Image processing device and image processing method

[0002] 1. Field of the Invention

[0003] The present invention relates to an image processing device and image processing method for converting a color image into a binary black and white image.

[0004] 2. Description of the Related Art

[0005] Recently facsimile machines and copying machines (photocopiers) capable of scanning color images have been developed, and demand for these machines has increased along with the spread of high quality color printers and displays.

[0006] These image processing devices capable of processing color images utilize an optical color sensor to scan in an image in the three primary colors (RGB). However, if an associated printer or display is set to a black-and-white output mode or is not provided with a color output or display function, the scanned color image is converted to a black and white image based on lightness of the original color image.

[0007] FIG. 4 of the accompanying drawings illustrates processing steps used in such a conversion. Here, image data scanned-in in the three primary colors (RGB) is converted into multi-bit-depth color data according to L*a*b* color, one of the CIE (International Council on Illumination) standard model color systems, and then further converted to binary image data.

[0008] Referring to FIG. 4, the RGB image data scanned in by a scanner Sc is converted to CIE LAB color-system data using an RGB-Lab converter 100. Next, lightness L data (multiple bit-depth data) is converted to binary data B&W using a dithering unit 101, and output from a printer Pr. The dithering unit 101 has only one dither pattern. The values inside the parentheses indicate the bit number for each data.

[0009] The dithering process divides an image into a plurality of small areas, and changes the density of the black pixels in each area to provide an intermediary gray-scale image having a plurality of pixel density levels as a whole. Specifically, an area defined by n×n pixels (a dot matrix) is used as a unit of gradation (or an original image is divided into a plurality of matrices), and a dither pattern in which the threshold values between white and black are dispersed is applied to each matrix to obtain a binary image. In the dithering process, the lightness L of each pixel in the matrix is compared with a corresponding threshold value of the dither pattern; the pixel is converted to white if the lightness of the pixel is brighter than the threshold, and the pixel is converted to black if its lightness is darker than the threshold.

[0010] FIG. 5 of the accompanying drawings illustrates the color spacing according to the Lab color system. In the Lab color system, lightness is represented as L and chromaticity, showing hue and saturation, is represented as “a” and “b”. Saturation “c” is given by (a²+b²)^0.5, and hue “h” is given by tan⁻¹(b/a). Further, in the color spacing of the Lab color system, “+a” represents red levels, “−a” represents the green level, “+b” represents the yellow level, and “−b” represents the blue level. As the color values extend to the limits of each range, the colors become more vivid, and as the values approach the center, the colors become softer.

[0011] In a conventional image processing system, even when the chromaticity “a”, “b” of dot matrices in a single original image differ, the matrices are converted into the same bilevel black and white data as long as their lightness L is the same. Therefore, the black and white image that is output upon color-to-monochrome conversion can look unnatural in comparison to the original color image.

[0012] An object of the present invention is to propose an image processing device and image processing method which can produce a natural-looking black-and-white image from an original color image.

[0013] In order to accomplish this object, the present invention uses different dither patterns according to the hue and/or saturation of an original color image (pixels). Specifically, a plurality of dither patterns are prepared beforehand, and a single dither pattern is then selected based on the hue and/or saturation of the original color image. The lightness is not considered in selecting the dither pattern.

[0014] According to one aspect of the invention, there is provided an image processing machine that detects lightness, hue and/or saturation of an original color image, presents a plurality of dither patterns, selects a single dither pattern depending upon the hue and/or saturation of the color image and carries out the dither process (or binary conversion of color image) to the lightness using the selected dither pattern. For example, if an original color image is expressed numerically in the Lab color system, two dither patterns may temporarily be provided, and a single dither pattern is selected depending upon the saturation “c” (c=(a²+b²)^0.5). Consequently, the resulting black-and-white image does not appear unnatural when output. The black-and-white image may be output onto recording paper or shown on an LCD or CRT display. The plurality of dither patterns may be stored in a memory of the image processing device beforehand.

[0015] The present invention may be modified as follows: only when the lightness of a particular area in the target image is equivalent to that of a neighboring area, hue and/or saturation is used to select the dithering pattern. When the chromaticities of adjoining regions in an original image differ and the lightness remains the same, the eye perceives the difference. Thus when, for example, the total lightness of the pixels in the neighboring dot matrices are equal, the chromaticity (hue and/or saturation) of the image is considered in the selection of the dithering pattern. Particularly, the dither pattern selection depending upon the hue and/or saturation may be performed when the neighboring regions on the same scanning line have the same lightness.

