JP2012054789A - Image processor, image processing method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the compression rate of image data.SOLUTION: The present invention relates to a method to improve the compression rate based on the similarity to a first color of three adjoining blocks, in a method of dividing image data into blocks of 2×2 pixels and converting into a pattern flag indicating a color arrangement pattern of each block and color data based on the color (pixel value) in each block. Furthermore, the number of bits of the pattern flag is not increased when the method of improving the compression rate based on the similarity to a previous adjacent block, and the method of improving the compression rate based on the similarity to the first color of the three adjoining blocks are used in combination.

Description

  The present invention relates to a technique for compressing an image in predetermined block units and a technique for performing image processing and restoration processing on an image compressed in block units.

  Conventionally, there is a high demand for high-resolution color images, and digital multifunction peripherals are increasingly handling images with a resolution of 1200 dpi or higher in order to meet the demand for higher image quality. As the resolution of an image increases, there is a problem that the number of pixels that require image processing increases dramatically and the processing load increases. Image processing devices such as digital cameras and facsimile machines, not limited to digital multifunction devices, compress color image data in order to save memory and hard disk space and shorten the time required for writing to them. Has been realized.

  As a color still image compression method, a JPEG method using discrete cosine transform or a method using Wavelet transform is often used. In this type of encoding method, generally, an image is encoded into a predetermined block (for example, 8 × 8 or 16 × 16 pixel unit), and high compression efficiency is achieved by performing discrete cosine transform, quantization, and entropy encoding. Have achieved. Since this type of encoding method is a variable-length encoding method, the code amount changes for each image to be encoded.

  When the above-described image compression is used, in order to refer to the pixel data and convert the pixel data, it is necessary to decode the compressed data. In other words, image processing cannot be performed with compressed data as it is, and decoding processing is necessarily required, and it is necessary to perform processing on a pixel basis for all pixels of high resolution data, which increases processing time. Invite.

  As a technique for performing compression processing without encoding pixel data, a known run-length compression method for storing pixel data and its continuous number or an edge is detected for each block, and two colors of the edge are stored. Thus, a technique for compressing is disclosed (for example, Patent Document 1).

JP 2008-271046 A

  In Patent Document 1, the inside of a block is made into two colors, and shape information relating to the arrangement of the two colors and color information of the two colors are stored. However, in order to achieve higher image quality and higher processing speed, the present applicant has made the following proposal in Japanese Patent Application No. 2009-212444.

  In the proposal, first, the image data is divided into blocks, and the color data of each pixel in the block is compared, so that the arrangement pattern information of the color data included in the block of interest and the number of colors included in the block of interest. Output color data information. In the output color data information, the first color data information corresponding to the pixel at a predetermined position in each block, the other color data information, and the arrangement pattern information are collected in different memory areas. It is characterized by storing. This realizes high image quality without reducing the image quality of blocks with more than two colors, and places the thinned image data at specific positions in a continuous memory area so that low-resolution images can be handled without decoding. By doing so, we are trying to achieve faster processing.

  That is, in the above proposal, an example is shown in which the compression processing is performed by comparing the color data included in the processing target block of 2 × 2 pixels (here, only by the comparison of the color data in the block described above). The compression processing will be referred to as “compression method closed with a block of 2 × 2 pixels”). In the above proposal, in addition to the compression method closed by the 2 × 2 pixel block described above, a process focusing on the similarity of the previous block to be processed is proposed. That is, in the above proposal, if the current processing target block has the same color arrangement as the previous processing target block, a flag indicating that it is the same as the previous block is stored as arrangement pattern information, and color data is stored. It is also considered that the compression rate is further improved by not doing so.

  In the present invention, the compression ratio is improved by applying another process based on the similarity with another adjacent block, which is different from the process based on the similarity with the previous processing target block described above. In addition, when the other process is combined with the process focusing on the similarity with the previous processing target block, the number of bits of the arrangement pattern information is increased so that it can be determined which process is applied. It is conceivable that a flag indicating that the process is applied is stored as arrangement pattern information. However, when the number of bits of the arrangement pattern information is increased, the data amount is increased correspondingly, and the compression rate may be reduced.

  Therefore, in addition to the compression method closed by the block of 2 × 2 pixels, when adding two or more other processes, in order to compress the arrangement pattern information with 4 bits, the remaining pattern before starting the compression process The processing to be assigned in one way must be determined in advance. However, the processing that can be expected to improve the compression rate differs depending on the image data to be compressed, and it is not known until the image processing is compressed which processing is effective. For this reason, it is necessary to determine which process is suitable for the image data to be compressed before starting the compression process, and to assign the process to the remaining one.

  An image processing apparatus according to the present invention divides image data into blocks each having a size of 2 × 2 pixels, a dividing unit that sequentially processes each of the divided blocks, and a processing target block divided by the dividing unit The first determination means for determining whether the arrangement of the color data in the block is the same as the arrangement of the color data in the previous block to be processed and the first determination means determine that the arrangement is not the same. If the color data of each pixel in the block to be processed is compared, the arrangement of the color data included in the block to be processed is selected from a plurality of predefined pattern flags indicating the arrangement pattern of the color data. A specifying means for specifying a pattern flag indicating a pattern, a color of a pixel at a predefined position in the processing target block is extracted as color data of a first color, and the block Extraction means for extracting color data other than the color data included in the first color, first count means for counting the number of appearances of each of the pattern flags specified by the specifying means, and two or more by the specifying means Regarding the processing target block identified as a pattern flag indicating that color data is arranged, the color data of three pixels other than the pixel at a predefined position in the processing target block are the three pixels. Based on the result of counting by the second counting means and the first counting means based on the result of counting by the first counting means, respectively Replace the pattern flag with the smallest amount of data increase caused by replacing the pattern from the multiple defined pattern flags. Selection means for selecting as a target pattern flag, and output means for outputting at least one of the pattern flag or the color data, and the output means is determined to be the same by the first determination means. For the processing target block, a pattern flag indicating that the color arrangement in the processing target block is the same as the color arrangement in the previous processing target block is output, and the first For the processing target block that is a pattern flag indicating that the pattern flag determined by the determining unit is not the same and the pattern flag specified by the specifying unit is a block composed of one color, the specifying unit specifies Output the pattern flag and the color data of the first color in the processing target block extracted by the extraction unit, and the first determination unit determines that they are not the same. In addition, for the processing target block determined to match by the second determination unit, the pattern flag selected by the selection unit and the first in the processing target block extracted by the extraction unit Color data of a color is output, and the first determination unit determines that they are not the same, and the second determination unit determines that they do not match, and the pattern flag specified by the specifying unit is the selection unit For the processing target block that is the replacement target pattern flag selected in step 4, the processing target block is a block in which four color data are arranged, with the processing target block being a block composed of four colors. And the color data of each of the four pixels in the processing target block are output, and the first determination unit determines that they are not the same, For the processing target block that is determined not to match by the second determination unit and the pattern flag specified by the specifying unit is not the replacement target pattern flag selected by the selection unit, the specifying unit A pattern flag indicating an arrangement pattern of colors included in the processing target block identified in Step 1, color data of the first color in the processing target block extracted by the extraction unit, and the processing target block The color data other than the color data of the first color are output.

