JP2012074897A - Image processing device, image processing method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a coding technique that achieves both high compression and high image quality through a simple process.SOLUTION: There is provided an image processing device comprising: compression means for setting a value of a pixel at a predetermined position included in a block of image data as a representative color and values of the other pixels as interpolation colors and outputting arrangement of the representative color and the interpolation colors; color subtraction means for performing a color subtraction process on each of a plurality of blocks; interpolation data amount calculation means for calculating a total amount of interpolation data, which is a sum of an amount of data of the interpolation colors obtained by performing a compression process on color-subtracted image data and an amount of data of the interpolation colors obtained by further performing a compression process on reduced image data formed of the representative color obtained by performing a compression process on image data; reduced amount calculation means for calculating a reduced amount, which is a difference between the amount of data of the interpolation colors obtained by performing the compression process on the color-subtracted image data and the amount of data of the interpolation colors obtained by performing the compression process on the image data; and selection means for selecting which of lossless coding and lossy coding is to be performed according to the total amount of interpolation data and the reduced amount.

Description

  The present invention relates to an image processing apparatus, an image processing method, and a program for compressing image data in predetermined block units. In particular, the present invention relates to a technique for easily and efficiently encoding high resolution image data.

  Conventionally, an image is divided into blocks each composed of m × n pixels, and this block has a configuration for lossless encoding and a configuration for lossy encoding. An image encoding technique for outputting as encoded data has been proposed. In such a technique, the lossy encoding method is selected to increase the compression rate in a natural image region where image quality degradation is relatively unnoticeable. On the other hand, the basic principle is to suppress visual image quality degradation by using a lossless encoding method for characters, line drawings, CG portions, etc., which are prone to image quality degradation. Patent Document 1 is known as the aforementioned method.

  As lossless encoding, there is a method of predicting a pixel value of a pixel of interest from values of surrounding pixels, obtaining a prediction error thereof, and encoding the prediction error. Alternatively, a run-length encoding technique that converts and encodes a continuous number of the same pixel value is often used. On the other hand, as irreversible encoding, a conversion encoding technique is generally used in which image data is converted into frequency space data using DCT, wavelet conversion, or the like, and coefficient values are encoded.

  In recent years, the resolution of image data has been increased with the increase in accuracy of image input / output devices. In order to process high-resolution image data at high speed, more hardware resources and processing time are required. Therefore, high-efficiency lossless and lossy encoding processing requires advanced calculation processing, and an image encoding technique that satisfies the three balances of compression performance, image quality, and calculation cost is required. As a method, for example, as in Patent Document 2, m × n divided image data is converted into reversible low-resolution data, and a reversible or irreversible compression method is selected according to the amount of interpolation data generated at the time of conversion. A method has been proposed. In addition, in order to further increase the compression rate, a technique has been proposed in which the data to be restored to the original resolution in the resolution conversion process is simply thinned to reduce the amount of data after encoding.

Japanese Patent Laid-Open No. 08-167030 JP 2010-141533 A

  In the method in which the resolution conversion is performed on the above-described image data and the lossless / lossy compression method is selected based on the interpolation data generated at the time of resolution conversion, there is a possibility that significant image quality degradation occurs. In other words, when the irreversible compression method is executed on the simple thinned data by resolution conversion, it is equivalent to performing irreversible image thinning for two times, and there is a problem of causing significant image quality degradation from the original image. is there.

  The present invention has been made in view of such problems, and an object thereof is to provide an encoding technique that achieves both high compression and high image quality by simple processing.

  In order to solve the above problems, the present invention has the following configuration. That is, an image processing apparatus that divides image data into blocks each having a predetermined number of pixels, and sequentially encodes the divided blocks. Among the pixels included in the block in the image data to be processed, A compression means for setting a pixel value at a predetermined position as a representative color, a pixel value other than the representative color as an interpolation color, and outputting an arrangement of the representative color and the interpolation color in the block as arrangement information; Color reduction means for performing color reduction processing for each block; data amount of interpolation color obtained by processing by the compression means for image data reduced by the color reduction means; and compression for the image data Obtained by further processing by the compression means on the reduced image data composed of the representative colors obtained by performing the processing by the means. Interpolation data amount calculation means for calculating the total interpolation data amount that is the sum of the inter-color data amount, and the interpolation color obtained by performing the processing by the compression means on the image data reduced in color by the color reduction means Reduction amount calculation means for calculating a reduction amount that is a difference between the data amount and the data amount of the interpolation color obtained by performing processing by the compression means on the image data, the overall interpolation data amount, and the A selection unit that selects whether to perform lossless encoding or lossy encoding according to the reduction amount, and when the selection unit selects lossless encoding, the compression unit outputs the When it is selected that lossless encoding means for losslessly encoding reduced image data composed of representative colors and irreversible encoding by the selection means are output by the compression means And a lossy encoding means for lossy encoding the reduced image data composed of serial representative colors.

