Camera Modules Patterned with Pi Filter Groups

[0001] The current application is a continuation of U.S. application Ser. No. 13/875,248, filed on May 1, 2013, which application claims priority to U.S. Provisional Application No. 61/641,165, filed on May 1, 2012, the disclosures of which are incorporated herein by reference.

[0002] The present invention relates generally to digital cameras and more specifically to filter patterns utilized in camera modules of array cameras.

[0003] Conventional digital cameras typically include a single focal plane with a lens stack. The focal plane includes an array of light sensitive pixels and is part of a sensor. The lens stack creates an optical channel that forms an image of a scene upon the array of light sensitive pixels in the focal plane. Each light sensitive pixel can generate image data based upon the light incident upon the pixel.

[0004] In a conventional color digital camera, an array of color filters is typically applied to the pixels in the focal plane of the camera's sensor. Typical color filters can include red, green and blue color filters. A demosaicing algorithm can be used to interpolate a set of complete red, green and blue values for each pixel of image data captured by the focal plane given a specific color filter pattern. One example of a camera color filter pattern is the Bayer filter pattern. The Bayer filter pattern describes a specific pattern of red, green and blue color filters that results in 50% of the pixels in a focal plane capturing green light, 25% capturing red light and 25% capturing blue light.

[0005] Systems and methods in accordance with embodiments of the invention pattern array camera modules with π filter groups. In one embodiment, an array camera module includes: an M×N imager array including a plurality of focal planes, each focal plane including an array of light sensitive pixels; an M×N optic array of lens stacks, where each lens stack corresponds to a focal plane, and where each lens stack forms an image of a scene on its corresponding focal plane; where each pairing of a lens stack and its corresponding focal plane thereby defines a camera; where at least one row in the M×N array of cameras includes at least one red color camera, at least one green color camera, and at least one blue color camera; and where at least one column in the M×N array of cameras includes at least one red color camera, at least one green color camera, and at least one blue color camera.

[0006] In another embodiment, M and N are each greater than two and at least one of M and N is even; color filters are implemented within the cameras in the array camera module such that the array camera module is patterned with at least one π filter group including: a 3×3 array of cameras including: a reference camera at the center of the 3×3 array of cameras; two red color cameras located on opposite sides of the 3×3 array of cameras; two blue color cameras located on opposite sides of the 3×3 array of cameras; and four green color cameras surrounding the reference camera.

[0007] In yet another embodiment, each of the four green color cameras surrounding the reference camera is disposed at a corner location of the 3×3 array of cameras.

[0008] In still another embodiment, M is four; N is four; the first row of cameras of the 4×4 array camera module includes, in the following order, a green color camera, a blue color camera, a green color camera, and a red color camera; the second row of cameras of the 4×4 array camera module includes, in the following order, a red color camera, a green color camera, a red color camera, and a green color camera; the third row of cameras of the 4×4 array camera module includes, in the following order, a green color camera, a blue color camera, a green color camera, and a blue color camera; and the fourth row of cameras of the 4×4 array camera module includes, in the following order, a blue color camera, a green color camera, a red color camera, and a green color camera.

[0009] In an even further embodiment, M is four; N is four; the first row of cameras of the 4×4 array camera module includes, in the following order, a red color camera, a green color camera, a blue color camera, and a green color camera; the second row of cameras of the 4×4 array camera module includes, in the following order a green color camera, a red color camera, a green color camera, and a red color camera; the third row of cameras of the 4×4 array camera module includes, in the following order, a blue color camera, a green color camera, a blue color camera, and a green color camera; and the fourth row of cameras of the 4×4 array camera module includes, in the following order, a green color camera, a red color camera, a green color camera, and a blue color camera.

[0010] In still another embodiment, the reference camera is a green color camera.

[0011] In still yet another embodiment, the reference camera is one of: a camera that incorporates a Bayer filter, a camera that is configured to capture infrared light, and a camera that is configured to capture ultraviolet light.

[0012] In a still yet further embodiment, each of the two red color cameras is located at a corner location of the 3×3 array of cameras, and each of the two blue color cameras is located at a corner location of the 3×3 array of cameras.