  Further, the image processing apparatus of the present invention divides image data into blocks each having a size of 2 × 2 pixels, dividing means for sequentially processing each divided block, and each pixel in the processing target block Specifying means for identifying a pattern flag indicating the arrangement pattern of the color data included in the processing target block from a plurality of predefined pattern flags indicating the arrangement pattern of the color data by comparing the color data of Extracting means for extracting a color of a pixel at a predetermined position in the processing target block as color data of a first color, and further extracting color data other than the color data of the first color included in the block And the color data of three pixels other than the pixel at the predefined position in the processing target block are respectively in three separate blocks adjacent to the three pixels. Determination means for determining whether or not the color data of the pixel at a predefined position in each of the two match, and for the processing target block determined to match by the determination means, three adjacent to the three pixels Outputting a pattern flag indicating that it matches the color data of a pixel at a predefined position in each of the other blocks, and the color data of the first color in the processing target block extracted by the extraction means, For a processing target block determined not to match by the determination unit, a pattern flag indicating an arrangement pattern of color data included in the processing target block specified by the specifying unit, and the extraction unit extracts the pattern flag. Output means for outputting color data of the first color and color data other than the color data of the first color.

  According to the present invention, it is possible to improve the compression rate of image data without increasing the number of bits of arrangement pattern information.

1 is a block diagram showing an example of the configuration of an image processing apparatus according to the present invention. The figure which shows the structural example of the controller which concerns on this invention. The figure which shows the combination of the arrangement pattern of the color of 2x2 pixel block which concerns on this invention. The figure which shows the relationship between the arrangement | positioning pattern which concerns on this invention, and a pattern flag. The flowchart which shows the number of pixels and pattern flag specific processing which concern on this invention. The figure which shows the data amount of the output data corresponding to the arrangement | positioning pattern which concerns on this invention. The figure which shows the example of writing to the memory space of the image data based on this invention. The figure which shows the pattern flag showing a coincidence with the previous block which concerns on this invention. The flowchart which shows the compression process showing a coincidence with the previous block which concerns on this invention. The figure which shows the example of writing to the memory space of the image data based on this invention. The figure which shows the positional relationship of the attention block and surrounding block which concerns on this invention. The figure which shows the pattern flag showing a match with the block of 3 directions which concerns on this invention. The flowchart of the compression process showing a match with the block of 3 directions which concerns on this invention. The figure which shows the example of the pattern exchange used by the compression process which concerns on this invention. The figure which shows the data increase of the pattern by replacement of the pattern which concerns on this invention. The figure which shows the reduction amount and increase amount of data by the replacement of the pattern which concerns on this invention. The flowchart which shows the compression process which replaces the pattern which concerns on this invention. The flowchart which shows the compression process and count process which concern on this invention. The flowchart which shows the pattern replacement process which concerns on this invention. The flowchart which shows the compression process after the pattern replacement based on this invention. The figure which shows the example of writing to the memory space of the image data based on this invention.

[Hardware configuration]
DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the invention will be described with reference to the drawings. In this embodiment, a digital multi-function peripheral (MFP) having a plurality of functions such as scanning, printing, and copying will be described as an example of the image processing apparatus.

  As shown in FIG. 1, the controller 101 is connected to a scanner 102 as an image input device and a printer 103 as an image output device. Further, the controller 101 is connected to a network 104 such as a LAN or a public line (WAN) to input / output image information and device information and develop an image of PDL data.

  The CPU 105 is a processor that controls the entire MFP in accordance with a program stored in a storage unit 107 described later. The memory 106 is a system work memory for operating the CPU 105, and is also an image memory for temporarily storing image data. The storage unit 107 is a hard disk drive, and stores system software, programs, image data, and the like.

  Next, detailed processing of each unit of the controller 101 will be described with reference to a configuration example of the controller 101 shown in FIG. First, a case where image data scanned by the scanner 102 is read will be described. The scanner image processing unit 201 receives RGB (red, green, blue) three-color image data read by the scanner 102, performs image processing such as shading processing and filter processing on the image data, and compresses the data. Performs image compression processing. The compressed data (compressed data) is stored in the memory 106 via the image memory bus by a DMAC (Direct Memory Access Controller) 203.

  Next, when printing scanned image data, the DMAC 211 transfers the compressed data stored in the memory 106 to the color processing unit 212 via the image memory bus. Then, the color processing unit 212 converts into a CMYK (cyan, magenta, yellow, black) color space. Thereafter, after the color processing unit 212 further performs color processing such as density adjustment and printer gamma correction on each value of CMYK, the data after the color processing is stored again in the memory 106 by the DMAC 211 via the image memory bus. To do. Thereafter, in order to perform image processing for printing, the DMAC 221 reads the compressed data stored in the memory 106 via the image memory bus, and the expansion unit 222 expands the raster image data. The raster CMYK image data is input to the print image processing unit 223, performs area gradation processing by a dither method or an error diffusion method, and outputs the result to the printer 103.

  When the scanned image data is transmitted to the network, the DMAC 211 transfers the compressed data stored in the memory 106 to the color processing unit 212 via the image memory bus. After the color processing unit 212 performs display gamma adjustment, paper background color adjustment, and the like, the color processing unit 212 performs conversion to a YCbCr (luminance, BLUE color difference, RED color difference) color space. Then, the DMAC 211 stores the data processed by the color processing unit 212 in the memory 106 again via the image memory bus. Thereafter, in order to perform image processing for transmission, the DMAC 231 transfers the compressed data stored in the memory 106 to the expansion unit 232 via the image memory bus. Then, the expansion unit 232 expands the compressed data into raster image data. After that, if the transmission processing unit 233 performs color image transmission on the raster YCbCr image data, JPEG compression processing is performed. If the monochrome binary image transmission is performed, binarization is performed on the Y data, and JBIG compression or the like is performed. And output to the network 104.

  When saving scanned image data, the DMAC 241 transfers the compressed data stored in the memory 106 to the disk spool high compression / decompression unit 242 via the image memory bus. In the disk spool high compression / decompression unit 242, the writing speed of the HDD is slower than that of the memory. Thereafter, the compressed data is stored in the storage unit 107 via the disk access controller 243. Further, when the compressed data stored in the storage unit 107 is transferred again to the memory 106, the above-described processing may be performed in reverse.

  Here, a case where PDL data sent from another device connected via the network 104 shown in FIG. Although the PDL interpretation unit is not shown in FIG. 2, the CPU 105 functioning as the PDL interpretation unit interprets the PDL data and outputs the display list as a result to the memory 106. Thereafter, the rendering unit 251 renders the display list stored in the memory 106 into raster RGB image data, and the compression unit 252 performs image compression processing. Then, the DMAC 253 stores the compressed data in the memory 106 via the image memory bus.

  It should be noted that the process of printing, transmitting to a network, and storing the PDL data can be realized by performing the same process as in the case of scanned image data.

[Placement pattern]
Before explaining the compression processing here, a combination of colors in a block having a predetermined number of pixels (4 pixels of 2 × 2 in this embodiment) is considered. Since the number of pixels is four, the maximum number of colors is four, and there are only combinations of one to four colors in the block. FIG. 3 shows the number of combinations of the four color arrangement patterns.

  First, when the number of colors constituting the block is one, four pixels are composed of the same color, so there is one combination (arrangement pattern). Next, consider a case where the number of colors constituting the block is two (each color is referred to as a first color and a second color). As shown in FIG. 3, when the two colors are laid out in four pixels, the pixel at a predetermined position (in this embodiment, the lower right pixel) is considered as the first color, except for the lower right pixel. Since the first color or the second color enters the remaining three pixels, a total of seven combinations are possible except for the case of four pixels having the same color.

