, offset, and clipping independently for each color component “c” (e.g., R, B, Gr, and Gb for Bayer) on the pixels used for statistics collection. For instance, as expressed by the following operation, the input value for the current pixel is first offset by a signed value, and then multiplied by a gain.
Y=(X+O[c])×G[c],  (11)
wherein X represents the input pixel value for a given color component c (e.g., R, B, Gr, or Gb), O[c] represents a signed 16-bit offset for the current color component c, and G[c] represents a gain value for the color component c. In one embodiment, the gain G[c] may be a 16-bit unsigned number with 2 integer bits and 14 fraction bits (e.g., 2.14 in floating point representation), and the gain G[c] may be applied with rounding. By way of example only, the gain G[c] may have a range of between 0 to 4× (e.g., 4 times the input pixel value).

[0250] Next, as shown by Equation 12 below, the computed value Y, which is signed, may then be then clipped to a minimum and maximum range:
Y=(Y<min[c])?min[c]:(Y>max[c])?max[c]:Y  (12)

[0251] The variables min[c] and max[c] may represent signed 16-bit “clipping values for the minimum and maximum output values, respectively. In one embodiment, the BLC logic 462 may also be configured to maintain a count of the number of pixels that were clipped above and below maximum and minimum, respectively, per color component.

[0252] Subsequently, the output of the BLC logic 462 is forwarded to the lens shading correction (LSC) logic 464. The LSC logic 464 may be configured to apply an appropriate gain on a per-pixel basis to compensate for drop-offs in intensity, which are generally roughly proportional to the distance from the optical center of the lens 88 of the imaging device 30. As can be appreciated, such drop-offs may be the result of the geometric optics of the lens. By way of example, a lens having ideal optical properties may be modeled as the fourth power of the cosine of the incident angle, cos⁴(θ), referred to as the cos⁴ law. However, because lens manufacturing is not perfect, various irregularities in the lens may cause the optical properties to deviate from the assumed cos⁴ model. For instance, the thinner edged of the lens usually exhibits the most irregularities. Additionally, irregularities in lens shading patterns may also be the result of a microlens array within an image sensor not being perfectly aligned with the color array filter. Further, the infrared (IR) filter in some lenses may cause the drop-off to be illuminant-dependent and, thus, lens shading gains may be adapted depending upon the light source detected.

[0253] Referring to FIG. 40, a three-dimensional profile 496 depicting light intensity versus pixel position for a typical lens is illustrated. As shown, the light intensity near the center 498 of the lens gradually drops off towards the corners or edges 500 of the lens. The lens shading irregularities depicted in FIG. 40 may be better illustrated by FIG. 41, which shows a colored drawing of an image 502 that exhibits drop-offs in light intensity towards the corners and edges. Particularly, it should be noted that the light intensity at the approximate center of the image appears to be brighter than the light intensity at the corners and/or edges of the image.

[0254] In accordance with embodiments of the present techniques, lens shading correction gains may be specified as a two-dimensional grid of gains per color channel (e.g., Gr, R, B, Gb for a Bayer filter). The gain grid points may be distributed at fixed horizontal and vertical intervals within the raw frame 278 (FIG. 19). As discussed above in FIG. 19, the raw frame 278 may include an active region 280 which defines an area on which processing is performed for a particular image processing operation. With regard to the lens shading correction operation, an active processing region, which may be referred to as the LSC region, is defined within the raw frame region 278. As will be discussed below, the LSC region must be completely inside or at the gain grid boundaries, otherwise results may be undefined.

[0255] For instance, referring to FIG. 42, an LSC region 504 and a gain grid 506 that may be defined within the raw frame 278 are shown. The LSC region 504 may have a width 508 and a height 510, and may be defined by an x-offset 512 and a y-offset 514 with respect to the boundary of the raw frame 278. Grid offsets (e.g., grid x-offset 516 and grid y-offset 518) from the base 520 of the grid gains 506 to the first pixel 522 in the LSC region 504 is also provided. These offsets may be within the first grid interval for a given color component. The horizontal (x-direction) and vertical (y-direction) grid point intervals 524 and 526, respectively, may be specified independently for each color channel.

[0256] As discussed above, assuming the use of a Bayer color filter array, 4 color channels of grid gains (R, B, Gr, and Gb) may be defined. In one embodiment, a total of 4K (4096) grid points may be available, and for each color channel, a base address for the start location of grid gains may be provided, such as by using a pointer. Further, the horizontal (524) and vertical (526) grid point intervals may be defined in terms of pixels at the resolution of one color plane and, in certain embodiments, may be provide for grid point intervals separated by a power of 2, such as by 8, 16, 32, 64, or 128, etc., in horizontal and vertical directions. As can be appreciated, by utilizing a power of 2, efficient implementation of gain interpolation using a shift (e.g., division) and add operations may be achieved. Using these parameters, the same gain values can be used even as the image sensor cropping region is changing. For instance, only a few parameters need to be updated to align the grid points to the cropped region (e.g., updating the grid offsets 524 and 526) instead of updating all grid gain values. By way of example only, this may be useful when cropping is used during digital zooming operations. Further, while the gain grid 506 shown in the embodiment of FIG. 42 is depicted as having generally equally spaced grid points, it should be understood that in other embodiments, the grid points may not necessarily be equally spaced. For instance, in some embodiments, the grid points may be distributed unevenly (e.g., logarithmically), such that the grid points are less concentrated in the center of the LSC region 504, but more concentrated towards the corners of the LSC region 504, typically where lens shading distortion is more noticeable.

[0257] In accordance with the presently disclosed lens shading correction techniques, when a current pixel location is located outside of the LSC region 504, no gain is applied (e.g., the pixel is passed unchanged). When the current pixel location is at a gain grid location, the gain value at that particular grid point may be used. However, when a current pixel location is between grid points, the gain may be interpolated using bi-linear interpolation. An example of interpolating the gain for the pixel location “G” on FIG. 43 is provided below.

[0258] As shown in FIG. 43, the pixel G is between the grid points G0, G1, G2, and G3, which may correspond to the top-left, top-right, bottom-left, and bottom-right gains, respectively, relative to the current pixel location G. The horizontal and vertical size of the grid interval is represented by X and Y, respectively. Additionally, ii and jj represent the horizontal and vertical pixel offsets, respectively, relative to the position of the top left gain G0. Based upon these factors, the gain corresponding to the position G may thus be interpolated as follows:

[0259] $\begin{array}{cc}G=\frac{\begin{array}{c}\left(G0\left(Y-\mathrm{jj}\right)\left(X-\mathrm{ii}\right)\right)+\left(G1\left(Y-\mathrm{jj}\right)\left(\mathrm{ii}\right)\right)+\\ \left(G2\left(\mathrm{jj}\right)\left(X-\mathrm{ii}\right)\right)+\left(G3\left(\mathrm{ii}\right)\left(\mathrm{jj}\right)\right)\end{array}}{\mathrm{XY}}& \left(13a\right)\end{array}$
The terms in Equation 13a above may then be combined to obtain the following expression:

[0260] $\begin{array}{cc}G=\frac{\begin{array}{c}G0\left[\mathrm{XY}-X\left(\mathrm{jj}\right)-Y\left(\mathrm{ii}\right)+\left(\mathrm{ii}\right)\left(\mathrm{jj}\right)\right]+\\ G\left[Y\left(\mathrm{ii}\right)-\left(\mathrm{ii}\right)\left(\mathrm{jj}\right)\right]+G2\left[X\left(\mathrm{jj}\right)-\left(\mathrm{ii}\right)\left(\mathrm{jj}\right)\right]+G3\left[\left(\mathrm{ii}\right)\left(\mathrm{jj}\right)\right]\end{array}}{\mathrm{XY}}& \left(13b\right)\end{array}$
In one embodiment, the interpolation method may be performed incrementally, instead of using a multiplier at each pixel, thus reducing computational complexity. For instance, the term (ii)(jj) may be realized using an adder that may be initialized to 0 at location (0, 0) of the gain grid 506 and incremented by the current row number each time the current column number increases by a pixel. As discussed above, since the values of X and Y may be selected as powers of two, gain interpolation may be accomplished using a simple shift operations. Thus, the multiplier is needed only at the grid point G0 (instead of at every pixel), and only addition operations are needed to determine the interpolated gain for the remaining pixels.

[0261] In certain embodiments, the interpolation of gains between the grid points may use 14-bit precision, and the grid gains may be unsigned 10-bit values with 2 integer bits and 8 fractional bits (e.g., 2.8 floating point representation). Using this convention, the gain may have a range of between 0 and 4×, and the gain resolution between grid points may be 1/256.