[0013] In another embodiment, at least one color filter is implemented on the imager array.

[0014] In a further embodiment, at least one color filter is implemented on a lens stack.

[0015] In another embodiment, a 3×3 array camera module includes: a 3×3 imager array including a 3×3 arrangement of focal planes, each focal plane including an array of light sensitive pixels; a 3×3 optic array of lens stacks, where each lens stack corresponds to a focal plane, and where each lens stack forms an image of a scene on its corresponding focal plane; where each pairing of a lens stack and its corresponding focal plane thereby defines a camera; where the 3×3 array of cameras includes: a reference camera at the center of the 3×3 array of cameras; two red color cameras located on opposite sides of the 3×3 array of cameras; two blue color cameras located on opposite sides of the 3×3 array of cameras; and four green color cameras, each located at a corner location of the 3×3 array of cameras; where each of the color cameras is achieved using a color filter.

[0016] In a further embodiment, at least one color filter is implemented on the imager array to achieve a color camera.

[0017] In a still yet further embodiment, at least one color filter is implemented within a lens stack to achieve a color camera.

[0018] In yet another embodiment, the reference camera is a green color camera.

[0019] In an even further embodiment, the reference camera is one of: a camera that incorporates a Bayer filter, a camera that is configured to capture infrared light, and a camera that is configured to capture ultraviolet light.

[0020] In another embodiment, a method of patterning an array camera module with at least one π filter group includes: evaluating whether an imager array of M×N focal planes, where each focal plane comprises an array of light sensitive pixels, includes any defective focal planes; assembling an M×N array camera module using: the imager array of M×N focal planes; an M×N optic array of lens stacks, where each lens stack corresponds with a focal plane, where the M×N array camera module is assembled so that: each lens stack and its corresponding focal plane define a camera; color filters are implemented within the array camera module such that the array camera module is patterned with at least one π filter group including: a 3×3 array of cameras including: a reference camera at the center of the 3×3 array of cameras; two red color cameras located on opposite sides of the 3×3 array of cameras; two blue color cameras located on opposite sides of the 3×3 array of cameras; and four green color cameras surrounding the reference camera; and where the array camera module is patterned with the at least one π filter group such that a camera that includes a defective focal plane is a green color camera.

[0021] In a further embodiment, at least one color filter is implemented on the imager array.

[0022] In a still further embodiment, at least one color filter is implemented within a lens stack.

[0023] In an even further embodiment, the reference camera is a green color camera.

[0024] In still yet another embodiment, the reference camera is one of: a camera that incorporates a Bayer filter, a camera that is configured to capture infrared light, and a camera that is configured to capture ultraviolet light.

[0040] Turning now to the drawings, systems and methods for patterning array cameras with π filter groups in accordance with embodiments of the invention are illustrated. In many embodiments, camera modules of an array camera are patterned with one or more π filter groups. The term patterned here refers to the use of specific color filters in individual cameras within the camera module so that the cameras form a pattern of color channels within the array camera. The term color channel or color camera can be used to refer to a camera that captures image data within a specific portion of the spectrum and is not necessarily limited to image data with respect to a specific color. The term Bayer camera can be used to refer to a camera that captures image data using the Bayer filter pattern on the image plane. In many embodiments, a color channel can include a camera that captures infrared light, ultraviolet light, extended color and any other portion of the visible spectrum appropriate to a specific application. The term π filter group refers to a 3×3 group of cameras including a central camera and color cameras distributed around the central camera to minimize occlusion zones. The central camera of a π filter group can be used as a reference camera when synthesizing an image using image data captured by an imager array. A camera is a reference camera when its viewpoint is used as the viewpoint of the synthesized image. The central camera of a π filter group is surrounded by color cameras in a way that minimizes occlusion zones for each color camera when the central camera is used as a reference camera. Occlusion zones are areas surrounding foreground objects not visible to cameras that are spatially offset from the reference camera due to the effects of parallax. In several embodiments, the central camera is a green camera while in other embodiments the central camera captures image data from any appropriate portion of the spectrum. In a number of embodiments, the central camera is a Bayer camera (i.e. a camera that utilizes a Bayer filter pattern to capture a color image). In many embodiments, a π filter group is a 3×3 array of cameras with a green color camera at each corner and a green color camera at the center which can serve as the reference camera with a symmetrical distribution of red and blue cameras around the central green camera. The symmetrical distribution can include arrangements where either red color cameras are directly above and below the center green reference camera with blue color cameras directly to the left and right, or blue color cameras directly above and below the green center reference camera with red color cameras directly to the left and right.