  Next, consider the case where the number of colors constituting the block is three (first to third colors). The number in the case where three colors are laid out in four pixels can be paraphrased as the number in which only one of the three colors is used twice, so the number in which two of the four pixels have the same color. Just ask. That is, the number in the case of three colors is a combination of taking two pixels from four pixels, for a total of six. In addition, when the number of colors constituting the block is four colors (first to fourth colors), there is only one pattern.

  Therefore, if the number of colors constituting the block is the total of all the cases of 1 to 4 colors, 15 patterns can be considered in total. Considering that a flag (identifier) is added to identify all these patterns, the flag data amount requires 4 bits (flags 0 to E). The relationship between these 15 patterns and flags is defined as shown in FIG. 4, and this flag is hereinafter referred to as a “pattern flag”.

[Compression processing]
As described above, processing performed by the compression units 202 and 252 will be described with reference to FIG. 5 based on possible combinations of 2 × 2 pixels. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like. As an input, for example, each of RGB has 256 gradations of 8 bits, and the data is described as an image of 24 bits per pixel in a dot-sequential manner of 8-bit data.

  First, a 2 × 2 pixel block is input, and 24-bit comparison is performed for all combinations of two pixels in the block (S501). As a result of the comparison, “1” is output when all bits match, and “0” is output when they do not match.

  Here, the coordinates are 1, 2, 3, 4 in the order of the positions of the lower right, the upper left, the upper right, and the lower left pixels in 2 × 2 pixels (401 shown in FIG. 4). Since there are six coordinates (402 shown in FIG. 4) of the combination of two pixels, that is, 2-3, 2-4, 2-1, 3-4, 1-3, and 1-4 in total, compare 6 times. The result is output 6 bits. As shown in the comparison result shown in FIG. 4, if all pixels have the same color, all comparison results output 1, and conversely, if all four pixels have disjoint pixel values, all comparison results are 0. Is output.

  In this example, since there are 15 patterns that can appear from the color match in 4 pixels, as shown in FIG. 4, a 4-bit pattern flag is specified according to the 6-bit comparison result (S502). Next, the number of colors appearing within the four pixels and the color data are extracted (S503). As shown in FIG. 4, since each pattern indicating the arrangement of each color in the block is associated with the 4-bit pattern flag (or 6-bit comparison result), the number of colors in each block Color data can be specified.

  In FIG. 4 of the present embodiment, it is defined that the color (pixel value) of the lower right pixel is the first color (first color data) in all patterns. When the pattern flag is 0, the number of colors is 1 and the color of the lower right pixel (pixel value) is extracted as the first color. When the pattern flag is 1 to 7, the number of colors is 2 and the color (pixel value) of the lower right pixel is extracted as the first color, and further, the second color defined according to each pattern flag The color (pixel value) of the pixel at the position where (second color data) exists is extracted. For example, when the pattern flag is 1, the color of the upper right pixel is extracted as the second color. When the pattern flag is 8 to D, the number of colors is 3, and the color (pixel value) of the lower right pixel is extracted as the first color, and further, the second color defined according to each pattern flag ( The color (pixel value) of the pixel at the position where the second color data) and the third color (third color data) exist is extracted. For example, when the pattern flag is 8, the color (pixel value) of the upper left pixel is extracted as the second color, and the color (pixel value) of the upper right pixel is extracted as the third color. When the pattern flag is E, the lower right pixel is the first color, the upper left pixel is the second color, the upper right pixel is the third color, and the lower left pixel is the fourth color. Extract as

  That is, based on the pattern flag (or comparison result), the number of colors of the block is specified (S504, S506, S508), and the corresponding pattern flag and color data are output (S505, S507, S509, S510). . The first color may be referred to as a representative color, and the second to fourth colors may be referred to as other colors.

  Data output by this processing will be described with reference to FIG. As shown in FIG. 6, for example, when the pattern flag is 0 (that is, the four pixels are configured with one color) (YES in S504), since the second and subsequent colors do not exist, the 4th bit of the pattern flag and the first color Are output (24-bit color data) (S505). If the pattern flag is 1 to 7 (that is, the four pixels are composed of two colors) (YES in S506), the coordinates of the pixel where the second color exists are calculated from the pattern flag, and the four bits of the pattern flag And pixel values for two colors (48-bit color data) are output (S507). If the pattern flag is 8 to D (that is, three colors) (YES in S508), 4 bits of the pattern flag and pixel values for three colors (72-bit color data) are output (S510). If the pattern flag is E (that is, 4 colors) (NO in S508), 4 bits of the pattern flag and pixel values for 4 colors (96-bit color data) are output (S511), and the process ends. (S512). In other words, the color data output from each block is the color data that did not appear when the coordinates in the block (1, 2, 3, 4 in the order from lower right to upper left, upper right, lower left) were scanned. This corresponds to output in order from the first color.

  In this way, the input data of 4 colors (96 bits) in a 2 × 2 pixel block is output as a 4-bit pattern flag and pixel values corresponding to the number of colors existing therein, thereby relatively simple processing. This makes it possible to reduce the amount of output data. In particular, in the case of a raster image having a large number of blocks having the same color pixel in a 2 × 2 pixel block (that is, a block having a small number of colors in each block), the compression rate of the output data amount is also increased. Also, by referring to the pattern flag, the number of colors in the block can be specified. By performing such processing on all the image blocks, the entire image can be compressed.

  Next, a process (memory storing process) in which the DMACs 203 and 253 write the pattern flag and color data obtained in this way into the memory 106 is performed. At this time, the DMAC changes the writing position of the pattern flag, the first color data, and the second to fourth color data (other than the first color data). The DMAC designates three addresses: a memory start address for writing a pattern flag, a memory start address for writing first color data, and a memory start address for writing data for the second and subsequent colors. That is, the pattern flags of each block are stored together in a pattern flag storage unit (memory area for storing pattern flags) on the memory. Further, the color data of the first color (first color) of each block is stored together in a first color storage unit (memory area for storing the first color data of each block) on the memory. Become. Further, the color data of the second to fourth colors of each block are collectively stored in the second, third, and fourth color storage units (memory areas for storing the color data of the second to fourth colors of each block) on the memory. Will be.

  FIG. 7 is a diagram illustrating an example of writing image data in the memory space by the DMAC. When an 8-bit RGB image having a size of M × N pixels is input to the compression unit, the data size of the pattern flag storage unit that stores pattern flag data is (M × N × 4/8) bytes. It becomes. Further, the data size of the first color storage unit in which the first color data is stored is (M × N × 24/8) bytes. The data sizes of the second, third, and fourth color storage units that store the second, third, and fourth color data differ depending on the raster image to be processed. This is because the number of blocks in which the second, third, and fourth colors are present differs depending on the image.

  Here, in the memory area (first color storage unit and second, third, and fourth color storage units) after the first color write start address, pixel color data is stored in the same number of bits. That is, unlike compressed data such as JPEG, the color (pixel value) of each pixel can be specified without decoding the compressed data and returning it to a raster image. Therefore, when performing color processing that is completed with one pixel input and one pixel output, such as color conversion using LUT, gamma correction processing, color space conversion processing using matrix operation, etc., there is no need to return to the original raster image. It becomes possible to directly process the data stored as shown in FIG. In the case where image processing in units of pixels is performed by the color processing unit 212 shown in FIG. 2, pixel data after the first color writing start address on the memory 106 is read via the DMAC 211, and after processing in units of pixels is completed, Write back to the memory 106. At this time, if the number of bits of the pixel does not change due to some pixel unit processing, the memory 106 can be saved by overwriting the same location on the memory 106.