[0262] The lens shading correction techniques may be further illustrated by the process 528 shown in FIG. 44. As shown, process 528 begins at step 530, at which the position of a current pixel is determined relative to the boundaries of the LSC region 504 of FIG. 42. Next, decision logic 532 determines whether the current pixel position is within the LSC region 504. If the current pixel position is outside of the LSC region 504, the process 528 continues to step 534, and no gain is applied to the current pixel (e.g., the pixel passes unchanged).

[0263] If the current pixel position is within the LSC region 504, the process 528 continues to decision logic 536, at which it is further determined whether the current pixel position corresponds to a grid point within the gain grid 504. If the current pixel position corresponds to a grid point, then the gain value at that grid point is selected and applied to the current pixel, as shown at step 538. If the current pixel position does not correspond to a grid point, then the process 528 continues to step 540, and a gain is interpolated based upon the bordering grid points (e.g., G0, G1, G2, and G3 of FIG. 43). For instance, the interpolated gain may be computed in accordance with Equations 13a and 13b, as discussed above. Thereafter, the process 528 ends at step 542, at which the interpolated gain from step 540 is applied to the current pixel.

[0264] As will be appreciated, the process 528 may be repeated for each pixel of the image data. For instance, as shown in FIG. 45, a three-dimensional profile depicting the gains that may be applied to each pixel position within a LSC region (e.g. 504) is illustrated. As shown, the gain applied at the corners 544 of the image may be generally greater than the gain applied to the center 546 of the image due to the greater drop-off in light intensity at the corners, as shown in FIGS. 40 and 41. Using the presently described lens shading correction techniques, the appearance of light intensity drop-offs in the image may be reduced or substantially eliminated. For instance, FIG. 46 provides an example of how the colored drawing of the image 502 from FIG. 41 may appear after lens shading correction is applied. As shown, compared to the original image from FIG. 41, the overall light intensity is generally more uniform across the image. Particularly, the light intensity at the approximate center of the image may be substantially equal to the light intensity values at the corners and/or edges of the image. Additionally, as mentioned above, the interpolated gain calculation (Equations 13a and 13b) may, in some embodiments, be replaced with an additive “delta” between grid points by taking advantage of the sequential column and row incrementing structure. As will be appreciated, this reduces computational complexity.

[0265] In further embodiments, in addition to using grid gains, a global gain per color component that is scaled as a function of the distance from the image center is used. The center of the image may be provided as an input parameter, and may be estimated by analyzing the light intensity amplitude of each image pixel in the uniformly illuminated image. The radial distance between the identified center pixel and the current pixel, may then be used to obtain a linearly scaled radial gain, G_r, as shown below:
G_r=G_p[c]×R,  (14)
wherein G_p[e] represents a global gain parameter for each color component c (e.g., R, B, Gr, and Gb components for a Bayer pattern), and wherein R represents the radial distance between the center pixel and the current pixel.

[0266] With reference to FIG. 47, which shows the LSC region 504 discussed above, the distance R may be calculated or estimated using several techniques. As shown, the pixel C corresponding to the image center may have the coordinates (x₀, y₀), and the current pixel G may have the coordinates (x_G, y_G). In one embodiment, the LSC logic 464 may calculate the distance R using the following equation:
R=√{square root over (x_G−x₀)²+(y_G−y₀)²)}{square root over (x_G−x₀)²+(y_G−y₀)²)}  (15)

[0267] In another embodiment, a simpler estimation formula, shown below, may be utilized to obtain an estimated value for R.
R=α×max(abs(x_G−x₀),abs(y_G−y₀))+β×min(abs(x_G−x₀),abs(y_G−y₀))  (16)
In Equation 16, the estimation coefficients α and β may be scaled to 8-bit values. By way of example only, in one embodiment, a may be equal to approximately 123/128 and 0 may be equal to approximately 51/128 to provide an estimated value for R. Using these coefficient values, the largest error may be approximately 4%, with a median error of approximately 1.3%. Thus, even though the estimation technique may be somewhat less accurate than utilizing the calculation technique in determining R (Equation 15), the margin of error is low enough that the estimated values or R are suitable for determining radial gain components for the present lens shading correction techniques.

[0268] The radial gain G_r may then be multiplied by the interpolated grid gain value G (Equations 13a and 13b) for the current pixel to determine a total gain that may be applied to the current pixel. The output pixel Y is obtained by multiplying the input pixel value X with the total gain, as shown below:
Y=(G×G_r×X)  (17)
Thus, in accordance with the present technique, lens shading correction may be performed using only the interpolated gain, both the interpolated gain and the radial gain components. Alternatively, lens shading correction may also be accomplished using only the radial gain in conjunction with a radial grid table that compensates for radial approximation errors. For example, instead of a rectangular gain grid 506, as shown in FIG. 42, a radial gain grid having a plurality of grid points defining gains in the radial and angular directions may be provided. Thus, when determining the gain to apply to a pixel that does not align with one of the radial grid points within the LSC region 504, interpolation may be applied using the four grid points that enclose the pixel to determine an appropriate interpolated lens shading gain.

[0269] Referring to FIG. 48, the use of interpolated and radial gain components in lens shading correction is illustrated by the process 548. It should be noted that the process 548 may include steps that are similar to the process 528, described above in FIG. 44. Accordingly, such steps have been numbered with like reference numerals. Beginning at step 530, the current pixel is received and its location relative to the LSC region 504 is determined. Next, decision logic 532 determines whether the current pixel position is within the LSC region 504. If the current pixel position is outside of the LSC region 504, the process 548 continues to step 534, and no gain is applied to the current pixel (e.g., the pixel passes unchanged). If the current pixel position is within the LSC region 504, then the process 548 may continue simultaneously to step 550 and decision logic 536. Referring first to step 550, data identifying the center of the image is retrieved. As discussed above, determining the center of the image may include analyzing light intensity amplitudes for the pixels under uniform illumination. This may occur during calibration, for instance. Thus, it should be understood that step 550 does not necessarily encompass repeatedly calculating the center of the image for processing each pixel, but may refer to retrieving the data (e.g., coordinates) of previously determined image center. Once the center of the image is identified, the process 548 may continue to step 552, wherein the distance between the image center and the current pixel location (R) is determined. As discussed above, the value of R may be calculated (Equation 15) or estimated (Equation 16). Then, at step 554, a radial gain component G_r may be computed using the distance R and global gain parameter corresponding to the color component of the current pixel (Equation 14). The radial gain component G_r may be used to determine the total gain, as will be discussed in step 558 below.

[0270] Referring back to decision logic 536, a determined whether the current pixel position corresponds to a grid point within the gain grid 504. If the current pixel position corresponds to a grid point, then the gain value at that grid point is determined, as shown at step 556. If the current pixel position does not correspond to a grid point, then the process 548 continues to step 540, and an interpolated gain is computed based upon the bordering grid points (e.g., G0, G1, G2, and G3 of FIG. 43). For instance, the interpolated gain may be computed in accordance with Equations 13a and 13b, as discussed above. Next, at step 558, a total gain is determined based upon the radial gain determined at step 554, as well as one of the grid gains (step 556) or the interpolated gain (540). As can be appreciated, this may depend on which branch decision logic 536 takes during the process 548. The total gain is then applied to the current pixel, as shown at step 560. Again, it should be noted that like the process 528, the process 548 may also be repeated for each pixel of the image data.

[0271] The use of the radial gain in conjunction with the grid gains may offer various advantages. For instance, using a radial gain allows for the use of single common gain grid for all color components. This may greatly reduce the total storage space required for storing separate gain grids for each color component. For instance, in a Bayer image sensor, the use of a single gain grid for each of the R, B, Gr, and Gb components may reduce the gain grid data by approximately 75%. As will be appreciated, this reduction in grid gain data may decrease implementation costs, as grid gain data tables may account for a significant portion of memory or chip area in image processing hardware. Further, depending upon the hardware implementation, the use of a single set of gain grid values may offer further advantages, such as reducing overall chip area (e.g., such as when the gain grid values are stored in an on-chip memory) and reducing memory bandwidth requirements (e.g., such as when the gain grid values are stored in an off-chip external memory).

[0272] Having thoroughly described the functionalities of the lens shading correction logic 464 shown in FIG. 37, the output of the LSC logic 464 is subsequently forwarded to the inverse black level compensation (IBLC) logic 466. The IBLC logic 466 provides gain, offset and clip independently for each color component (e.g., R, B, Gr, and Gb), and generally performs the inverse function to the BLC logic 462. For instance, as shown by the following operation, the value of the input pixel is first multiplied by a gain and then offset by a signed value.
Y=(X×G[c])+O[c],  (18)
wherein X represents the input pixel value for a given color component c (e.g., R, B, Gr, or Gb), O[c] represents a signed 16-bit offset for the current color component c, and G[c] represents a gain value for the color component c. In one embodiment, the gain G[c] may have a range of between approximately 0 to 4× (4 times the input pixel value X). It should be noted that these variables may be the same variables discussed above in Equation 11. The computed value Y may be clipped to a minimum and maximum range using, for example, Equation 12. In one embodiment, the IBLC logic 466 may be configured to maintain a count of the number of pixels that were clipped above and below maximum and minimum, respectively, per color component.