  According to the present invention, it is possible to perform encoding that realizes high compression performance with reduced image quality degradation for high resolution image data.

1 is a diagram showing an overview of an MFP system. The figure which shows the outline | summary of a controller. The figure which shows the compression processing whole flow. The figure which enumerated the pattern of the block when dividing an image into blocks. The figure which enumerated the pattern of the block and its identifier. The figure which shows the processing flow performed by resolution conversion / interpolation data generation. The figure which shows a mode that the pattern of a block is converted into a flag. The figure which shows the relationship between the input and output with respect to a compression process. The figure which shows arrangement | positioning on the memory space of compressed data. The figure which shows the flow of the image compression by a color reduction method. The figure which shows the detail of an encoding selection means with a flow.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. FIG. 1 is an overall configuration diagram of a digital multi-function peripheral system (hereinafter referred to as MFP) that performs scanning, printing, and copying according to an embodiment of the present invention.

  The controller 101 is connected to a scanner 102 that is an image input device and a printer 103 that is an image output device. On the other hand, by connecting to a network 104 such as a LAN or a public line (WAN), input / output of image information and device information and image development of PDL (Page Description Language) data can be performed. A CPU 105 is a processor that controls the entire system. In the present embodiment, each process is realized by the CPU 105 reading and executing a program stored in the HDD storage unit 107 or the like.

  The memory 106 is a system work memory for the CPU 105 to operate, and is also an image memory for temporarily storing image data. The HDD storage unit 107 is a hard disk drive and stores system software and image data. Examples of the external input device 108 include a touch panel and a keyboard. The external input device 108 is a user interface device for a user using the MFP to perform various print settings.

  The operation of the controller unit in FIG. 1 will be described in detail with reference to FIG. A case of reading scan data will be described. In the controller 101 that has received read image data of three colors RGB (RED, GREEN, BLUE) from the scanner 102, first, the scanner image processing unit 201 performs image processing such as shading processing and filter processing. Then, the compression unit 202 performs image compression processing. The compressed data is stored in the memory 106 via a DMAC (Direct Memory Access Controller) 203.

  When printing scan data, the compressed data stored in the memory 106 is input to the color processing unit 212 via the DMAC 211 and converted into a CMYK (Cyan, Magenta, Yellow, Black) color space. Thereafter, color processing for adjustment such as density adjustment and printer gamma correction is performed on these CMYK values, and then stored again in the memory 106 via the DMAC 211. Thereafter, in order to perform image processing for printing, the compressed data stored in the memory 106 is read via the DMAC 221 and expanded into raster image data by the expansion unit 222. Raster CMYK image data is input to the print image processing unit 223, where area gradation processing is performed by a dither method or an error diffusion method, and is output to the printer 103. Note that the area gradation processing is not limited to the above method, and other methods may be used as long as the present invention can be applied.

  When the scan data is transmitted to the network, the compressed data stored in the memory 106 is input to the color processing unit 212 via the DMAC 211 to perform color conversion. Specifically, after performing display gamma adjustment, paper background color adjustment, and the like, conversion to a YCbCr (luminance, BLUE color difference, RED color difference) color space is performed. Thereafter, it is stored again in the memory 106 via the DMAC 211. Thereafter, in order to perform image processing for transmission, the compressed data stored in the memory 106 is read via the DMAC 231, and is expanded into raster image data by the expansion unit 232. For the raster YCbCr image data, the transmission processing unit 233 performs JPEG compression if the image data to be processed is color image transmission, and binarizes the Y data for monochrome binary image transmission. And perform JBIG compression. Then, the transmission processing unit 233 outputs the compressed image data to the network 104.