  In this way, by directly using compressed data, transfer efficiency on the memory bus is improved, and processing is performed for a smaller number of color data than the number of pixels of the original raster image, enabling high-speed processing. become.

[Data storage format]
As shown in FIG. 7, the image data is divided into a pattern flag, a first color, and other colors, and is discretely stored in the memory. Therefore, in the first color storage unit, pixel values (color data) obtained by dividing the raster image into blocks of 2 × 2 pixels and sampling the pixel at the lower right coordinate of each block are continuously stored in the memory. Will exist.

  The MFP also has functions such as preview display of accumulated PDL image data and scanned image data, and network transmission described above. For example, even if the print resolution is 600 dpi, the resolution up to that is not usually required at the time of preview or transmission, and 300 dpi or less is often sufficient. As described above, when it is necessary to obtain reduced data, a raster image having a half size can be easily obtained by extracting only the color data of the first color stored in the first color storage unit. Can do.

  For example, a description will be given of the time of reduced transmission when a 600 dpi raster image is converted into data as shown in FIG. When 300 dpi is specified as the resolution of the image to be transmitted, the data stored in the first color storage unit may be extracted as it is and transmitted. If a resolution greater than 300 dpi sampled in the first color storage unit, such as 400 dpi, is designated, the following processing is performed. That is, based on the data stored in the pattern flag storage unit, the first color storage unit, and the second, third, and fourth color storage units, the compressed data is expanded once and scaled using a known scaling process. Then, the scaled image is transmitted. If transmission resolution less than 300 dpi is specified, scaling processing is performed to the specified resolution using only the data in the first color storage unit. In this way, data can be read while switching according to the required image size. Hereinafter, the above-described compression processing is referred to as normal compression processing.

[Deployment processing]
Next, the expansion units 222 and 232 that are paired with the compression units 202 and 252 will be described. The developing units 222 and 232 perform processing for returning the pattern flag and color data as described above to raster image data. The compressed data pattern flag write start address, first color write start address, and second, third and fourth color write start addresses of the compressed data arranged on the memory 106 as shown in FIG. specify. The DMACs 221 and 231 read data from three addresses and transfer them to the expansion units 222 and 232.

  The expansion units 222 and 232 first interpret the 4-bit pattern flag and calculate the number of colors in the block. In addition to the first color data, the second, third, and fourth color data are read in accordance with the number of colors, and the first color and the first color are determined according to the color data arrangement pattern (FIG. 4) defined in advance for each pattern flag. Rearrange 2 to 4 color data. As a result, the 2 × 2 pixel block is expanded and decoded.

  In addition, when the image size is halved by the developing units 222 and 232, the pattern flag and the second, third, and fourth color data are not required as described above, and therefore the first color is included in the DMACs 221 and 231. Specify only the write start address. As a result, only the first color data is read from the memory 106 to form an image. By processing in this way, it becomes possible to save the bandwidth of the memory bus.

  In the present embodiment, the pattern indicating the arrangement of the color data and the pattern flag are associated as shown in FIG. 4, but the present invention is not limited to this. The first color is not limited to the lower right pixel. For example, the pattern indicating the arrangement of the color data is associated with the pattern flag so that the pixel value of the upper left pixel of the 2 × 2 pixel block is the first color. It may be defined in advance. In the first embodiment, the block size of 2 × 2 pixels has been described. However, the present invention is not limited to this. In the description of compression, 8 bits for each of RGB has been described as an example of image data, but data having a CMYK color space, GRAY scale data, or a pixel value other than 8 bits may be used.

[Compression processing by repeat processing]
Next, a process for further improving the compression rate by adding to the normal compression process will be described. An image that is expected to be highly compressed by a normal compression process is an image having many regions having the same color of adjacent pixel values in the image. Focusing on such an image area, not only the inside of a 2 × 2 pixel block but also an adjacent 2 × 2 pixel block often has the same color.

  Therefore, in addition to the compression method closed within the block described above, a compression method that focuses on the similarity between adjacent blocks will be described. In FIG. 4, the number of pattern flags that can be taken by a 2 × 2 block is expressed by 4 bits as 15 patterns. Since a 4-bit pattern flag can represent up to 16 patterns, there is still room for one pattern to be used.

  Therefore, if the pixel value of the block immediately before the currently processed block is stored and it can be determined that all pixels match the current block, the pixel value (color data) of that block is not output and the previous block is not output. Only flag the same as the block. By assigning the remaining 4 bits to this pattern flag, the data can be further compressed without changing the number of bits of the pattern flag.

  FIG. 8 is a diagram showing the number of bits when a flag (referred to as a repeat flag) indicating that the pattern is the same as the previous block is added as a new pattern flag. In this example, “F” representing a match with the previous block (no new pixel value output) is added as a pattern flag. As shown in FIG. 8, the block of the pattern flag F can be expressed by 4 bits of the pattern flag. That is, since the block having the pattern flag F can be expanded with reference to the pixel value (color data) and the pattern flag of the previous block, it is not necessary to store the pixel value (color data). The compression rate is improved. The above process is referred to as a repeat process below.

[Repeat processing flow]
Here, the repeat process will be described in detail with reference to FIG. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like. First, when processing is started, raster image data is input, and a block of 2 × 2 pixels is cut out therefrom (S901). Next, if the target block (processing target block) is the first block, there is no block that has been processed and stored before (YES in S902), so the compression process is performed in the same way as the normal compression process described above. (S906). On the other hand, when it is determined in S902 that the target block is not the first block (NO in S902), the pixel value of the target block is compared with the block that has been processed one time before and stored (S903). Thereby, the first determination is performed. When the pixel value of the block stored here and the pixel value of the block of interest match (YES in S904), it is determined that the same data as the previous block stored immediately before can be used. Then, the pattern flag F indicating that it is the same as the previous block is specified (S905), and output of pixel values (color data) is omitted. That is, when it is determined that the pixel value of the block to be processed matches the block to be processed before that, the pattern flag F (repeat flag) indicating the repetition of the previous block is determined for the block to be processed. ) Is output. Thereafter, the process proceeds to S909.

  If the stored block does not match the pixel value (NO in S904), the normal compression process described above is performed (S906). Thereafter, the data is expanded (S907), and the expanded 2 × 2 pixel block is stored and updated as data of the previous block (S908). In this embodiment, after the compression process is performed in S906, the data of the previous block is updated in S907 to S908. However, the present invention is not limited to this. For example, before performing the compression processing in S906, the data of the previous block may be updated using the pixel value of the processing target block.

  Next, for a block whose pixel value does not match the stored block, the pattern flag and pixel value (color data) as a result of performing the normal compression processing in S906 are output, and the stored block and pixel Only the pattern flag F is output for the blocks whose values match (S909). This process is repeated until the last block (YES in S910), and the entire image compression process is terminated.

  Since the number of bits of the pattern flag has not increased to 4 bits, it is possible to improve the compression rate with certainty compared to the normal compression processing described above, but the size of the color data of the first color is Like the three- and four-color data, it becomes indefinite (differs depending on the image).

  Accordingly, as shown in FIG. 10, the second, third, and fourth color data cannot be stored in the memory space continuous with the first color data, and the first color, the second, third, and fourth color data are stored. Must be given discretely. In the above-described color processing unit 212 shown in FIG. 2, pixel data after the first color writing head address on the memory 106 is read via the DMAC 211, processed in units of pixels, and then written back onto the memory 106. However, since the first color and the second, third, and fourth colors are not continuous in the repeat process, the color processing unit includes the second, third, and fourth write start addresses in addition to the first color address. It is necessary to specify in the DMAC 211 of 212.