[0041] Camera modules of dimensions greater than a 3×3 array of cameras can be patterned with π filter groups in accordance with many embodiments of the invention. In many embodiments, patterning a camera module with π filter groups enables an efficient distribution of cameras around a reference camera that reduces occlusion zones. In several embodiments, patterns of π filter groups can overlap with each other such that two overlapping π filter groups on a camera module share common cameras. When overlapping π filter groups do not span all of the cameras in the camera module, cameras that are not part of a π filter group can be assigned a color to minimize occlusion zones in the resulting camera array.

[0042] In several embodiments, patterning a camera module with π filter groups can result in reference cameras that are not in the center of the camera module. Additionally, color cameras surrounding the reference camera need not be uniformly distributed but need only be distributed in a way to minimize occlusion zones of each color from the perspective of the reference camera. Utilization of a reference camera in a π filter group to synthesize an image from captured image data can be significantly less computationally intensive than synthesizing an image using the same image data from a virtual viewpoint.

[0043] High quality images or video can be captured by an array camera including a camera module patterned with π filter groups utilizing a subset of cameras within the camera module (i.e. not requiring that all cameras on a camera module be utilized). Similar techniques can also be used for efficient generation of stereoscopic 3D images utilizing image data captured by subsets of the cameras within the camera module.

[0044] Patterning camera modules with π filter groups also enables robust fault tolerance in camera modules with multiple π filter groups as multiple possible reference cameras can be utilized if a reference camera begins to perform sub optimally. Patterning camera modules with π filter groups also allows for yield improvement in manufacturing camera modules as the impact of a defective focal plane on a focal plane array can be minimized by simply changing the pattern of the color lens stacks in an optic array. Various π filter groups and the patterning of camera modules with π filter groups in accordance with embodiments of the invention are discussed further below.

[0045] In many embodiments, an array camera includes a camera module and a processor. An array camera with a camera module patterned with π filter groups in accordance with an embodiment of the invention is illustrated in FIG. 1. The array camera 100 includes a camera module 102 as an array of individual cameras 104 where each camera 104 includes a focal plane with a corresponding lens stack. An array of individual cameras refers to a plurality of cameras in a particular arrangement, such as (but not limited to) the square arrangement utilized in the illustrated embodiment. The camera module 102 is connected 106 to a processor 108. In the illustrated embodiment, a camera 104 labeled as “R” refers to a red camera with a red filtered color channel, “G” refers to a green camera with a green filtered color channel and “B” refers to a blue camera with a blue filtered color channel. Although a specific array camera is illustrated in FIG. 1, any of a variety of different array camera configurations can be utilized in accordance with many different embodiments of the invention.

[0046] Camera modules in accordance with embodiments of the invention can be constructed from an imager array and an optic array. A camera module in accordance with an embodiment of the invention is illustrated in FIG. 2. The camera module 200 includes an imager array 230 including an array of focal planes 240 along with a corresponding optic array 210 including an array of lens stacks 220. Within the array of lens stacks, each lens stack 220 creates an optical channel that forms an image of a scene on an array of light sensitive pixels within a corresponding focal plane 240. Each pairing of a lens stack 220 and focal plane 240 forms a single camera 104 within the camera module. Each pixel within a focal plane 240 of a camera 104 generates image data that can be sent from the camera 104 to the processor 108.