[0273] Thereafter, the output of the IBLC logic 466 is received by the statistics collection block 468, which may provide for the collection of various statistical data points about the image sensor(s) 90, such as those relating to auto-exposure (AE), auto-white balance (AWB), auto-focus (AF), flicker detection, and so forth. A brief overview discussing the significance of AWB, AE, and AF statistics is provided below.

[0274] With regard to white balancing, the image sensor response at each pixel may depend on the illumination source, since the light source is reflected from objects in the image scene. Thus, each pixel value recorded in the image scene is related to the color temperature of the light source. When a white object is illuminated under a low color temperature, it may appear reddish in the captured image. Conversely, a white object that is illuminated under a high color temperature may appear bluish in the captured image. The goal of white balancing is, therefore, to adjust RGB values such that the image appears to the human eye as if it were taken under canonical light. Thus, in the context of imaging statistics relating to white balance, color information about white objects are collected to determine the color temperature of the light source. In general, white balance algorithms may include two main steps. First, the color temperature of the light source is estimated. Second, the estimated color temperature is used to adjust color gain values and/or determine/adjust coefficients of a color correction matrix. Such gains may be a combination of analog and digital image sensor gains, as well as ISP digital gains.

[0275] Next, auto-exposure generally refers to a process of adjusting pixel integration time and gains to control the luminance of the captured image. For instance, auto-exposure may control the amount of light from the scene that is captured by the image sensor(s) by setting the integration time. Further, auto-focus may refer to determining the optimal focal length of the lens in order to substantially optimize the focus of the image. Thus, these various types of statistics, among others, may be determined and collected via the statistics collection block 468. As shown, the output STATS0 of the statistics collection block 468 may be sent to the memory 108 and routed to the control logic 84 or, alternatively, may be sent directly to the control logic 84.

[0276] As discussed above, the control logic 84 may process the collected statistical data to determine one or more control parameters for controlling the imaging device 30 and/or the image processing circuitry 32. For instance, such the control parameters may include parameters for operating the lens of the image sensor 90 (e.g., focal length adjustment parameters), image sensor parameters (e.g., analog and/or digital gains, integration time), as well as ISP pipe processing parameters (e.g., digital gain values, color correction matrix (CCM) coefficients). Additionally, as mentioned above, in certain embodiments, statistical processing may occur at a precision of 8-bits and, thus, raw pixel data having a higher bit-depth may be down-sampled to an 8-bit format for statistics purposes. As discussed above, down-sampling to 8-bits (or any other lower-bit resolution) may reduce hardware size (e.g., area) and also reduce processing complexity, as well as allow for the statistics data to be more robust to noise (e.g., using spatial averaging of the image data).

[0277] Before proceeding with a detailed discussion of the ISP pipe logic 82 downstream from the ISP front-end logic 80, it should understood that the arrangement of various functional logic blocks in the statistics processing units 120 and 122 (e.g., logic blocks 460, 462, 464, 466, and 468) and the ISP front-end pixel processing unit 130 (e.g., logic blocks 298 and 300) are intended to illustrate only one embodiment of the present technique. Indeed, in other embodiments, the logic blocks illustrated herein may be arranged in different ordering, or may include additional logic blocks that may perform additional image processing functions not specifically described herein. Further, it should be understood that the image processing operations performed in the statistics processing units (e.g., 120 and 122), such as lens shading correction, defective pixel detection/correction, and black level compensation, are performed within the statistics processing units for the purposes of collecting statistical data. Thus, processing operations performed upon the image data received by the statistical processing units are not actually reflected in the image signal 109 (FEProcOut) that is output from the ISP front-end pixel processing logic 130 and forwarded to the ISP pipe processing logic 82.

[0278] Before continuing, it should also be noted, that given sufficient processing time and the similarity between many of the processing requirements of the various operations described herein, it is possible to reconfigure the functional blocks shown herein to perform image processing in a sequential manner, rather than a pipe-lined nature. As will be understood, this may further reduce the overall hardware implementation costs, but may also increase bandwidth to external memory (e.g., to cache/store intermediate results/data).

[0279] Having described the ISP front-end logic 80 in detail above, the present discussion will now shift focus to the ISP pipe processing logic 82. Generally, the function of the ISP pipe logic 82 is to receive raw image data, which may be provided from the ISP front-end logic 80 or retrieved from memory 108, and to perform additional image processing operations, i.e., prior to outputting the image data to the display device 28.

[0280] A block diagram showing an embodiment of the ISP pipe logic 82 is depicted in FIG. 49. As illustrated, the ISP pipe logic 82 may include raw processing logic 562, RGB processing logic 564, and YCbCr processing logic 566. The raw processing logic 562 may perform various image processing operations, such as defective pixel detection and correction, lens shading correction, demosaicing, as well as applying gains for auto-white balance and/or setting a black level, as will be discussed further below. As shown in the present embodiment, the input signal 570 to the raw processing logic 562 may be the raw pixel output 109 (signal FEProcOut) from the ISP front-end logic 80 or the raw pixel data 112 from the memory 108, depending on the present configuration of the selection logic 568.

[0281] As a result of demosaicing operations performed within the raw processing logic 562, the image signal output 572 may be in the RGB domain, and may be subsequently forwarded to the RGB processing logic 564. For instance, as shown in FIG. 49, the RGB processing logic 564 receives the signal 578, which may be the output signal 572 or an RGB image signal 574 from the memory 108, depending on the present configuration of the selection logic 576. The RGB processing logic 564 may provide for various RGB color adjustment operations, including color correction (e.g., using a color correction matrix), the application of color gains for auto-white balancing, as well as global tone mapping, as will be discussed further below. The RGB processing logic 566 may also provide for the color space conversion of RGB image data to the YCbCr (luma/chroma) color space. Thus, the image signal output 580 may be in the YCbCr domain, and may be subsequently forwarded to the YCbCr processing logic 566.

[0282] For instance, as shown in FIG. 49, the YCbCr processing logic 566 receives the signal 586, which may be the output signal 580 from the RGB processing logic 564 or a YCbCr signal 582 from the memory 108, depending on the present configuration of the selection logic 584. As will be discussed in further detail below, the YCbCr processing logic 566 may provide for image processing operations in the YCbCr color space, including scaling, chroma suppression, luma sharpening, brightness, contrast, and color (BCC) adjustments, YCbCr gamma mapping, chroma decimation, and so forth. The image signal output 588 of the YCbCr processing logic 566 may be sent to the memory 108, or may be output from the ISP pipe processing logic 82 as the image signal 114 (FIG. 7). The image signal 114 may be sent to the display device 28 (either directly or via memory 108) for viewing by the user, or may be further processed using a compression engine (e.g., encoder 118), a CPU/GPU, a graphics engine, or the like.

[0283] In accordance with embodiments of the present techniques, the ISP pipe logic 82 may support the processing of raw pixel data in 8-bit, 10-bit, 12-bit, or 14-bit formats. For instance, in one embodiment, 8-bit, 10-bit, or 12-bit input data may be converted to 14-bit at the input of the raw processing logic 562, and raw processing and RGB processing operations may be performed with 14-bit precision. In the latter embodiment, the 14-bit image data may be down-sampled to 10 bits prior to the conversion of the RGB data to the YCbCr color space, and the YCbCr processing (logic 566) may be performed with 10-bit precision.

[0284] In order to provide a comprehensive description of the various functions provided by the ISP pipe processing logic 82, each of the raw processing logic 562, RGB processing logic 564, and YCbCr processing logic 566, as well as internal logic for performing various image processing operations that may be implemented in each respective unit of logic 562, 564, and 566, will be discussed sequentially below, beginning with the raw processing logic 562. For instance, referring now to FIG. 50, a block diagram showing a more detailed view of an embodiment of the raw processing logic 562 is illustrated, in accordance with an embodiment of the present technique. As shown, the raw processing logic 562 includes the gain, offset, and clamping (GOC) logic 590, defective pixel detection/correction (DPDC) logic 592, the noise reduction logic 594, lens shading correction logic 596, GOC logic 598, and demosaicing logic 600. Further, while the examples discussed below assume the use of a Bayer color filter array with the image sensor(s) 90, it should be understood that other embodiments of the present technique may utilize different types of color filters as well.