  Here, the block one block to the left is defined as the previous block, and matching and mismatching with the block are held as states, but this is not restrictive. For example, it is possible to define a match with a wider area by expanding the match with the upper block or the number of bits of the pattern flag. When the number of bits of the pattern flag is expanded, it is set within a range where the data amount can be reduced rather than holding the color data.

[Expanding process for compressed data with repeat flag]
Next, a decompression process paired with the compression process will be described. The expansion process is a process for returning from the pattern flag and pixel data to the raster image data as described above. Designate the DMACs 221 and 231 with three addresses of the compressed data pattern flag write start address, the first color write start address, and the second, third and fourth color write start addresses arranged on the memory 106 as shown in FIG. To do. When it is identified as the same as the previous block with reference to the pattern flag, the output of the block of interest uses the output of the previous block itself, and other flags can be developed by the method described above. At this time, since one block may be used in the next block, it is necessary to buffer it.

  However, when the image size is halved by the expansion process, only the first color data is read from the memory 106 as described above, and the matching data with the adjacent block can be expanded only by composing the image. Can not. For this reason, in the development processing of the present embodiment, a 1/2 size image is restored based on the first color data while referring to the pattern flag. The second, third and fourth color data need not be used. Specifically, when the pattern flag is referred to and identified as the same as the previous block, the previous color data in the first color data can be repeatedly output to be developed. As a result, it is possible to save the bandwidth of the memory bus.

[Compression based on similarity]
Next, compression processing that focuses on the similarity between adjacent blocks, which is different from repeat processing, will be described. Similar to the repeat process, this compression process is a process that further improves the compression ratio by expanding the pixel value of the currently processed block with reference to the pixel values of the surrounding blocks.

  FIG. 11 is a diagram showing the positional relationship between pixels of a block currently being processed and surrounding blocks. As shown in FIG. 11, the pixel at coordinate 2 of the block currently being processed is located at the lower right of the pixel at coordinate 1 of the upper left block, and the pixel at coordinate 3 is the pixel at coordinate 1 of the upper block. The pixel at coordinate 4 is arranged adjacent to the right of the pixel at coordinate 1 in the left block. This is the same as the arrangement indicated by 401 in FIG. For the pixel at coordinate 2 located at the upper left of the pixel at coordinate 1 (first color pixel) in the processing target block, the pixel at first coordinate (pixel of the first color) in another block adjacent in the upper left direction. ) To determine whether they match. Similarly, with respect to the pixel of coordinate 3 positioned above the pixel of coordinate 1 (first color pixel) in the processing target block, the pixel of coordinate 1 (first color of the first color) in another block adjacent in the upward direction. By comparing with (pixel). Furthermore, with respect to the pixel of coordinate 4 located to the left of the pixel of coordinate 1 (first color pixel) in the processing target block, the pixel of coordinate 1 (first color pixel) in another block adjacent in the left direction. ) To determine whether they match. As described above, when the pixel values of the coordinates 1 of the three adjacent blocks coincide with the coordinates 2 to 4 in the block currently being processed, the following processing is performed. The pixel value other than the coordinate 1 (first color pixel) of the block currently being processed is not output, but only the first color pixel value and the flag that matches the first color pixel in the three directions are output. By assigning the remaining one pattern to this pattern flag, it is possible to further compress the data without changing the number of bits of the pattern flag.

  FIG. 12 is a diagram showing the number of bits when a flag indicating that pixels in three directions match (referred to as a three-way match flag) is added as a new pattern flag when repeat processing is not performed. In this example, “F” is added as a pattern flag, which represents a match with pixels in the surrounding three directions (pixel values other than the first color are not output). As shown in FIG. 12, a block whose pattern flag is F can be expressed by 4 bits of the pattern flag and 24 bits of the pixel value of the first color. That is, since the block having the pattern flag F can be expanded with reference to the pixel values of the first color and the pattern flags of the three surrounding blocks, the pixel values of the second to fourth colors are stored. This is unnecessary and the compression rate is improved. In this case, it is possible to reduce to the same data amount as when the pattern flag is 0 (the color data in the block is only one color). That is, the data amount of the block compressed based on the similarity is 28 bits that is the sum of the pattern flag (4 bits) and the color data for one color (24 bits). The above process is hereinafter referred to as a three-way matching process.

[Compression processing flow based on similarity]
Here, the description of the three-way matching process will be described in detail with reference to FIG. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like. First, when the process is started, raster image data is input, and a 2 × 2 pixel block is cut out therefrom (S1301). Next, if it is the uppermost or leftmost block of the image data (YES in S1302), there is a portion where no block exists around it, so the compression process is performed in the same way as the normal compression process described above (S1306). On the other hand, if the block is not the upper end or the left end block (NO in S1302), the pixel values of the first color of the blocks arranged at the upper left, the upper, and the left, and the respective colors other than the first color currently processed The pixel values are compared (S1303). If all pixel value comparison results of pixels other than the first color match (YES in S1304), the pattern flag F indicating three-way matching is specified, and output of pixel values other than the first color is omitted (S1306). ). That is, when it is determined that the pixel values other than the first color of the block to be processed match the pixel values of the first color of the surrounding blocks, a pattern flag F (three-way match flag) indicating three-way match is output. . This process is repeated until the last block (S1308), and the entire image compression process is terminated.

  Similar to the repeat processing, the number of bits of the pattern flag does not increase from 4 bits, so that the compression rate can be improved reliably as compared with the normal compression processing described above. The method for storing data in the memory space is the same as the method for storing data in the memory space for normal compression processing as shown in FIG.

[Expansion processing for compressed data having a three-way match flag]
Next, the decompression process paired with the compression process will be described. The expansion process is a process for returning from the pattern flag and pixel data to the raster image data as described above. The compressed data pattern flag write start address, first color write start address, second, third and fourth color write start addresses of the compressed data arranged on the memory 106 as shown in FIG. To do. When it is identified that three directions coincide with each other with reference to the pattern flag, the output of the block currently being processed is expanded by referring to the block arranged adjacently or diagonally above, and the other flags In this case, it can be expanded by the method described above. At this time, in order to refer to the pixel values of adjacent three-direction blocks, it is necessary to buffer one line above the block currently processed and the previous block.

  The repeat process and the three-direction matching process described above are processes that further improve the compression ratio by assigning these processes to the remaining one of the 4-bit arrangement pattern information as compared with the case of only the normal compression process. The compression ratio can be further improved by combining these processes rather than performing the repeat process and the three-way matching process alone. However, in order to perform processing by combining these, it is necessary to increase the number of bits of arrangement pattern information by increasing the number of patterns.

[Compression processing according to the present invention]
Here, a method for performing compression processing by combining repeat processing and three-way matching processing without increasing the number of bits of the arrangement pattern information according to the present invention will be described. In order to add the above-described two processes to the normal compression process without increasing the number of bits of the arrangement pattern information, the process already assigned to each pattern flag and the process to be newly added are switched. That's fine.