[0047] In several embodiments, color filters in individual cameras can be used to pattern the camera module with π filter groups. These cameras can be used to capture data with respect to different colors, or a specific portion of the spectrum. In contrast to applying color filters to the pixels of the camera, color filters in many embodiments of the invention are included in the lens stack. For example, a green color camera can include a lens stack with a green light filter that allows green light to pass through the optical channel. In many embodiments, the pixels in each focal plane are the same and the light information captured by the pixels is differentiated by the color filters in the corresponding lens stack for each filter plane. Although a specific construction of a camera module with an optic array including color filters in the lens stacks is described above, camera modules including π filter groups can be implemented in a variety of ways including (but not limited to) by applying color filters to the pixels of the focal planes of the camera module similar to the manner in which color filters are applied to the pixels of a conventional color camera. In several embodiments, at least one of the cameras in the camera module can include uniform color filters applied to the pixels in its focal plane. In many embodiments, a Bayer filter pattern is applied to the pixels of one of the cameras in a camera module. In a number of embodiments, camera modules are constructed in which color filters are utilized in both the lens stacks and on the pixels of the imager array.

[0048] In several embodiments, an array camera generates image data from the multiple focal planes and uses a processor to synthesize one or more images of a scene. In certain embodiments, the image data captured by a single focal plane in the sensor array can constitute a low resolution image (the term low resolution here is used only to contrast with higher resolution images), which the processor can use in combination with other low resolution image data captured by the camera module to construct a higher resolution image through Super Resolution processing. Super Resolution processes that can be used to synthesize high resolution images using low resolution images captured by an array camera are discussed in U.S. patent application Ser. No. 12/967,807 entitled “Systems and Methods for Synthesizing High Resolution Images Using Super-Resolution Processes”, filed Dec. 14, 2010, the disclosure of which is hereby incorporated by reference in its entirety.

[0049] Although specific array cameras are discussed above, many different array cameras are capable of utilizing π filter groups in accordance with embodiments of the invention. Camera modules utilizing π filter groups in accordance with embodiments of the invention are described in further detail below.

Patterning with π Filter Groups

[0050] Camera modules can be patterned with π filter groups in accordance with embodiments of the invention. In several embodiments, π filter groups utilized as part of a camera module can each include a central camera that can function as a reference camera surrounded by color cameras in a way that reduces occlusion zones for each color. In certain embodiments, the camera module is arranged in a rectangular format utilizing the RGB color model where a reference camera is a green camera surrounded by red, green and blue cameras. In several embodiments, a number of green cameras that is twice the number of red cameras and twice the number of blue cameras surround the reference camera. However, any set of colors from any color model can be utilized to detect a useful range of colors in addition to the RGB color model, such as the cyan, magenta, yellow and key (CMYK) color model or red, yellow and blue (RYB) color model.

[0051] In several embodiments, two π filter groups can be utilized in the patterning of a camera module when the RGB color model is used. One π filter group is illustrated in FIG. 3A and the other π filter group is illustrated FIG. 3B. Either of these π filter groups can be used to pattern any camera module with dimensions greater than a 3×3 array of cameras.

[0052] In embodiments with a 3×3 camera module, patterning of the camera module with π filter group includes only a single π filter group. A π filter group on a 3×3 camera module in accordance with an embodiment of the invention is illustrated in FIG. 3A. The π filter group 300 includes a green camera at each corner, a green reference camera in the center notated within a box 304, blue cameras above and below the reference camera, and red cameras to the left and right sides of the reference camera. In this configuration, the number of green cameras surrounding the central reference camera is twice the number of red cameras and twice the number of blue cameras. An alternative to the π filter group described in FIG. 3A is illustrated in FIG. 3B in accordance with an embodiment of the invention. This π filter group also includes green cameras at the corners with a green reference camera 352 at the center, as denoted with a box. However, unlike FIG. 3A, the red cameras shown in FIG. 3B are above and below, and the blue cameras are to the left and right side of the reference camera. As with the π filter group shown in FIG. 3A, the π filter group in FIG. 3B includes a central reference camera surrounded by a number of green cameras that is twice the number of red cameras and twice the number of blue cameras. As discussed above, the reference camera need not be a green camera. In several embodiments, the configurations in FIGS. 3A and 3B can be modified to include a central camera that employs a Bayer color filter. In other embodiments, the central camera is an infrared camera, an extended color camera and/or any other type of camera appropriate to a specific application. In further embodiments, any of a variety of color cameras can be distributed around the reference camera in a manner that reduces occlusion zones with respect to each color channel.