[0016] FIG. 1 illustrates a diagram outlining the flow of operations for binary conversion of an image in the image processing device of the present invention.

[0017] FIG. 1A illustrates another image processing device according to the present invention.

[0018] FIGS. 2A to 2C illustrate a dithering process (binary conversion process) applied to a relatively vivid color: specifically, FIG. 2A illustrates a dot matrix of original image, FIG. 2B illustrates a Bayer dither pattern and FIG. 2C illustrates a resulting binary data.

[0019] FIGS. 3A to 3C illustrate another dithering process which is applied to a relatively dark or soft color: specifically, FIG. 3A is a dot matrix, FIG. 3B is a vortex dither pattern and FIG. 3C is a resulting binary data.

[0020] FIG. 4 is a diagram showing the order of operations for binary conversion of a color image in a conventional image processing device.

[0021] FIG. 5 is a diagram of the color space in the Lab color system.

[0022] Embodiments of the present invention will now be described in reference to FIGS. 1 to 3C and 5 of the accompanying drawings.

[0023] Referring to FIG. 1, illustrated is a block diagram of an image processing device of the present embodiment. This image processing device may be applied to facsimile machines, photocopiers and the like which converts a color image into a black-and-white image. The numbers in parentheses indicate the bit depth for each variable.

[0024] An original color image is scanned in by a scanner Sc and RGB data is output to an RGB-Lab converter 1 for conversion to Lab color data. Then, the RGB-Lab converter 1 outputs data having lightness L to a dithering unit 2 to provide two sets of binary data B&W. In this embodiment, two dither patterns P1, P2 are already prepared in the dithering unit 2. In the dithering unit 2, two dithering processes are performed using the two dither patterns P1, P2, respectively. The two sets of binary data B&W are output to a selecting unit 5.

[0025] In the meantime, the RGB-Lab converter 1 outputs data about “a” and “b” to a saturation calculating unit 3 which calculates the saturation “c” of the image according to the equation c=(a²+b²)^0.5(see FIG. 5). A decision unit 4 compares the saturation “c” calculated by the saturation calculating unit 3 with the predetermined threshold value, and based upon this result, the selecting unit 5 selects one of the two sets of binary data B&W. The selected binary data is output to a printer Pr. It should be noted here that the saturation calculating unit 3 is only necessary to provide a certain value uniquely corresponding to the saturation value “c”. For instance, it is satisfactory for the saturation calculating unit 3 to merely provide (a²+b²). The decision unit 4 can compare this value to a threshold value, thereby shortening the operation (calculation for acquiring the root of (a²and b²) is omitted).

[0026] The present embodiment thus uses the saturation “c” as a parameter to select an appropriate set of data from two candidates. The printer Pr prints the selected black-and-white image. Since the image to be printed is determined based on the saturation, the original color image is converted into a binary image that does not look unnatural to the eye. It should be noted that only two dither patterns P1, P2, are utilized in the illustrated embodiment, but three or more dither patterns may be used for the binary conversion in the dithering unit 2. Further, it is also possible to select one dither pattern before the dithering process and to output only one binary image from the dithering unit 2. This modification is illustrated in FIG. 1A. As depicted, the data from the decision unit 4 is fed to the dithering unit 2, the dither pattern P1 or P2 is selected based on the data from the decision unit 4, and the dithering process is performed with the selected dither pattern. A single binary image B&W is output to the printer Pr from the dithering unit 2.

[0027] It should also be noted that the color image of the present embodiment has been represented according to a Lab color system, but other color systems like XYZ (Yxy) can be used as well. The binary image converted by the image processing device of the present embodiment is also suitable, in addition to being output by printer Pr, for being displayed on an LCD or CRT display.

[0028] Next, the selection of the dither pattern P1, P2 based on the saturation-based data determined by the decision unit 4 will be described in reference to FIGS. 2A to 2C and 3A to 3C.

[0029] Generally, a dithering process is applied to each of small matrices in a single original image. In the following description, one of such matrices is focused.