  FIG. 14A shows assignment of the initial pattern to the pattern flag. Here, pattern flags 0 to F (16 patterns) are assigned to 15 patterns in normal compression processing and one pattern for repeat processing. Next, FIG. 14B shows an example of replacing the pattern flag. In this example, the pattern assigned to the pattern flag D is replaced with the three-way matching process, and the pattern flag D is assigned to the three-way matching process. Next, a pattern flag E which is a pattern of four colors is newly assigned to the pattern assigned to the initial pattern flag D. That is, when a block corresponding to the initial pattern flag D is input, the block is handled as a four-color block (E). In the case of this pattern, a 4-bit pattern flag and a 4-bit pixel value of 96 bits are output. Therefore, even if the patterns are replaced, the input block that is a pattern corresponding to the initial pattern flag D outputs the pixel values for four pixels, so that the image quality is not deteriorated.

  However, when pattern replacement is performed, the pattern replaced in the initial pattern is treated as a four-color block (E) as described above. For this reason, when compressing with the initial pattern, pixel values (color data) that did not need to be output are also output, and the compression rate is reduced accordingly.

  FIG. 15 shows output data when the input data is compressed with the initial pattern and the pattern after replacement. FIG. 15 shows an example in which a two-color block is treated as a four-color block. Since a block treated as four colors must output pixel values for four pixels, pixel data (48 bits) for two pixels increases.

  Therefore, in order to improve the compression rate reliably, determine how much the pixel data will increase by replacing the pattern, and how much the compression rate will improve by adding the three-way matching process. It is necessary to change the pattern. That is, the higher one of the normal pattern and the pattern after replacement is applied to the block.

  FIG. 16 shows the count value of the number of appearances of each pattern when image data is processed, and the data reduction amount and data increase amount due to pattern replacement. Here, a method for determining a pattern to be replaced will be described with reference to FIG. FIG. 16 shows 3 types extracted from 16 types of combinations of normal compression processing and repeat processing. First, a two-color pattern a will be described as an example. It is assumed that the number of occurrences of the pattern a in the entire image is 30 times, of which the number of matches in the three directions is 10 times. By adding the three-direction matching process, ten of the blocks in the pattern a can reduce pixel data (24 bits) for one color. Therefore, in the case of the pattern a, the number of bytes can be reduced by matching in three directions (1 × 10 × 24/8 = 30). On the other hand, if the three-direction matching process is added by replacing the pattern a, the pattern a is handled as a block of four colors, so that it is necessary to output extra data for two colors as described above. That is, when the pattern a and the three-direction matching process are exchanged, (2 × (30−10) × 24/8 = 120) bytes of data increase. Similarly, when the patterns b and c are considered, the result is as shown in FIG.

  If the number of three-way matching times in the remaining patterns not shown is zero, the data reduced by the three-way matching process is a total of 150 bytes. If the total data reduction amount exceeds the data increase amount for each pattern, it can be determined that the compression rate is improved by replacing the three-way matching process and the pattern. By replacing the pattern with the smallest data increase amount among the patterns in which the total data reduction amount exceeds the data increase amount with the three-way match, the compression ratio can be expected to be improved. In FIG. 16, the pattern b is selected as a replacement target pattern, and the byte compression rate is improved for the entire image (150-30 = 120). If the total data reduction amount does not exceed the data increase amount of each pattern generated by the pattern replacement, the pattern is not replaced.

  When pattern replacement is performed, a pattern replacement flag indicating the replaced pattern flag and an additional process flag “1” indicating that the three-way matching process has been added are output and held. Since the pattern used for the compression can be determined from the pattern replacement flag and the additional processing flag, it is possible to restore the original image by performing a development process.

  In this way, the number of occurrences of each pattern is counted using a counter, and the pattern assigned to the pattern flag is replaced so as to improve the compression rate without increasing the number of bits of the arrangement pattern information. Processing can be added. Thereby, the first counting means is realized.

[Processing Flow According to the Present Invention]
Here, the processing method of the present invention will be described in detail with reference to FIG. In the present embodiment, a case will be described in which normal compression processing and repeat processing are represented by 4-bit arrangement pattern information and three-way matching processing is combined. Here, 15 patterns of normal compression processing and combinations of repeat processing are defined as initial patterns. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like.

  First, when the process is started, raster image data is input, and a 2 × 2 pixel block is cut out therefrom (S1701). Next, while performing compression processing on the block extracted in S1701, the number of appearances of each pattern flag is counted in parallel (S1702). This process will be described later with reference to FIG. The processing so far is repeated up to the last block of the image data (YES in S1703), the entire image is compressed, and stored in the storage unit. Here, it is determined from the compression result whether the target compression rate has been reached (S1704). The target compression rate here is a target compression rate for the input image data. This target compression rate may be defined in advance or specified by the user. As a result of the determination, if the target compression rate is achieved by the processing on the image data in S1702 (YES in S1704), the compression processing is terminated. On the other hand, if the target compression rate has not been achieved (NO in S1704), pattern replacement processing is performed to improve the compression rate (S1705). This process will be described later with reference to FIG. As a result of the pattern replacement process, if the compression rate improves (data amount decreases) (YES in S1706), the compression process is performed using the next replaced pattern (S1707), and the process ends. . This process will be described later with reference to FIG. If the compression rate does not improve as a result of the pattern replacement process (NO in S1706) (NO in S1706), compression is performed by an irreversible compression means (not shown) such as JPEG (S1708). Exit. If it is determined in S1704 that the target compression rate cannot be achieved (NO in S1704), in S1707, compression processing is performed using the assignment after replacing the pattern flag for the original image data. . In step S1708, the original image data is compressed using an irreversible compression method. Note that the image data to which the irreversible compression by the irreversible compression method is applied may be the original image data, or may be applied to the image data once compressed in S1702. Therefore, the compressed data compressed in S1702 and stored in the storage unit may be discarded or used for further compression or the like.

[Compression & Counting]
Here, the compression processing and count processing (S1702) in the above-described processing will be described in detail with reference to FIG. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like. Further, the counter value here is held in a storage unit such as the memory 106. First, when processing is started, it is first determined whether the block currently being processed is the first block (S1801). If it is the first block, there is no block that has been previously stored by applying compression processing (in S1801). YES), normal compression processing is performed (S1807). On the other hand, if it is not the first block (NO in S1801), the previous block is processed and the pixel value is compared with the stored block (S1802). If the pixel value matches the stored block (YES in S1803), a pattern flag (F) indicating the same as the previous block is created (S1804), and the pattern flag is output (S1805). . Thereafter, 1 is added to the counter corresponding to the repeat flag (S1806).

  If the stored block does not match the pixel value (NO in S1803), normal compression processing is performed (S1807). Thereafter, the data is expanded (S1808), and the expanded block is stored with the expanded 2 × 2 pixel block as the previous block data (S1809). Next, pixel data and a pattern flag are output (S1810). Then, 1 is added to the counter corresponding to the determined pattern flag (S1811).

  Next, it is determined how many times the identified pattern has been matched in three directions (S1812). If the determined pattern is one color (YES in S1812), there is no change in the number of pixels to be output even if the one-color pattern is replaced with the three-direction match, so there is no need to perform the counting process. Therefore, the process ends here. If the determined pattern is not one color (NO in S1812), it is determined whether the three directions match (S1813). Here, if the currently processed block is the uppermost or leftmost block of the image (YES in S1813), there are places where there are no blocks around and there is no coincidence in three directions, so the processing ends here. On the other hand, if the block is not the upper end or the left end block (NO in S1813), the pixel value of the first color of the block arranged next to or obliquely above and the pixel value of each block other than the first color of the currently processed block Are compared (S1814). If the pixel values are not matched as a result of the comparison (NO in S1815), the process ends here. If all the comparison results match (YES in S1815), 1 is added to the counter indicating the three-way match (S1816), and the process ends. The second counting means is realized by the processing of S1816.