[0053] Any camera module with dimensions at and above 3×3 cameras can be patterned with one or more π filter groups, where cameras not within a π filter group are assigned a color that reduces or minimizes the likelihood of occlusion zones within the camera module given color filter assignments of the π filter groups. A 4×4 camera module patterned with two π filter groups in accordance with an embodiment of the invention is illustrated in FIG. 4. The camera module 400 includes a first π filter group 402 of nine cameras centered on a reference green camera 404. A second π filter group 410 is diagonally located one camera shift to the lower right of the first π filter group. The second π filter group shares the four center cameras 412 of the camera module 400 with the first π filter group. However, the cameras serve different roles (i.e. different cameras act as reference cameras in the two π filter groups). As illustrated in FIG. 4, the two cameras at the corners 406 and 408 of the camera module are not included in the two π filter groups, 402 and 410. The color filters utilized within these cameras are determined based upon minimization of occlusion zones given the color filter assignments of the cameras that are part of the two π filter groups, 402 and 410. Due to the patterning of the π filter groups, there is an even distribution of blue color cameras around the reference camera, but there is no red color camera above the reference camera. Therefore, selecting the upper right corner camera 406 to be red provides red image data from a viewpoint above the reference camera and the likelihood of occlusion zones above and to the right of the foreground images in a scene for the reference camera 404 and the center camera of the second π filter group is minimized. Similarly, selecting the lower left corner camera 408 to be blue provides blue image data from a viewpoint to the left of the reference camera and the likelihood of occlusion zones below and to the left of the foreground images in a scene for the reference camera 404 and the center camera of the second π filter group is minimized. Thereby, a camera module with dimensions greater than 3×3 can be patterned with π filter groups with colors assigned to cameras not included in any π filter group to reduce and/or minimize occlusion zones as discussed above. Although specific π filter groups are discussed above, any of a variety of π filter groups can pattern a camera module in accordance with many different embodiments of the invention.

Multiple Reference Camera Options with Equivalent Performance

[0054] The use of multiple π filter groups to pattern a camera module in accordance with embodiments of the invention enables multiple cameras to be used as the reference camera with equivalent performance. A 4×4 camera module with two π filter groups in accordance with an embodiment of the invention is illustrated in FIG. 5. The camera module 500 includes two π filter groups 502, 506 where the central camera of each π filter group 504, 508 can act as a reference camera. Irrespective of the reference camera that is selected, the distribution of cameras around the reference camera is equivalent due to the use of π filter groups. Thereby, if a camera module 500 detects a defect with the a reference camera 504, the camera module 500 can switch to using the camera at the center of another π filter group as a reference camera 508 to avoid the defects of the first reference camera 504. Furthermore, patterning with π filter groups does not require that the reference camera or a virtual viewpoint be at the center of a camera module but rather that the reference camera is surrounded by color cameras in a way that reduces occlusion zones for each color. Although a specific camera module is discussed above, camera modules of any number of different dimensions can be utilized to create multiple reference camera options in accordance with embodiments of the invention.

[0055] Manufacturing Yield Improvement

[0056] Manufacturing processes inherently involve variations that can result in defects. In some instances the manufacturing defects may be severe enough to render an entire focal plane within an imager array inoperable. If the failure of the focal plane results in the discarding of the imager array, then the cost to manufacture array cameras is increased. Patterning camera modules with π filter groups can provide high manufacturing yield because the allocation of color filters in the optical channels of the optic array can be used to reduce the impact that a faulty focal plane has with respect to the creation of occlusion zones in the images synthesized using the image data captured by the array camera.

[0057] In many embodiments, the light sensed by the pixels in a focal plane of an imager array is determined by a color filter included in the optical channel that focuses light onto the focal plane. During manufacture, defects in a focal plane can be detected. When a defect is detected, the color filter pattern of the optical channels in the optic array can be determined so that the defective focal plane does not result in an increase in the size of occlusion zones. Typically, this means patterning camera modules with π filter groups in such a way that the presence of the defective focal plane does not reduce the number of red or blue cameras in the camera array (i.e. a filter pattern is used that results in a green channel being assigned to the defective focal plane, which reduces the number of green cameras in the camera array by one camera).