  When storing the scan data, the compressed data stored in the memory 106 is input to the disk spool high compression / decompression unit 242 via the DMAC 241. In the disk spool high compression / decompression unit 242, since the writing speed to the HDD is slower than the writing speed to the memory, higher compression JPEG compression is performed. Thereafter, the compressed data is stored in the HDD storage unit 107 via the disk access controller 243. When the stored data is expanded again in the memory 106, the reverse process is performed.

  A case where PDL data is written to the memory will be described. Although not shown in FIG. 2, the PDL data sent from the network in FIG. 1 is interpreted by the CPU 105 and a display list is output to the memory 106. Thereafter, the rendering list 251 renders the display list stored in the memory 106 into raster format RGB image data, and the compression unit 252 performs image compression processing. The compressed data is stored in the memory 106 via the DMAC 253. When PDL data is printed, transmitted to a network, and stored, it can be realized by performing processing similar to that of scan data. Hereinafter, the details of the compression unit of raster image data, which is a feature of the present invention, will be described.

[Overall compression processing flow]
First, the overall compression processing flow will be described with reference to FIG. In the present embodiment, an example in which the resolution of input image data is 1200 dpi and reduced image data having a resolution of 300 dpi is generated by performing resolution conversion twice. In the present embodiment, this processing flow is realized by the CPU 105 reading and executing a program stored in a storage unit such as the HDD storage unit 107. In the following processing of the present embodiment, “packet” refers to a unit of image data composed of 64 × 64 pixels in image data to be processed. The “tile” refers to a unit of image data composed of 2 × 2 pixels in the image data to be processed. Here, for convenience, the size of the image data is defined as “packet” and “tile”, but the present invention is not limited to this, and may be changed as necessary.

  In S300, the CPU 105 performs an initialization process before starting the encoding. Here, header information to be included in a code string is prepared for image data to be encoded, and initialization processing of various memories and variables is performed. The header information prepared here includes an encoding method, an attribute of data to be handled, and the like, but is not particularly limited in the present embodiment.

  After the initialization process in S300, in S301, the CPU 105 sequentially inputs the image data from the rendering unit 251 and the scanner image processing unit 201 to the compression units 252 and 202. As a result, image data for one line to be encoded is input to the line buffer. Image data handling and the like will be described later. Here, “line” refers to a pixel group in the scanning direction (main scanning direction) when compressing image data. In this embodiment, one line has a configuration of 1 × N pixels ( Here, N is the number of pixels in the main scanning direction of the image data).

  Next, in S302, the CPU 105 cuts out packet data (64 × 64 pixels) of interest from the image data for one line. Subsequently, in S303, the CPU 105 generates 1/2 reduced tile data by sub-sampling the upper left corner pixel in each pixel block in the tile data of interest (2 × 2 pixels). That is, 600 dpi image data is generated from 1200 dpi image data as input image data. This process is resolution conversion from 1200 dpi to 600 dpi, and the obtained data is referred to as “600 dpi packet data”. In the present embodiment, the position of the pixel to be subjected to sub-sampling is the pixel at the upper left corner in the block. However, the present invention is not limited to this, and a pixel at another position may be used. .

  In S304, the CPU 105 generates 1200 dpi interpolation data for restoring the original 1200 dpi resolution packet data from the 600 dpi packet data. Details of the interpolation data and the generation process will be described later with reference to FIGS. In step S <b> 305, the CPU 105 performs a color reduction process on the 1200 dpi image data in order to reduce interpolation data at the time of resolution conversion. Details of the color reduction processing will be described later with reference to FIG. In step S306, the CPU 105 generates reduced color 1200 dpi interpolation data for the data subjected to the reduced color process. Details of this processing will be described later with reference to FIGS.

  Subsequently, in S307, the CPU 105 generates 1/2 reduced tile data by sub-sampling the upper left corner pixel in each pixel block in the tile data of interest with respect to the 600 dpi packet data. That is, 300 dpi image data is generated from 600 dpi image data that is an input image. This process is resolution conversion from 600 dpi to 300 dpi, and the obtained data is referred to as “300 dpi packet data”. Also in the sub-sampling here, the pixel position is not limited to the upper left corner as in S303. In S308, the CPU 105 generates 600 dpi interpolation data for restoring the original 600 dpi resolution packet data from the 300 dpi packet data.