[0285] The input signal 570, which may be a raw image signal, is first received by the gain, offset, and clamping (GOC) logic 590. The GOC logic 590 may provide similar functions and may be implemented in a similar manner with respect to the BLC logic 462 of the statistics processing unit 120 of the ISP front-end logic 80, as discussed above in FIG. 37. For instance, the GOC logic 590 may provide digital gain, offsets and clamping (clipping) independently for each color component R, B, Gr, and Gb of a Bayer image sensor. Particularly, the GOC logic 590 may perform auto-white balance or set the black level of the raw image data. Further, in some embodiments, the GOC logic 590 may also be used correct or compensate for an offset between the Gr and Gb color components.

[0286] In operation, the input value for the current pixel is first offset by a signed value and multiplied by a gain. This operation may be performed using the formula shown in Equation 11 above, wherein X represents the input pixel value for a given color component R, B, Gr, or Gb, O[c] represents a signed 16-bit offset for the current color component c, and G[c] represents a gain value for the color component c. The values for G[c] may be previously determined during statistics processing (e.g., in the ISP front-end block 80). In one embodiment, the gain G[c] may be a 16-bit unsigned number with 2 integer bits and 14 fraction bits (e.g., 2.14 floating point representation), and the gain G[c] may be applied with rounding. By way of example only, the gain G[c] may have a range of between 0 to 4X.

[0287] The computed pixel value Y (which includes the gain G[c] and offset O[c]) from Equation 11 is then be clipped to a minimum and a maximum range in accordance with Equation 12. As discussed above, the variables min[c] and max[c] may represent signed 16-bit “clipping values” for the minimum and maximum output values, respectively. In one embodiment, the GOC logic 590 may also be configured to maintain a count of the number of pixels that were clipped above and below maximum and minimum ranges, respectively, for each color component.

[0288] Subsequently, the output of the GOC logic 590 is forwarded to the defective pixel detection and correction logic 592. As discussed above with reference to FIG. 37 (DPDC logic 460), defective pixels may attributable to a number of factors, and may include “hot” (or leaky) pixels, “stuck” pixels, and “dead pixels, wherein hot pixels exhibit a higher than normal charge leakage relative to non-defective pixels, and thus may appear brighter than non-defective pixel, and wherein a stuck pixel appears as always being on (e.g., fully charged) and thus appears brighter, whereas a dead pixel appears as always being off. As such, it may be desirable to have a pixel detection scheme that is robust enough to identify and address different types of failure scenarios. Particularly, when compared to the front-end DPDC logic 460, which may provide only dynamic defect detection/correction, the pipe DPDC logic 592 may provide for fixed or static defect detection/correction, dynamic defect detection/correction, as well as speckle removal.

[0289] In accordance with embodiments of the presently disclosed techniques, defective pixel correction/detection performed by the DPDC logic 592 may occur independently for each color component (e.g., R, B, Gr, and Gb), and may include various operations for detecting defective pixels, as well as for correcting the detected defective pixels. For instance, in one embodiment, the defective pixel detection operations may provide for the detection of static defects, dynamics defects, as well as the detection of speckle, which may refer to the electrical interferences or noise (e.g., photon noise) that may be present in the imaging sensor. By analogy, speckle may appear on an image as seemingly random noise artifacts, similar to the manner in which static may appear on a display, such as a television display. Further, as noted above, dynamic defection correction is regarded as being dynamic in the sense that the characterization of a pixel as being defective at a given time may depend on the image data in the neighboring pixels. For example, a stuck pixel that is always on maximum brightness may not be regarded as a defective pixel if the location of the stuck pixel is in an area of the current image that is dominate by bright white colors. Conversely, if the stuck pixel is in a region of the current image that is dominated by black or darker colors, then the stuck pixel may be identified as a defective pixel during processing by the DPDC logic 592 and corrected accordingly.

[0290] With regard to static defect detection, the location of each pixel is compared to a static defect table, which may store data corresponding to the location of pixels that are known to be defective. For instance, in one embodiment, the DPDC logic 592 may monitor the detection of defective pixels (e.g., using a counter mechanism or register) and, if a particular pixel is observed as repeatedly failing, the location of that pixel is stored into the static defect table. Thus, during static defect detection, if it is determined that the location of the current pixel is in the static defect table, then the current pixel is identified as being a defective pixel, and a replacement value is determined and temporarily stored. In one embodiment, the replacement value may be the value of the previous pixel (based on scan order) of the same color component. The replacement value may be used to correct the static defect during dynamic/speckle defect detection and correction, as will be discussed below. Additionally, if the previous pixel is outside of the raw frame 278 (FIG. 19), then its value is not used, and the static defect may be corrected during the dynamic defect correction process. Further, due to memory considerations, the static defect table may store a finite number of location entries. For instance, in one embodiment, the static defect table may be implemented as a FIFO queue configured to store a total of 16 locations for every two lines of image data. The locations in defined in the static defect table will, nonetheless, be corrected using a previous pixel replacement value (rather than via the dynamic defect detection process discussed below). As mentioned above, embodiments of the present technique may also provide for updating the static defect table intermittently over time.

[0291] Embodiments may provide for the static defect table to be implemented in on-chip memory or off-chip memory. As will be appreciated, using an on-chip implementation may increase overall chip area/size, while using an off-chip implementation may reduce chip area/size, but increase memory bandwidth requirements. Thus, it should be understood that the static defect table may be implemented either on-chip or off-chip depending on specific implementation requirements, i.e., the total number of pixels that are to be stored within the static defect table.

[0292] The dynamic defect and speckle detection processes may be time-shifted with respect to the static defect detection process discussed above. For instance, in one embodiment, the dynamic defect and speckle detection process may begin after the static defect detection process has analyzed two scan lines (e.g., rows) of pixels. As can be appreciated, this allows for the identification of static defects and their respective replacement values to be determined before dynamic/speckle detection occurs. For example, during the dynamic/speckle detection process, if the current pixel was previously marked as being a static defect, rather than applying dynamic/speckle detection operations, the static defect is simply corrected using the previously assessed replacement value.

[0293] With regard to dynamic defect and speckle detection, these processes may occur sequentially or in parallel. The dynamic defect and speckle detection and correction that is performed by the DPDC logic 592 may rely on adaptive edge detection using pixel-to-pixel direction gradients. In one embodiment, the DPDC logic 592 may select the eight immediate neighbors of the current pixel having the same color component that are within the raw frame 278 (FIG. 19) are used. In other words, the current pixels and its eight immediate neighbors P0, P1, P2, P3, P4, P5, P6, and P7 may form a 3×3 area, as shown below in FIG. 51.

[0294] It should be noted, however, that depending on the location of the current pixel P, pixels outside the raw frame 278 are not considered when calculating pixel-to-pixel gradients. For example, with regard to the “top-left” case 602 shown in FIG. 51, the current pixel P is at the top-left corner of the raw frame 278 and, thus, the neighboring pixels P0, P1, P2, P3, and P5 outside of the raw frame 278 are not considered, leaving only the pixels P4, P6, and P7 (N=3). In the “top” case 604, the current pixel P is at the top-most edge of the raw frame 278 and, thus, the neighboring pixels P0, P1, and P2 outside of the raw frame 278 are not considered, leaving only the pixels P3, P4, P5, P6, and P7 (N=5). Next, in the “top-right” case 606, the current pixel P is at the top-right corner of the raw frame 278 and, thus, the neighboring pixels P0, P1, P2, P4, and P7 outside of the raw frame 278 are not considered, leaving only the pixels P3, P5, and P6 (N=3). In the “left” case 608, the current pixel P is at the left-most edge of the raw frame 278 and, thus, the neighboring pixels P0, P3, and P5 outside of the raw frame 278 are not considered, leaving only the pixels P1, P2, P4, P6, and P7 (N=5).

[0295] In the “center” case 610, all pixels P0-P7 lie within the raw frame 278 and are thus used in determining the pixel-to-pixel gradients (N=8). In the “right” case 612, the current pixel P is at the right-most edge of the raw frame 278 and, thus, the neighboring pixels P2, P4, and P7 outside of the raw frame 278 are not considered, leaving only the pixels P0, P1, P3, P5, and P6 (N=5). Additionally, in the “bottom-left” case 614, the current pixel P is at the bottom-left corner of the raw frame 278 and, thus, the neighboring pixels P0, P3, P5, P6, and P7 outside of the raw frame 278 are not considered, leaving only the pixels P1, P2, and P4 (N=3). In the “bottom” case 616, the current pixel P is at the bottom-most edge of the raw frame 278 and, thus, the neighboring pixels P5, P6, and P7 outside of the raw frame 278 are not considered, leaving only the pixels P0, P1, P2, P3, and P4 (N=5). Finally, in the “bottom-right” case 618, the current pixel P is at the bottom-right corner of the raw frame 278 and, thus, the neighboring pixels P2, P4, P5, P6, and P7 outside of the raw frame 278 are not considered, leaving only the pixels P0, P1, and P3 (N=3).