[0058] A process for detecting faulty focal planes before combining an optic array and imager array to create a camera module in accordance with embodiments of the invention is illustrated in FIG. 6A. In the illustrated process, the color filter patterns are patterned on the optics array and not on the pixels of the imager array. By manufacturing different types of optics arrays (with different filter patterns, a process can systematically choose a specific optics array to force the faulty focal plane to pair with a color of a certain filter to ensure that the size of the occlusion zones in a given color channel are reduced and/or minimized. The process 600 includes testing (602) an imager array for faulty focal planes. After testing (602) the imager array, a decision (604) is made as to whether a faulty focal plane is detected on the imager array. If a faulty focal plane is detected, then an optic array is selected based upon the location of the faulty focal plane (606). In many embodiments, an optic array is selected that reduces the effect of the faulty focal plane by assigning color filters to the operational focal planes in a way that minimizes the impact of the faulty focal plane on the creation of occlusion zones within images synthesized using image data captured by the imager array. Further discussion of selecting different optic arrays that reduce occlusion zones when there is a faulty focal plane is provided below with reference to FIGS. 6B and 6C. After selecting (606) an optic array based upon the location of the faulty focal plane, the selected optic array is combined (608) with the imager array to create a camera module. If a faulty focal plane is not detected, then any of a variety of optic arrays including filter patterns based on π filter groups can be combined (608) with the tested imager array to create a camera module. As is discussed further below, a typical process can include a default optic array including a first filter pattern based on π filter groups and a second filter pattern based on π filter groups can be utilized when specific defects are detected that would result in the faulty focal plane reducing the number of color cameras (or even specific color cameras such as color cameras around the outside of the camera module) in the camera module when the first filter pattern is used.

[0059] The manner in which modifying color filter assignments can reduce the impact of a faulty focal plane is illustrated in FIGS. 6B and 6C. A camera module with a faulty red camera is illustrated in FIG. 6B. The camera module 620 includes a first π filter group 628 with a possible reference camera 622 at the center, a second π filter group 632 with a possible reference camera 630 at the center and a faulty red camera 624 below both π filter groups 628 and 632. There is a lack of red image data below both the possible reference cameras 622 and 630 due to the faulty red camera. Therefore, irrespective of which of the two cameras at the center of a π filter group is chosen as the reference camera. Accordingly, combining an optic array including the filter pattern illustrated in FIG. 6B to an imager with the indicated faulty focal plane results in a defective red camera that prevents the capture of red color information below any reference camera, increasing the likelihood of occlusion zones below foreground objects. However, an optic array patterned using π filter groups in different locations can result in all of the blue and red color filters being assigned to cameras that are active. In this way, the faulty focal plane only impacts the number of green cameras and does so in a way that reduces the likelihood of occlusion zones in an image synthesized using the image data captured by the resulting camera module. Stated another way, yield can be improved under certain circumstances by combining the imager array that includes the faulty focal plane with an optic array that assigns the color filters of the active cameras based on π filter groups in a way that results in color information being captured around the reference camera in a way that minimizes the likelihood of occlusion zones given the location of the faulty focal plane.

[0060] A camera module with the faulty focal plane of FIG. 6B but with an optic array patterned with π filter groups in such a way that the faulty focal plane does not reduce the capture of red or blue image data around the reference camera module is illustrated in FIG. 6C. Relative to the pattern of the optic array of FIG. 6B, the optic array of FIG. 6C is flipped along the center vertical bisecting axis 626 of the optic array and includes two π filter groups 628′ and 632′. The lens stack associated with the faulty focal plane is green 654, as opposed to red 624 in FIG. 6B. As there are multiple green cameras below all possible reference cameras 652, 656 in FIG. 6C, the loss of a green camera 654 is less impactful as opposed to the impact from the loss of the red camera 624 in FIG. 6B. Therefore, the impact of faulty focal planes on an imager array can be reduced by combining the faulty imager array with an optic array specifically selected to assign color filters to the focal planes in the imager array in a manner that reduces the likelihood that the faulty focal plane will create an occlusion zone in any of the color channels captured by the resulting camera module. Although the example above discusses reducing red occlusion zones, the impact of a defective focal plane in any of the locations in an imager array can be similarly minimized by appropriate selection of a filter pattern based on π filter groups. Although specific examples of camera modules patterned with π filter groups to minimize yield loss due to faulty focal planes are described above, any of a variety of alternative color filter patterns including π filter groups can be utilized to increase manufacturing yield in accordance with embodiments of the invention.