[Pattern replacement process]
Next, the pattern replacement process (S1705) in the process described above will be described in detail with reference to FIG. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like. First, when the processing is started, the count result of S1702 (FIG. 18) is read from the storage unit, and the total data reduction amount when the patterns are replaced is calculated (S1901). Next, the data increase amount for each pattern when the patterns are replaced is calculated (S1902). The calculation method for S1901 and S1902 is as described above. Thereafter, the results calculated in S1901 and S1902 are compared. Here, when there is a pattern in which the total data reduction amount is larger than the data increase amount due to replacement (YES in S1903), the data increase amount is the smallest among the two-color or three-color patterns (pattern flags 2 to D). The pattern flag is selected as a replacement target pattern flag (S1904). The pattern selected in S1904 is treated as a four-color pattern (pattern flag E), is replaced with the three-direction matching process, and information indicating the replacement arrangement pattern is held in the storage unit (S1905). Here, a flag indicating the pattern selected in S1904 (hereinafter referred to as a pattern replacement flag) and a flag indicating which process has been added (hereinafter referred to as an additional processing flag) are output (S1906). This implements a flag output means. Then, this process ends. On the other hand, if there is no pattern in which the total data reduction amount is greater than the data increase amount (NO in S1903), this process ends without performing pattern replacement. Note that the third determination unit is realized by the processing of S1903.

[Compression after pattern replacement]
Next, the compression processing (S1707) after pattern replacement in the above-described processing will be described in detail with reference to FIG. Here, when it is determined that the compression rate for the image data can be improved by replacing the pattern, it is determined that the pattern is to be replaced, and compression is performed using the pattern after the allocation by the replacement. Indicates. In the following processing, for example, the CPU 105 reads and executes a program stored in the storage unit 107 or the like. First, when processing is started, it is determined whether the currently processed block is the first block. If the first block is not the first block (NO in S2001), the pixel values of the currently processed block and the storage block are compared. (S2002). If the pixel value matches the stored block (YES in S2003), the pattern flag (F) indicating that it is the same as the previous block is specified (S2004), and the output of the pixel value is omitted. A pattern flag is output (S2005).

  On the other hand, if the current block is the first block (YES in S2001), or if the pixel value of the currently processed block and the storage block do not match (NO in S2003), then: The process moves to the process of specifying the pattern flag of the block currently being processed. First, the pixel values of all the combinations of two pixels in the currently processed block are compared (S2007). Then, a pattern is specified and a pattern flag is created (S2008).

  Thereafter, the number of colors of the block currently being processed is determined (S2009). If the currently processed block is a one-color pattern (YES in S2010), the first color data (pixel value) and the pattern flag (0) are output (S2011). If it is not a one-color pattern (NO in S2010), it is determined whether the block currently being processed is the uppermost or leftmost block of image data (S2014). If it is not the top or left end block of the image (NO in S2014), the pixel of coordinates 2-4 of the block currently being processed and three other blocks that are adjacent or diagonally above (ie, 3) The first color data of the direction block) is compared (S2015). Thereby, a second determination unit is realized. If the pixel values in the three directions match (YES in S2016), the three directions are based on the color data (pixel value) of the first color of the processing target block and the pattern replacement flag output in S1906. The matching process and a pattern flag indicating the replaced pattern are output (S2017).

  If the block currently being processed is the top or left block of the image (YES in S2014), or if the pixel values in the three directions do not match in S2016 (NO in S2016), the next Perform the following process. First, it is determined whether the pattern of the block currently being processed is a block selected as a replacement target in S1904 (S2018). If the pattern has been replaced (YES in S2018), all the pixel values of the four pixels in the currently processed block are output (that is, the color of each pixel is the color data of the first color, The data is output as the second to fourth color data), and the four-color pattern flag (E) is output (S2019).

  On the other hand, if the currently processed block is not a replaced pattern (NO in S2018), it is specified which of the 13 types of patterns (other than repeat, one color, and three-direction matching) matches (S2020). ). Then, the first color data, the second to fourth color data corresponding to the identified pattern, and the pattern flags (2 to E) are output (S2021).

  After performing the above processing (S2011, S2017, S2019, S2021) in this way, data is expanded (S2012), and the expanded block is stored as the previous block (S2013). Then, the above processing is repeated until the processing of the last block is completed (S2006), and the compression processing after pattern replacement is completed.

  Here, the compression process when the repeat process and the three-way matching process are added to the normal compression process is described as an example, but the present invention is not limited to this. A process different from the repeat process or the three-way matching process may be added, or there may be three or more additional processes. When three or more processes are added, it can be dealt with by increasing the number of bits of the additional process flag.

[Write configuration to memory space]
Next, FIG. 21 shows an example of writing image data into the memory space. As shown in FIG. 21, the image data is divided into a pattern replacement flag, an additional processing flag, a pattern flag, the first color, and other colors, and stored in the memory in a distributed manner. Writing image data to the memory space is almost the same as normal compression processing, but the present embodiment is different in that the pattern replacement flag and the additional processing flag are written at the address before the pattern flag in the memory space.

  Next, a decompression process paired with the compression process will be described. The write start address of compressed data arranged on the memory 106 as shown in FIG. 21 is designated to the DMACs 221 and 231. Here, the compressed data refers to a pattern replacement flag, an additional processing flag, a pattern flag, and pixel values of the first color, the second, third, and fourth colors. With reference to the pattern replacement flag and the additional pattern flag, it is specified which pattern is used for the compression processing. The processing after the pattern is specified can be developed by the method described above.

  According to the present embodiment, in consideration of the total data reduction amount due to the pattern replacement, by replacing the two-color or three-color pattern and the three-direction matching process, the arrangement pattern information remains 4 bits, and the normal compression process is performed. It is possible to combine repeat processing and three-way matching processing. Then, after determining whether or not the compression rate is improved by pattern replacement, the pattern replacement processing and the compression processing using the replaced pattern are performed, so that the compression rate can be reliably improved.

<Other embodiments>
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program. It is a process to be executed.

Claims (8)