[0296] Thus, depending upon the position of the current pixel P, the number of pixels used in determining the pixel-to-pixel gradients may be 3, 5, or 8. In the illustrated embodiment, for each neighboring pixel (k=0 to 7) within the picture boundary (e.g., raw frame 278), the pixel-to-pixel gradients may be calculated as follows:
G_k=abs(P−P_k), for 0≦k≦7 (only for k within the raw frame)  (19)
Additionally, an average gradient, G_av, may be calculated as the difference between the current pixel and the average, P_av, of its surrounding pixels, as shown by the equations below:

[0297] $\begin{array}{cc}{P}_{\mathrm{av}}=\frac{\left(\sum _k^N{P}_k\right)}{N},\mathrm{wherein}N=3,5,\mathrm{or}8\left(\mathrm{depending}\mathrm{on}\mathrm{pixel}\mathrm{position}\right)& \left(20a\right)\\ G_{\mathrm{av}}=\mathrm{abs}\left(P-P_{\mathrm{av}}\right)& \left(20b\right)\end{array}$
The pixel-to-pixel gradient values (Equation 19) may be used in determining a dynamic defect case, and the average of the neighboring pixels (Equations 20a and 20b) may be used in identifying speckle cases, as discussed further below.

[0298] In one embodiment, dynamic defect detection may be performed by the DPDC logic 592 as follows. First, it is assumed that a pixel is defective if a certain number of the gradients G_k are at or below a particular threshold, denoted by the variable dynTh (dynamic defect threshold). Thus, for each pixel, a count (C) of the number of gradients for neighboring pixels inside the picture boundaries that are at or below the threshold dynTh is accumulated. The threshold dynTh may be a combination of a fixed threshold component and a dynamic threshold component that may depend on the “activity” present the surrounding pixels. For instance, in one embodiment, the dynamic threshold component for dynTh may be determined by calculating a high frequency component value P_hf based upon summing the absolute difference between the average pixel values P_av (Equation 20a) and each neighboring pixel, as illustrated below:

[0299] $\begin{array}{cc}{P}_{\mathrm{hf}}=\frac{8}{N}\sum _k^N\mathrm{abs}\left(P_{\mathrm{av}}-P_k\right)\mathrm{wherein}N=3,5,\mathrm{or}8& \left(20c\right)\end{array}$
In instances where the pixel is located at an image corner (N=3) or at an image edge (N=5), the P_hf may be multiplied by the 8/3 or 8/5, respectively. As can be appreciated, this ensures that the high frequency component P_hf is normalized based on eight neighboring pixels (N=8).

[0300] Once P_hf is determined, the dynamic defect detection threshold dynTh may be computed as shown below:
dynTh=dynTh₁+(dynTh₂×P_hf),  (21)
wherein dynTh₁ represents the fixed threshold component, and wherein dynTh₂ represents the dynamic threshold component, and is a multiplier for P_hf in Equation 21. A different fixed threshold component dynTh₁ may be provided for each color component, but for each pixel of the same color, dynTh₁ is the same. By way of example only, dynTh₁ may be set so that it is at least above the variance of noise in the image.

[0301] The dynamic threshold component dynTh₂ may be determined based on some characteristic of the image. For instance, in one embodiment, dynTh₂ may be determined using stored empirical data regarding exposure and/or sensor integration time. The empirical data may be determined during calibration of the image sensor (e.g., 90), and may associate dynamic threshold component values that may be selected for dynTh₂ with each of a number of data points. Thus, based upon the current exposure and/or sensor integration time value, which may be determined during statistics processing in the ISP front-end logic 80, dynTh₂ may be determined by selecting the dynamic threshold component value from the stored empirical data that corresponds to the current exposure and/or sensor integration time value. Additionally, if the current exposure and/or sensor integration time value does not correspond directly to one of the empirical data points, then dynTh₂ may be determined by interpolating the dynamic threshold component values associated with the data points between which the current exposure and/or sensor integration time value falls. Further, like the fixed threshold component dynTh₁, the dynamic threshold component dynTh₂ may have different values for each color component. Thus, composite threshold value dynTh may vary for each color component (e.g., R, B, Gr, Gb).

[0302] As mentioned above, for each pixel, a count C of the number of gradients for neighboring pixels inside the picture boundaries that are at or below the threshold dynTh is determined. For instance, for each neighboring pixel within the raw frame 278, the accumulated count C of the gradients G_k that are at or below the threshold dynTh may be computed as follows:

[0303] $\begin{array}{cc}C=\sum _k^N\left(G_k\le \mathrm{dynTh}\right),& \left(22\right)\end{array}$

[0304] for 0≦k≦7 (only for k within the raw frame)

[0305] Next, if the accumulated count C is determined to be less than or equal to a maximum count, denoted by the variable dynMaxC, then the pixel may be considered as a dynamic defect. In one embodiment, different values for dynMaxC may be provided for N=3 (corner), N=5 (edge), and N=8 conditions. This logic is expressed below:
if(C≦dynMaxC), then the current pixel P is defective.  (23)

[0306] As mentioned above, the location of defective pixels may be stored into the static defect table. In some embodiments, the minimum gradient value (min(G_k)) calculated during dynamic defect detection for the current pixel may be stored and may be used to sort the defective pixels, such that a greater minimum gradient value indicates a greater “severity” of a defect and should be corrected during pixel correction before less severe defects are corrected. In one embodiment, a pixel may need to be processed over multiple imaging frames before being stored into the static defect table, such as by filtering the locations of defective pixels over time. In the latter embodiment, the location of the defective pixel may be stored into the static defect table only if the defect appears in a particular number of consecutive images at the same location. Further, in some embodiments, the static defect table may be configured to sort the stored defective pixel locations based upon the minimum gradient values. For instance, the highest minimum gradient value may indicate a defect of greater “severity.” By ordering the locations in this manner, the priority of static defect correction may be set, such that the most severe or important defects are corrected first. Additionally, the static defect table may be updated over time to include newly detected static defects, and ordering them accordingly based on their respective minimum gradient values.

[0307] Speckle detection, which may occur in parallel with the dynamic defect detection process described above, may be performed by determining if the value G_av (Equation 20b) is above a speckle detection threshold spkTh. Like the dynamic defect threshold dynTh, the speckle threshold spkTh may also include fixed and dynamic components, referred to by spkTh₁ and spkTh₂, respectively. In general, the fixed and dynamic components spkTh₁ and spkTh₂ may be set more “aggressively” compared to the dynTh₁ and dynTh₂ values, in order to avoid falsely detecting speckle in areas of the image that may be more heavily textured and others, such as text, foliage, certain fabric patterns, etc. Accordingly, in one embodiment, the dynamic speckle threshold component spkTh₂ may be increased for high-texture areas of the image, and decreased for “flatter” or more uniform areas. The speckle detection threshold spkTh may be computed as shown below:
spkTh=spkTh₁+(spkTh₂×P_hf),  (24)
wherein spkTh₁ represents the fixed threshold component, and wherein spkTh₂ represents the dynamic threshold component. The detection of speckle may then be determined in accordance with the following expression:
if(G_av>spkTh), then the current pixel P is speckled.  (25)

[0308] Once defective pixels have been identified, the DPDC logic 592 may apply pixel correction operations depending on the type of defect detected. For instance, if the defective pixel was identified as a static defect, the pixel is replaced with the stored replacement value, as discussed above (e.g., the value of the previous pixel of the same color component). If the pixel was identified as either a dynamic defect or as speckle, then pixel correction may be performed as follows. First, gradients are computed as the sum of the absolute difference between the center pixel and a first and second neighbor pixels (e.g., computation of G_k of Equation 19) for four directions, a horizontal (h) direction, a vertical (v) direction, a diagonal-positive direction (dp), and a diagonal-negative direction (dn), as shown below:
G_h=G₃+G₄  (26)
G_v=G₁+G₆  (27)
G_dp=G₂+G₅  (28)
G_dn=G_O+G₇  (29)

[0309] Next, the corrective pixel value P_c may be determined via linear interpolation of the two neighboring pixels associated with the directional gradient G_h, G_v, G_dp, and G_dn that has the smallest value. For instance, in one embodiment, the logic statement below may express the calculation of P_C:

[0310] $\begin{array}{cc}\mathrm{if}\left(\min ==G_h\right)P_C=\frac{P_3+P_4}{2};\mathrm{else}\mathrm{if}\left(\min ==G_v\right)P_C=\frac{P_1+P_6}{2};\mathrm{else}\mathrm{if}\left(\min ==G_{\mathrm{dp}}\right)P_C=\frac{P_2+P_5}{2};\mathrm{else}\mathrm{if}\left(\min ==G_{\mathrm{dn}}\right)P_C=\frac{P_0+P_7}{2};& \left(30\right)\end{array}$
The pixel correction techniques implemented by the DPDC logic 592 may also provide for exceptions at boundary conditions. For instance, if one of the two neighboring pixels associated with the selected interpolation direction is outside of the raw frame, then the value of the neighbor pixel that is within the raw frame is substituted instead. Thus, using this technique, the corrective pixel value will be equivalent to the value of the neighbor pixel within the raw frame.