[0061] In many embodiments, Super Resolution processes can be used to synthesize high resolution images using low resolution images captured by an array camera including pairs of stereoscopic 3D images as disclosed in U.S. patent application Ser. No. 12/967,807 entitled “Systems and Methods for Synthesizing High Resolution Images Using Super-Resolution Processes”, filed Dec. 14, 2010, the disclosure of which is incorporated by reference above. Stereoscopic 3D image pairs are two images of a scene from spatially offset viewpoints that can be combined to create a 3D representation of the scene. The use of a filter pattern including π filter groups can enable the synthesis of stereoscopic 3D images in a computationally efficient manner. Image data captured by less than all of the cameras in the array camera can be used to synthesize each of the images that form the stereoscopic 3D image pair.

[0062] Patterning with π filter groups enables an efficient distribution of cameras around a reference camera that reduces occlusion zones and reduces the amount of image data captured by the camera module that is utilized to synthesize each of the images in a stereoscopic 3D image pair. In many embodiments, different subsets of the cameras are used to capture each of the images that form the stereoscopic 3D image pair and each of the subsets includes a π filter group. In many embodiments, the images that form the stereoscopic 3D image pair are captured from a virtual viewpoint that is slightly offset from the camera at the center of the π filter group. The central camera of a π filter group is surrounded by color cameras in a way that minimizes occlusion zones for each color camera when the central camera is used as a reference camera. When the virtual viewpoint is proximate the center of a π filter group, the benefits of the distribution of color cameras around the virtual viewpoint are similar.

[0063] A left virtual viewpoint for a stereoscopic 3D image pair captured using a camera module patterned using π filter groups is illustrated in FIG. 7A. The left virtual viewpoint 704 is taken from image data from the 12 circled cameras G₁-G₃, G₅-G₇, B₁-B₂, B₄, and R₂-R₃ that form a 3×4 array. The virtual viewpoint is offset relative to the green camera G₃, which is the center of a π filter group 706. A right virtual viewpoint used to capture the second image in the stereoscopic pair using the camera module shown in FIG. 5A is illustrated in FIG. 7B. The right virtual viewpoint 754 is taken from image data from the 12 circled cameras B₁-B₃, G₂-G₄, G₆-G₈, R₁, R₃-R₄ that form a 3×4 array. The virtual viewpoint is offset relative to the green camera G₆, which is the center of a π filter group 756. Therefore, a single array camera can capture 3D images of a scene using image data from a subset of the cameras to synthesize each of the images that form the stereoscopic pair. By utilizing the image data captured by less than all of the cameras in the camera module, the computational complexity of generating the stereoscopic 3D image pair is reduced. In addition, the location of the view points of each of the images proximate a camera that is the center of a π filter group reduces the likelihood of occlusion zones in the synthesized images.

[0064] Although specific viewpoints and subsets of cameras for synthesizing stereoscopic 3D image pairs are illustrated in FIGS. 7A and 7B, stereoscopic image pairs can be generated using subsets of cameras in any of a variety of camera modules in accordance with embodiments of the invention.

[0065] Array cameras with camera modules patterned with π filter groups can utilize less than all of the available cameras in operation in accordance with many embodiments of the invention. In several embodiments, using fewer cameras can minimize the computational complexity of generating an image using an array camera and can reduce the power consumption of the array camera. Reducing the number of cameras used to capture image data can be useful for applications such as video, where frames of video can be synthesized using less than all of the image data that can be captured by a camera module. In a number of embodiments, a single π filter group can be utilized to capture an image. In many embodiments, image data captured by a single π filter group is utilized to capture a preview image prior to capturing image data with a larger number of cameras. In several embodiments, the cameras in a single IF filter group capture video image data. Depending upon the requirements of a specific application, image data can be captured using additional cameras to increase resolution and/or provide additional color information and reduce occlusions.