[0030] The dither pattern P1 (FIG. 2B) is used to create a scattered distribution of black pixels inside the binary pixel matrix of the selected area of the image when the saturation value “c” is high since a high saturation value means the color appears vividly. When the saturation value “c” is low, on the other hand, the dither pattern P2 (FIG. 3B) is employed to create a dense distribution of black pixels inside the matrix of binary data, since a low saturation value “c” represents a dark color. In the embodiment shown in FIG. 1, the decision unit 4 and selecting unit 5 in combination decide between the two patterns based on the specified threshold level.

[0031] The dithering process conducted inside the dithering unit 2 when a vivid color area of original image is converted to a binary image is illustrated in FIGS. 2A to 2C and that when a dark color area is converted is illustrated in FIGS. 3A to 3C. The dot matrix Dm (a 4×4 pixel grid) shown in FIG. 2A or 3A which is comprised of multi-bit pixel data Pd for the lightness L of each pixel is compared to the dither pattern P1 (FIG. 2B) or P2 (FIG. 3B). The dither pattern is a matrix of threshold values to determine whether each of the pixels of the original color matrix Dm should be converted to black or white in the binary conversion. The dither pattern value (or threshold value) is compared with the lightness value L of the original color data Pd, and if the former is smaller than the latter, that pixel Pd is turned to 1 (Pdb or black). Otherwise, the pixel is turned to 0 (Pdw or white). FIGS. 2C and 3C illustrate the resultant matrices Dm1, Dm2 composed of binary data Pdb and Pdw.

[0032] The dither patterns P1 and P2 are, of course, not limited to the patterns shown in FIGS. 2B and 3B. For example, how the threshold values are dispersed or distributed in the dither pattern P1/P2 may not be changed, but the threshold values themselves may be changed; for instance, if the matrix Dm of the original color image includes pixel data Pd with very high saturation values “c”, the threshold values in the dither pattern P1 are increased respectively. By raising each threshold value in the dither pattern P1, the resulting binary image Dm1 includes more white pixels Pdw.

[0033] The present embodiment may be modified as follows: the selection of dither patterns P1, P2 based on the saturation value “c” as described above is executed only when the lightness level (total value or average) L of two neighboring areas of the original image to be converted to binary format are equal. Such an embodiment might be implemented because the eye particularly perceives the difference when the two continuous areas of the original image have the same lightness L but different chromaticities (saturations “c”). As described earlier, if these areas are converted to binary images based on only the lightness, the acquired binary images are the same black-and-white images. These looks unnatural. Therefore, the saturation “c” should be used in the binary conversion to output a natural looking binary image.

[0034] For instance, when two neighboring matrices Dm in the original color image have the equal total lightness value, the saturation “c” is considered in selecting the dither pattern P1 or P2. In this modification, only one dither pattern is provided from the beginning of the dithering process if the total lightness of a matrix Dm is not equal to that of an adjacent matrix Dm. In order to have the total lightness of the matrix Dm, the RGB-Lab converter 1 calculates the sum of the lightness values L for the pixel data Pd of each of the matrices Dm. When the total lightness level of the neighboring matrices are equal, the saturation calculating unit 3 selects one of the dither patterns P1, P2 based upon the calculated saturation value “c”. It should be noted that the neighboring matrices Dm are not limited to two matrices located in the scanning direction. Since there are eight matrices (i.e., the one above, the one below, the right one, the left one . . . ) around a certain matrix Dm, arbitrary one or more adjacent matrices among these eight matrices may be considered as its neighboring matrices.

[0035] Considering the lightness values L of all the peripheral matrices, however, requires temporary memory for storage of these data, so comparing the lightness values L of the left and right matrices or just one of these matrices is more practical. The 4×4 matrix Dm would only require a memory to store 4 lines (scanning lines) of data.

[0036] Further, in order to further simplify the selection of either of the dither patterns P1, P2, the neighboring areas are selected from the same scanning line. For example, a total lightness of predetermined number of dots (e.g., four dots) on a particular scanning line is compared with that of next four dots on the same scanning line to determine whether or not the dither pattern selection based on the saturation “c” is needed.

[0037] The embodiments described above utilize only the saturation as the dither pattern determining parameter, but hue or a combination of hue and saturation can be used as well.