[0311] It should be noted that the defective pixel detection/correction techniques applied by the DPDC logic 592 during the ISP pipe processing is more robust compared to the DPDC logic 460 in the ISP front-end logic 80. As discussed in the embodiment above, the DPDC logic 460 performs only dynamic defect detection and correction using neighboring pixels in only the horizontal direction, whereas the DPDC logic 592 provides for the detection and correction of static defects, dynamic defects, as well as speckle, using neighboring pixels in both horizontal and vertical directions.

[0312] As will be appreciated, the storage of the location of the defective pixels using a static defect table may provide for temporal filtering of defective pixels with lower memory requirements. For instance, compared to many conventional techniques which store entire images and apply temporal filtering to identify static defects over time, embodiments of the present technique only store the locations of defective pixels, which may typically be done using only a fraction of the memory required to store an entire image frame. Further, as discussed above, the storing of a minimum gradient value (min(G_k)), allows for an efficient use of the static defect table prioritizing the order of the locations at which defective pixels are corrected (e.g., beginning with those that will be most visible).

[0313] Additionally, the use of thresholds that include a dynamic component (e.g., dynTh₂ and spkTh₂) may help to reduce false defect detections, a problem often encountered in conventional image processing systems when processing high texture areas of an image (e.g., text, foliage, certain fabric patterns, etc.). Further, the use of directional gradients (e.g., h, v, dp, dn) for pixel correction may reduce the appearance of visual artifacts if a false defect detection occurs. For instance, filtering in the minimum gradient direction may result in a correction that still yields acceptable results under most cases, even in cases of false detection. Additionally, the inclusion of the current pixel P in the gradient calculation may improve the accuracy of the gradient detection, particularly in the case of hot pixels.

[0314] The above-discussed defective pixel detection and correction techniques implemented by the DPDC logic 592 may be summarized by a series of flow charts provided in FIGS. 52-54. For instance, referring first to FIG. 52, a process 620 for detecting static defects is illustrated. Beginning initially at step 622, an input pixel P is received at a first time, T₀. Next, at step 624, the location of the pixel P is compared to the values stored in a static defect table. Decision logic 626 determines whether the location of the pixel P is found in the static defect table. If the location of P is in the static defect table, then the process 620 continues to step 628, wherein the pixel P is marked as a static defect and a replacement value is determined. As discussed above, the replacement value may be determined based upon the value of the previous pixel (in scan order) of the same color component. The process 620 then continues to step 630, at which the process 620 proceeds to the dynamic and speckle detection process 640, illustrated in FIG. 53. Additionally, if at decision logic 626, the location of the pixel P is determined not to be in the static defect table, then the process 620 proceeds to step 630 without performing step 628.

[0315] Continuing to FIG. 53, the input pixel P is received at time T1, as shown by step 642, for processing to determine whether a dynamic defect or speckle is present. Time T1 may represent a time-shift with respect to the static defect detection process 620 of FIG. 52. As discussed above, the dynamic defect and speckle detection process may begin after the static defect detection process has analyzed two scan lines (e.g., rows) of pixels, thus allowing time for the identification of static defects and their respective replacement values to be determined before dynamic/speckle detection occurs.

[0316] The decision logic 644 determines if the input pixel P was previously marked as a static defect (e.g., by step 628 of process 620). If P is marked as a static defect, then the process 640 may continue to the pixel correction process shown in FIG. 54 and may bypass the rest of the steps shown in FIG. 53. If the decision logic 644 determines that the input pixel P is not a static defect, then the process continues to step 646, and neighboring pixels are identified that may be used in the dynamic defect and speckle process. For instance, in accordance with the embodiment discussed above and illustrated in FIG. 51, the neighboring pixels may include the immediate 8 neighbors of the pixel P (e.g., P0-P7), thus forming a 3×3 pixel area. Next, at step 648, pixel-to-pixel gradients are calculated with respect to each neighboring pixel within the raw frame 278, as described in Equation 19 above. Additionally, an average gradient (G_av) may be calculated as the difference between the current pixel and the average of its surrounding pixels, as shown in Equations 20a and 20b.

[0317] The process 640 then branches to step 650 for dynamic defect detection and to decision logic 658 for speckle detection. As noted above, dynamic defect detection and speckle detection may, in some embodiments, occur in parallel. At step 650, a count C of the number of gradients that are less than or equal to the threshold dynTh is determined. As described above, the threshold dynTh may include fixed and dynamic components and, in one embodiment, may be determined in accordance with Equation 21 above. If C is less than or equal to a maximum count, dynMaxC, then the process 640 continues to step 656, and the current pixel is marked as being a dynamic defect. Thereafter, the process 640 may continue to the pixel correction process shown in FIG. 54, which will be discussed below.

[0318] Returning back the branch after step 648, for speckle detection, the decision logic 658 determines whether the average gradient G_av is greater than a speckle detection threshold spkTh, which may also include a fixed and dynamic component. If G_av is greater than the threshold spkTh, then the pixel P is marked as containing speckle at step 660 and, thereafter, the process 640 continues to FIG. 54 for the correction of the speckled pixel. Further, if the output of both of the decision logic blocks 652 and 658 are “NO,” then this indicates that the pixel P does not contain dynamic defects, speckle, or even static defects (decision logic 644). Thus, when the outputs of decision logic 652 and 658 are both “NO,” the process 640 may conclude at step 654, whereby the pixel P is passed unchanged, as no defects (e.g., static, dynamic, or speckle) were detected.

[0319] Continuing to FIG. 54, a pixel correction process 670 in accordance with the techniques described above is provided. At step 672, the input pixel P is received from process 640 of FIG. 53. It should be noted that the pixel P may be received by process 670 from step 644 (static defect) or from steps 656 (dynamic defect) and 660 (speckle defect). The decision logic 674 then determines whether the pixel P is marked as a static defect. If the pixel P is a static defect, then the process 670 continues and ends at step 676, whereby the static defect is corrected using the replacement value determined at step 628 (FIG. 52).

[0320] If the pixel P is not identified as a static defect, then the process 670 continues from decision logic 674 to step 678, and directional gradients are calculated. For instance, as discussed above with reference to Equations 26-29, the gradients may be computed as the sum of the absolute difference between the center pixel and first and second neighboring pixels for four directions (h, v, dp, and dn). Next, at step 680, the directional gradient having the smallest value is identified and, thereafter, decision logic 682 assesses whether one of the two neighboring pixels associated with the minimum gradient is located outside of the image frame (e.g., raw frame 278). If both neighboring pixels are within the image frame, then the process 670 continues to step 684, and a pixel correction value (P_C) is determined by applying linear interpolation to the values of the two neighboring pixels, as illustrated by Equation 30. Thereafter, the input pixel P may be corrected using the interpolated pixel correction value P_C, as shown at step 690.

[0321] Returning to the decision logic 682, if it is determined that one of the two neighboring pixels are located outside of the image frame (e.g., raw frame 165), then instead of using the value of the outside pixel (Pout), the DPDC logic 592 may substitute the value of Pout with the value of the other neighboring pixel that is inside the image frame (Pin), as shown at step 686. Thereafter, at step 688, the pixel correction value P_c is determined by interpolating the values of Pin and the substituted value of Pout. In other words, in this case, P_C may be equivalent to the value of Pin. Concluding at step 690, the pixel P is corrected using the value P_C. Before continuing, it should be understood that the particular defective pixel detection and correction processes discussed herein with reference to the DPDC logic 592 are intended to reflect only one possible embodiment of the present technique. Indeed, depending on design and/or cost constraints, a number of variations are possible, and features may be added or removed such that the overall complexity and robustness of the defect detection/correction logic is between the simpler detection/correction logic 460 implemented in the ISP front-end block 80 and the defect detection/correction logic discussed here with reference to the DPDC logic 592.