[0066] A π filter group within a camera module that is utilized to capture image data that can be utilized to synthesize an image is illustrated in FIG. 8. In the illustrated embodiments, the reference camera is boxed and utilized cameras are encompassed in a dotted line. The camera module 800 includes a π filter group of cameras generating image data G₁-G₂, G₅-G₆, B₁-B₂ and R₂-R₃ with reference camera G₃. FIG. 8 illustrates how the cameras in a π filter group can be utilized to capture images. Image data can be acquired using additional cameras for increased resolution and to provide additional color information in occlusion zones. Accordingly, any number and arrangement of cameras can be utilized to capture image data using a camera module in accordance with many different embodiments of the invention.

[0067] Color filter patterns for any array of cameras having dimensions greater than 3×3 can be constructed in accordance with embodiments of the invention. In many embodiments, processes for constructing color filter patterns typically involve assigning color filters to the cameras in a camera module to maximize the number of overlapping π filter groups. In the event that there are cameras that cannot be included in a π filter group, then color filters can be assigned to the cameras based upon minimizing occlusions around the camera that is to be used as the reference camera for the purposes of synthesizing high-resolution images.

[0068] A process for assigning color filters to cameras in a camera module in accordance with an embodiment of the invention is illustrated in FIG. 9. The process 900 includes selecting (902) a corner of the array, assigning (904) a π filter group to the selected corner. The π filter group occupies a 3×3 grid. Color filters can be assigned (906) to the remaining cameras in such a way to maximize the number of overlapping π filter groups within the array. In the event that there are cameras to which color filters are not assigned, the cameras are assigned (908) color filters that reduce the likelihood of occlusion zones in images synthesized from the viewpoint of a camera selected as the reference camera for the array. At which point, all of the cameras in the array are assigned color filters. As noted above, the presence of multiple π filter groups provides benefits including (but not limited to) robustness to failures in specific cameras within the array and the ability to synthesize images with fewer than all of the cameras in the camera module utilizing image data captured by at least one π filter group.

[0069] The process of generating a simple filter pattern for a 5×5 array using π filter groups is illustrated in FIGS. 10A-10D. The process starts with the selection of the top left corner of the array. A π filter group is assigned to the 3×3 group of cameras in the top left corner (cameras G₁-G₅, B₁-B₂, and R₁-R₂). A second overlapping π filter group is created by adding three green cameras and a blue camera and a red camera (G₆-G₈ and B₃ and R₃). A third overlapping π filter group is created by adding another three green cameras and a blue camera and a red camera (G₉-G₁₁ and B₄ and R₄). A fifth and sixth π filter groups are created by adding a single green camera, blue camera and red camera (G₁₂, B₅, R₅ and G₁₃, B₆, R₆). In the event that central camera (G₆) fails, a camera at the center of another π filter group can be utilized as the reference camera (e.g. G₃).

[0070] A similar process for generating a simple filter pattern for a 4×5 array using π filter groups is illustrated in FIGS. 11A-11D. The process is very similar with the exception that two cameras are not included in π filter groups. Due to the fact that there are no blue cameras below the camera G₆ (which is the center of a π filter group), the cameras that do not form part of a π filter group are assigned as blue cameras (B₅ and B₆). As can readily be appreciated similar processes can be applied to any array larger than a 3×3 array to generate a color filter pattern incorporating π filter groups in accordance with embodiments of the invention.

[0071] While the above description contains many specific embodiments of the invention, these should not be construed as limitations on the scope of the invention, but rather as an example of one embodiment thereof. It is therefore to be understood that the present invention may be practiced otherwise than specifically described, without departing from the scope and spirit of the present invention. Thus, embodiments of the present invention should be considered in all respects as illustrative and not restrictive.