A dividing unit that divides image data into blocks each having a size of 2 × 2 pixels, and sequentially processes each of the divided blocks;
First determination means for determining whether or not the arrangement of the color data in the processing target block divided by the dividing means is the same as the arrangement of the color data in the previous block to be processed;
If it is determined by the first determination means that they are not the same, by comparing the color data of each pixel in the processing target block, from among a plurality of predefined pattern flags indicating the arrangement pattern of the color data, A specifying means for specifying a pattern flag indicating an arrangement pattern of color data included in the processing target block;
Extraction means for extracting the color of a pixel at a predefined position in the processing target block as color data of the first color, and further extracting color data other than the color data of the first color included in the block When,
First counting means for counting the number of appearances of each pattern flag specified by the specifying means;
With respect to the processing target block specified as a pattern flag indicating that two or more color data are arranged by the specifying unit, three pixels other than the pixel at a predefined position in the processing target block Second determination means for determining whether the color data matches the color data of the pixel at a predefined position in each of three different blocks adjacent to the three pixels;
Selection for selecting, as a replacement target pattern flag, a pattern flag with the smallest amount of data increase caused by replacing a pattern from the plurality of predefined pattern flags based on the result counted by the first counting means Means,
Output means for outputting at least one of the pattern flag or the color data,
The output means includes
For the processing target block that is determined to be the same by the first determination unit, the color arrangement in the processing target block is the same as the color arrangement in the block that was previously processed. Output a pattern flag indicating
For the processing target block that is a pattern flag that is determined not to be the same by the first determination unit and that indicates that the pattern flag specified by the specifying unit is a block composed of one color, the specification Outputting the pattern flag specified by the means and the color data of the first color in the processing target block extracted by the extraction means;
The processing target block that is determined to be not the same by the first determination unit and that is matched by the second determination unit is extracted by the pattern flag selected by the selection unit and the extraction unit. Output the color data of the first color in the processing target block.
The replacement target pattern flag that is determined not to be the same by the first determination unit, is determined not to match by the second determination unit, and the pattern flag specified by the specifying unit is selected by the selection unit For the processing target block, the processing target block is a block configured with four colors, the processing target block is a block in which four color data are arranged, and the processing target block Output color data for each of the four pixels in the block,
The replacement target pattern flag that is determined not to be the same by the first determination unit, is determined not to match by the second determination unit, and the pattern flag specified by the specification unit is selected by the selection unit For a non-processing target block, a pattern flag indicating an arrangement pattern of colors included in the processing target block specified by the specifying means, and the first in the processing target block extracted by the extracting means. An image processing apparatus that outputs color data of one color and color data other than the color data of the first color in the processing target block. The pixel at the predefined position is a lower right pixel in a block having a size of 2 × 2 pixels,
The second determination unit is such that the color data of the upper left pixel in the processing target block matches the color data of the lower right pixel of the block adjacent to the upper left of the processing target block, and
The color data of the upper right pixel in the processing target block matches the color data of the lower right pixel of the block adjacent to the processing target block, and
The color data of the lower left pixel in the processing target block and the color data of the lower right pixel of a block adjacent to the left of the processing target block are determined. An image processing method described in 1. Based on the second counting means for counting the number of blocks determined to match by the second determining means, the result counted by the first counting means and the result counted by the second counting means, the pattern flag is set. A third determination means for determining whether or not the data amount decreases when the data is replaced;
When the selection means determines that the third determination means decreases, the selection means replaces a pattern from among the plurality of predefined pattern flags based on the result counted by the first count means. The image processing apparatus according to claim 1, wherein a pattern flag with the smallest data increase amount is selected as the replacement target pattern flag.   The image processing apparatus according to claim 3, further comprising an irreversible compression unit that compresses the image data using a lossy compression method when the third determination unit determines that the data amount does not decrease. . A dividing unit that divides image data into blocks each having a size of 2 × 2 pixels, and sequentially processes each of the divided blocks;
By comparing the color data of each pixel in the processing target block, the arrangement pattern of the color data included in the processing target block is indicated from among a plurality of predefined pattern flags indicating the arrangement pattern of the color data. A specifying means for specifying a pattern flag;
Extracting means for extracting a color of a pixel at a predefined position in the processing target block as color data of a first color, and further extracting color data other than the color data of the first color included in the block; ,
The color data of three pixels other than the pixel at the predefined position in the processing target block is the color data of the pixel at the predefined position in each of three separate blocks adjacent to the three pixels. Determining means for determining whether or not they match,
For a processing target block determined to match by the determination means, a pattern flag indicating that the color data of a pixel at a predefined position in each of three different blocks adjacent to the three pixels matches , Outputting the color data of the first color in the processing target block extracted by the extraction means,
For processing target blocks determined not to match by the determination means, a pattern flag indicating an arrangement pattern of color data included in the processing target blocks specified by the specifying means and the extraction means are extracted. An image processing apparatus comprising: output means for outputting the color data of the first color and color data other than the color data of the first color. A dividing step of dividing the image data into blocks each having a size of 2 × 2 pixels, and sequentially dividing each of the divided blocks;
The first determination means determines whether or not the arrangement of the color data in the processing target block divided in the dividing step is the same as the arrangement of the color data in the previous block to be processed. 1 determination process;
A plurality of pre-defined pattern flags indicating color data arrangement patterns by comparing the color data of each pixel in the processing target block when the specifying means determines that they are not the same in the first determination step A specifying step for specifying a pattern flag indicating an arrangement pattern of color data included in the processing target block,
The extracting means extracts the color of a pixel at a predefined position in the processing target block as color data of the first color, and further extracts color data other than the color data of the first color included in the block. An extraction process to extract;
A first counting unit that counts the number of appearances of each of the pattern flags identified in the identifying step;
A pixel at a predefined position in the processing target block with respect to the processing target block specified by the second determination unit as a pattern flag indicating that two or more color data are arranged in the specifying step A second determination step of determining whether the color data of the other three pixels matches the color data of the pixel at a predefined position in each of three different blocks adjacent to the three pixels,
Based on the result of the counting in the first counting step, the selection means selects a pattern flag having the smallest data increase amount when the pattern is replaced from among the plurality of predefined pattern flags. A selection step to select as,
An output unit having an output step of outputting at least one of the pattern flag or the color data;
In the output step,
For the processing target block that is determined to be the same in the first determination step, the color arrangement in the processing target block is the same as the color arrangement in the block that was previously processed. Output a pattern flag indicating
For the processing target block which is a pattern flag indicating that the pattern flag specified in the first determination step is not the same and the pattern flag specified in the specific step is a block composed of one color, the specification Outputting the pattern flag specified in the process and the color data of the first color in the processing target block extracted in the extraction process;
For the processing target block determined to be not the same in the first determination step and determined to match in the second determination step, it is extracted in the pattern flag selected in the selection step and the extraction step. Output the color data of the first color in the processing target block.
The replacement target pattern flag that is determined not to be the same in the first determination step, is determined not to match in the second determination step, and the pattern flag specified in the specific step is selected in the selection step For the processing target block, the processing target block is a block configured with four colors, the processing target block is a block in which four color data are arranged, and the processing target block Output color data for each of the four pixels in the block,
The replacement target pattern flag that is determined not to be the same in the first determination step, is determined not to match in the second determination step, and the pattern flag specified in the specific step is selected in the selection step For a non-processing target block, a pattern flag indicating an arrangement pattern of colors included in the processing target block specified in the specifying step, and the first in the processing target block extracted in the extraction step. An image processing method comprising: outputting color data of one color and color data other than the color data of the first color in the processing target block. A dividing step of dividing the image data into blocks each having a size of 2 × 2 pixels, and sequentially dividing each of the divided blocks;
The specifying unit compares the color data of each pixel in the processing target block, so that the color data included in the processing target block can be selected from a plurality of predefined pattern flags indicating the arrangement pattern of the color data. A specific step of specifying a pattern flag indicating an arrangement pattern;
The extraction unit extracts the color of a pixel at a predefined position in the processing target block as first color data, and further extracts color data other than the first color data included in the block An extraction process to
The determination unit determines that the color data of the three pixels other than the pixel at the predefined position in the processing target block is the pixel data at the predefined position in each of three separate blocks adjacent to the three pixels. A determination step of determining whether or not each color data matches,
The output means is
For a processing target block determined to match in the determination step, a pattern flag indicating that it matches the color data of a pixel at a predefined position in each of three different blocks adjacent to the three pixels , Outputting color data of the first color in the processing target block extracted in the extraction step,
For processing target blocks determined not to match in the determination step, a pattern flag indicating an arrangement pattern of color data included in the processing target block specified in the specifying step and the extraction step are extracted. An image processing method comprising: outputting the color data of the first color and color data other than the color data of the first color.   A program for causing a computer to function as the image processing apparatus according to any one of claims 1 to 5.

Images