[0322] Referring back to FIG. 50, the corrected pixel data is output from the DPDC logic 592 and then received by the noise reduction logic 594 for further processing. In one embodiment, the noise reduction logic 594 may be configured to implements two-dimensional edge-adaptive low pass filtering to reduce noise in the image data while maintaining details and textures. The edge-adaptive thresholds may be set (e.g., by the control logic 84) based upon the present lighting levels, such that filtering may be strengthened under low light conditions. Further, as briefly mentioned above with regard to the determination of the dynTh and spkTh values, noise variance may be determined ahead of time for a given sensor so that the noise reduction thresholds can be set just above noise variance, such that during the noise reduction processing, noise is reduced without significantly affecting textures and details of the scene (e.g., avoid/reduce false detections). Assuming a Bayer color filter implementation, the noise reduction logic 594 may process each color component Gr, R, B, and Gb independently using a separable 7-tap horizontal filter and a 5-tap vertical filter. In one embodiment, the noise reduction process may be carried out by correcting for non-uniformity on the green color components (Gb and Gr), and then performing horizontal filtering and vertical filtering.

[0323] Green non-uniformity (GNU) is generally characterized by a slight brightness difference between the Gr and Gb pixels given a uniformly illuminated flat surface. Without correcting or compensating for this non-uniformity, certain artifacts, such as a “maze” artifact, may appear in the full color image after demosaicing. During the green non-uniformity process may include determining, for each green pixel in the raw Bayer image data, if the absolute difference between a current green pixel (G1) and the green pixel to the right and below (G2) the current pixel is less than a GNU correction threshold (gnuTh). FIG. 55 illustrates the location of the G1 and G2 pixels in a 2×2 area of the Bayer pattern. As shown, the color of the pixels bordering G1 may be depending upon whether the current green pixel is a Gb or Gr pixel. For instance, if G1 is Gr, then G2 is Gb, the pixel to the right of G1 is R (red), and the pixel below G1 is B (blue). Alternatively, if G1 is Gb, then G2 is Gr, and the pixel to the right of G1 is B, whereas the pixel below G1 is R. If the absolute difference between G1 and G2 is less than the GNU correction threshold value, then current green pixel G1 is replaced by the average of G1 and G2, as shown by the logic below:

[0324] $\begin{array}{cc}\mathrm{if}\left(\mathrm{abs}\left(G1-G2\right)\le \mathrm{gnuTh}\right);G1=\frac{G1+G2}{2}& \left(31\right)\end{array}$
As can be appreciated, the application of green non-uniformity correction in this manner may help to prevent the G1 and G2 pixels from being averaged across edges, thus improving and/or preserving sharpness.

[0325] Horizontal filtering is applied subsequent to green non-uniformity correction and may, in one embodiment, provide a 7-tap horizontal filter. Gradients across the edge of each filter tap are computed, and if it is above a horizontal edge threshold (horzTh), the filter tap is folded to the center pixel, as will be illustrated below. The horizontal filter may process the image data independently for each color component (R, B, Gr, Gb) and may use unfiltered values as inputs values.

[0326] By way of example, FIG. 56 shows a graphical depiction of a set of horizontal pixels P0 to P6, with a center tap positioned at P3. Based upon the pixels shown in FIG. 56, edge gradients for each filter tap may be calculated as follows:
Eh0=abs(P0−P1)  (32)
Eh1=abs(P1−P2)  (33)
Eh2=abs(P2−P3)  (34)
Eh3=abs(P3−P4)  (35)
Eh4=abs(P4−P5)  (36)
Eh5=abs(P5−P6)  (37)
The edge gradients Eh0-Eh5 may then be utilized by the horizontal filter component to determine a horizontal filtering output, P_horz, using the formula shown in Equation 38 below:

[0327] $\begin{array}{cc}{P}_{\mathrm{horz}}=C0\times \left[\left(\mathrm{Eh}2>\mathrm{horzTh}\left[c\right]\right)?P3:\left(\mathrm{Eh}1>\mathrm{horzTh}\left[c\right]\right)?P2:\left(\mathrm{Eh}0>\mathrm{horzTh}\left[c\right]\right)?P1:P0\right]+C1\times \left[\left(\mathrm{Eh}2>\mathrm{horzTh}\left[c\right]\right)?P3:\left(\mathrm{Eh}1>\mathrm{horzTh}\left[c\right]\right)?P2:P1\right]+C2\times \left[\left(\mathrm{Eh}2>\mathrm{horzTh}\left[c\right]\right)?P3:P2\right]+C3\times P3+C4\times \left[\left(\mathrm{Eh}3>\mathrm{horzTh}\left[c\right]\right)?P3:P4\right]+C5\times \left[\left(\mathrm{Eh}3>\mathrm{horzTh}\left[c\right]\right)?P3:\left(\mathrm{Eh}4>\mathrm{horzTh}\left[c\right]\right)?P4:P5\right]+C6\times \left[\left(\mathrm{Eh}3>\mathrm{horzTh}\left[c\right]\right)?P3:\left(\mathrm{Eh}4>\mathrm{horzTh}\left[c\right]\right)?P4:\left(\mathrm{Eh}5>\mathrm{horzTh}\left[c\right]\right)?P5:P6\right],& \left(38\right)\end{array}$
wherein horzTh[c] is the horizontal edge threshold for each color component c (e.g., R, B, Gr, and Gb), and wherein C0-C6 are the filter tap coefficients corresponding to pixels P0-P6, respectively. The horizontal filter output P_horz may be applied at the center pixel P3 location. In one embodiment, the filter tap coefficients C0-C6 may be 16-bit two's complement values with 3 integer bits and 13 fractional bits (3.13 in floating point). Further, it should be noted that the filter tap coefficients C0-C6 need not necessarily be symmetrical with respect to the center pixel P3.

[0328] Vertical filtering is also applied by the noise reduction logic 594 subsequent to green non-uniformity correction and horizontal filtering processes. In one embodiment, the vertical filter operation may provide a 5-tap filter, as shown in FIG. 57, with the center tap of the vertical filter located at P2. The vertical filtering process may occur in a similar manner as the horizontal filtering process described above. For instance, gradients across the edge of each filter tap are computed, and if it is above a vertical edge threshold (vertTh), the filter tap is folded to the center pixel P2. The vertical filter may process the image data independently for each color component (R, B, Gr, Gb) and may use unfiltered values as inputs values.

[0329] Based upon the pixels shown in FIG. 57, vertical edge gradients for each filter tap may be calculated as follows:
Ev0=abs(P0−P1)  (39)
Ev1=abs(P1−P2)  (40)
Ev2=abs(P2−P3)  (41)
Ev3=abs(P3−P4)  (42)
The edge gradients Ev0-Ev5 may then be utilized by the vertical filter to determine a vertical filtering output, P_vert, using the formula shown in Equation 43 below:

[0330] $\begin{array}{cc}{P}_{\mathrm{vert}}=C0\times \left[\left(\mathrm{Ev}1>\mathrm{vertTh}\left[c\right]\right)?P2:\left(\mathrm{Ev}0>\mathrm{vertTh}\left[c\right]\right)?P1:P0\right]+C1\times \left[\left(\mathrm{Ev}1>\mathrm{vertTh}\left[c\right]\right)?P2:P1\right]+C2\times P2+C3\times \left[\left(\mathrm{Ev}2>\mathrm{vertTh}\left[c\right]\right)?P2:P3\right]+C4\times \left[\left(\mathrm{Ev}2>\mathrm{vertTh}\left[c\right]\right)?P2:\left(\mathrm{Eh}3>\mathrm{vertTh}\left[c\right]\right)?P3:P4\right],& \left(43\right)\end{array}$
wherein vertTh[c] is the vertical edge threshold for each color component c (e.g., R, B, Gr, and Gb), and wherein C0-C4 are the filter tap coefficients corresponding to the pixels P0-P4 of FIG. 57, respectively. The vertical filter output P_vert may be applied at the center pixel P2 location. In one embodiment, the filter tap coefficients C0-C4 may be 16-bit two's complement values with 3 integer bits and 13 fractional bits (3.13 in floating point). Further, it should be noted that the filter tap coefficients C0-C4 need not necessarily be symmetrical with respect to the center pixel P2.

[0331] Additionally, with regard to boundary conditions, when neighboring pixels are outside of the raw frame 278 (FIG. 19), the values of the out-of-bound pixels are replicated with the value of same color pixel at the edge of the raw frame. This convention may be implemented for both horizontal and vertical filtering operations. By way of example, referring again to FIG. 56, in the case of horizontal filtering, if the pixel P2 is an edge pixel at the left-most edge of the raw frame, and the pixels P0 and P1 are outside of the raw frame, then the values of the pixels P0 and P1 are substituted with the value of the pixel P2 for horizontal filtering.

[0332] Referring again back to the block diagram of the raw processing logic 562 shown in FIG. 50, the output of the noise reduction logic 594 is subsequently sent to the lens shading correction (LSC) logic 596 for processing. As discussed above, lens shading correction techniques may include applying an appropriate gain on a per-pixel basis to compensate for drop-offs in light intensity, which may be the result of the geometric optics of the lens, imperfections in manufacturing, misalignment of the microlens array and the color array filter, and so forth. Further, the infrared (IR) filter in some lenses may cause the drop-off to be illuminant-dependent and, thus, lens shading gains may be adapted depending upon the light source detected.

[0333] In the depicted embodiment, the LSC logic 596 of the ISP pipe 82 may be implemented in a similar manner, and thus provide generally the same functions, as the LSC logic 464 of the ISP front-end block 80, as discussed above with reference to FIGS. 40-48. Accordingly, in order to avoid redundancy, it should be understood that the LSC logic 596 of the presently illustrated embodiment is configured to operate in generally the same manner as the LSC logic 460 and, as such, the description of the lens shading correction techniques provided above will not be repeated here. However, to generally summarize, it should be understood that the LSC logic 596 may process each color component of the raw pixel data stream independently to determine a gain to apply to the current pixel. In accordance with the above-discussed embodiments, the lens shading correction gain may be determined based upon a defined set of gain grid points distributed across the imaging frame, wherein the interval between each grid point is defined by a number of pixels (e.g., 8 pixels, 16 pixels etc.). If the location of the current pixel corresponds to a grid point, then the gain value associated with that grid point is applied to the current pixel. However, if the location of the current pixel is between grid points (e.g., G0, G1, G2, and G3 of FIG. 43), then the LSC gain value may be calculated by interpolation of the grid points between which the current pixel is located (Equations 13a and 13b). This process is depicted by the process 528 of FIG. 44. Further, as mentioned above with respect to FIG. 42, in some embodiments, the grid points may be distributed unevenly (e.g., logarithmically), such that the grid points are less concentrated in the center of the LSC region 504, but more concentrated towards the corners of the LSC region 504, typically where lens shading distortion is more noticeable.

[0334] Additionally, as discussed above with reference to FIGS. 47 and 48, the LSC logic 596 may also apply a radial gain component with the grid gain values. The radial gain component may be determined based upon distance of the current pixel from the center of the image (Equations 14-16). As mentioned, using a radial gain allows for the use of single common gain grid for all color components, which may greatly reduce the total storage space required for storing separate gain grids for each color component. This reduction in grid gain data may decrease implementation costs, as grid gain data tables may account for a significant portion of memory or chip area in image processing hardware.

[0335] Next, referring again to the raw processing logic block diagram 562 of FIG. 50, the output of the LSC logic 596 is then passed to a second gain, offset, and clamping (GOC) block 598. The GOC logic 598 may be applied prior to demosaicing (by logic block 600) and may be used to perform auto-white balance on the output of the LSC logic 596. In the depicted embodiment, the GOC logic 598 may be implemented in the same manner as the GOC logic 590 (and the BLC logic 462). Thus, in accordance with the Equation 11 above, the input received by the GOC logic 598 is first offset by a signed value and then multiplied by a gain. The resulting value is then clipped to a minimum and a maximum range in accordance with Equation 12.

[0336] Thereafter, the output of the GOC logic 598 is forwarded to the demosaicing logic 600 for processing to produce a full color (RGB) image based upon the raw Bayer input data. As will be appreciated, the raw output of an image sensor using a color filter array, such as a Bayer filter is “incomplete” in the sense that each pixel is filtered to acquire only a single color component. Thus, the data collected for an individual pixel alone is insufficient to determine color. Accordingly, demosaicing techniques may be used to generate a full color image from the raw Bayer data by interpolating the missing color data for each pixel.

[0337] Referring now to FIG. 58, a graphical process flow 692 that provides a general overview as to how demosaicing may be applied to a raw Bayer image pattern 694 to produce a full color RGB is illustrated. As shown, a 4×4 portion 696 of the raw Bayer image 694 may include separate channels for each color component, including a green channel 698, a red channel 700, and a blue channel 702. Because each imaging pixel in a Bayer sensor only acquires data for one color, the color data for each color channel 698, 700, and 702 may be incomplete, as indicated by the “?” symbols. By applying a demosaicing technique 704, the missing color samples from each channel may be interpolated. For instance, as shown by reference number 706, interpolated data G′ may be used to fill the missing samples on the green color channel Similarly, interpolated data R′ may (in combination with the interpolated data G′ 706) be used to fill the missing samples on the red color channel 708, and interpolated data B′ may (in combination with the interpolated data G′ 706) be used to fill the missing samples on the blue color channel 710. Thus, as a result of the demosaicing process, each color channel (R, G, B) will have a full set of color data, which may then be used to reconstruct a full color RGB image 712.

[0338] A demosaicing technique that may be implemented by the demosaicing logic 600 will now be described in accordance with one embodiment. On the green color channel, missing color samples may be interpolated using a low pass directional filter on known green samples and a high pass (or gradient) filter on the adjacent color channels (e.g., red and blue). For the red and blue color channels, the missing color samples may be interpolated in a similar manner, but by using low pass filtering on known red or blue values and high pass filtering on co-located interpolated green values. Further, in one embodiment, demosaicing on the green color channel may utilize a 5×5 pixel block edge-adaptive filter based on the original Bayer color data. As will be discussed further below, the use of an edge-adaptive filter may provide for the continuous weighting based on gradients of horizontal and vertical filtered values, which reduce the appearance of certain artifacts, such as aliasing, “checkerboard,” or “rainbow” artifacts, commonly seen in conventional demosaicing techniques.

[0339] During demosaicing on the green channel, the original values for the green pixels (Gr and Gb pixels) of the Bayer image pattern are used. However, in order to obtain a full set of data for the green channel, green pixel values may be interpolated at the red and blue pixels of the Bayer image pattern. In accordance with the present technique, horizontal and vertical energy components, respectively referred to as Eh and Ev, are first calculated at red and blue pixels based on the above-mentioned 5×5 pixel block. The values of Eh and Ev may be used to obtain an edge-weighted filtered value from the horizontal and vertical filtering steps, as discussed further below.

[0340] By way of example, FIG. 59 illustrates the computation of the Eh and Ev values for a red pixel centered in the 5×5 pixel block at location (j, i), wherein j corresponds to a row and i corresponds to a column. As shown, the calculation of Eh considers the middle three rows (j−1, j, j+1) of the 5×5 pixel block, and the calculation of Ev considers the middle three columns (i−1, i, i+1) of the 5×5 pixel block. To compute Eh, the absolute value of the sum of each of the pixels in the red columns (i−2, i, i+2) multiplied by a corresponding coefficient (e.g., −1 for columns i−2 and i+2; 2 for column i) is summed with the absolute value of the sum of each of the pixels in the blue columns (i−1, i+1) multiplied by a corresponding coefficient (e.g., 1 for column i−1; −1 for column i+1). To compute Ev, the absolute value of the sum of each of the pixels in the red rows (j−2, j, j+2) multiplied by a corresponding coefficient (e.g., −1 for rows j−2 and j+2; 2 for row j) is summed with the absolute value of the sum of each of the pixels in the blue rows (j−1, j+1) multiplied by a corresponding coefficient (e.g., 1 for row j−1; −1 for row j+1). These computations are illustrated by Equations 44 and 45 below:

[0341] $\begin{array}{cc}\mathrm{Eh}=\mathrm{abs}[2(\left(P\left(j-1,i\right)+P\left(j,i\right)+P\left(j+1,i\right)\right)-\left(P\left(j-1,i-2\right)+P\left(j,i-2\right)+P\left(j+1,i-2\right)\right)-\left(P\left(j-1,i+2\right)+P\left(j,i+2\right)+P\left(j+1,i+2\right)\right]+\mathrm{abs}[\left(P\left(j-1,i-1\right)+P\left(j,i-1\right)+P\left(j+1,i-1\right)\right)-\left(P\left(j-1,i+1\right)+P\left(j,i+1\right)+P\left(j+1,i+1\right)\right]& \left(44\right)\\ \mathrm{Ev}=\mathrm{abs}[2\left(P\left(j,i-1\right)+P\left(j,i\right)+P\left(j,i+1\right)\right)-\left(P\left(j-2,i-1\right)+P\left(j-2,i\right)+P\left(j-2,i+1\right)\right)-(P\left(j+2,i-1\right)+P\left(j+2,i\right)+P\left(j+2,i+1\right]+\mathrm{abs}[\left(P\left(j-1,i-1\right)+P\left(j-1,i\right)+P\left(j-1,i+1\right)\right)-\left(P\left(j+1,i-1\right)+P\left(j+1,i\right)+P\left(j+1,i+1\right)\right]& \left(45\right)\end{array}$
Thus, the total energy sum may be expressed as: Eh+Ev. Further, while the example shown in FIG. 59 illustrates the computation of Eh and Ev for a red center pixel at (j, i), it should be understood that the Eh and Ev values may be determined in a similar manner for blue center pixels.