  In step S <b> 309, the CPU 105 determines whether to apply a lossless or lossy encoding method for 300 dpi packet data from the structure information of the interpolation data. Details of the selection method will be described later. If the selected encoding method is lossless encoding (YES in S310), 300 dpi packet data is losslessly encoded in S311 to generate encoded data. If the selected encoding method is lossy encoding (NO in S310), in S312, CPU 105 performs lossy encoding of 600 dpi packet data to generate encoded data.

  Subsequently, in S313, the CPU 105 combines the tile header, information identifying reversible / irreversible, interpolation data, and 1/2 reduced tile encoded data for the tile of interest to form tile encoded data. In step S <b> 314, the CPU 105 determines whether the tile on which the encoding process has been performed is the last packet of one line that is currently read. If it is not the last packet (NO in S314), the process is repeated from S302 on the next packet. If it is the last packet of one line (YES in S314), the process proceeds to S315.

  In step S315, the CPU 105 determines whether the currently read line is the last line of the image. If it is not the last line (NO in S315), the process returns to S301, and the CPU 105 reads the next line data and continues the processing. If it is the last line (YES in S315), the process proceeds to S316. When the process proceeds to S316, the encoding process has been completed for all the packets of the encoding target image data, and the CPU 105 generates final encoded data from the encoded data of all tiles, and performs DMAC253. Data is stored in the memory 106 via / 203. Thus, the present processing flow ends.

[Resolution conversion]
Hereinafter, details of the resolution conversion unit of raster image data will be described. This corresponds to the processing of S303 and S307 in FIG.

  First, in the image data to be processed, a raster image in units of pages is divided into blocks each composed of 2 × 2 pixels, and data is compressed in units of the divided blocks. Here, the block is called a “tile”. Prior to the description of the processing, the number of combinations of the arrangement of the colors will be considered according to the number of colors occupied in the 2 × 2 4-pixel data. Since the number of pixels is four, the number of colors occupied there is a maximum of four colors, and there is only a combination of one to four colors in the tile as the color arrangement pattern. The number of combinations of color arrangements that can be taken by these four color patterns will be described with reference to FIG. In the present embodiment, in a block composed of 2 × 2 pixels, a pixel value (color value) at a predetermined position is a representative color, and values of pixels other than the representative color are interpolation colors. The interpolation color is any number of 0 to 3 colors in a 2 × 2 pixel block. Therefore, the representative color is also referred to as the first color, and the interpolation color is also referred to as the second color, the third color, and the fourth color, respectively.

  In FIG. 4, what is composed of a collection of tiles each composed of 2 × 2 pixels is a “packet”, and in this embodiment, is composed of 64 × 64 pixels. First, consider one of the tiles included in the packet. When the color of the pixel included in the tile is one color, four pixels are configured in the same color, so there are only one combination of the arrangements.

  Next, consider a case where two colors of pixels are included in a tile. In the combination in which two colors are laid out in four pixels, if the value of the upper left pixel in the tile is considered as the first color and the other value as the second color, the remaining three pixels other than the upper left are changed to the first color or second Color enters. Therefore, except for the case of four pixels having the same color, a total of seven arrangement combinations are conceivable.

  Next, consider a case where pixels of three colors are included in a tile. The combination in which three colors are laid out in four pixels can be paraphrased as a case where only one of the three colors is used twice. Therefore, what is necessary is just to obtain | require the combination from which two pixels become the same color among the coordinates of four pixels. That is, the number in the case of three colors is a combination of taking two coordinates from four coordinates, and there are six ways in total.

  Finally, the values of the pixels included in the tile are all different, and in the case of four colors, there is only one pattern as in the case of one color.

  As described above, in the combination of the colors of the pixels included in the tile, a total of 15 patterns can be considered when the numbers for all of these 1 to 4 colors are added up. Considering adding a flag to identify all these patterns, 4 bits are required as the data amount. FIG. 5 shows this, and hereinafter, a flag that is arrangement information indicating an arrangement pattern of colors in a block is referred to as a “pattern flag”. The pattern flag here takes one of the values 0 to E.

[Compression processing flow]
Based on the possible combinations of 2 × 2 pixels as described above, a processing flow performed in resolution conversion / interpolation data generation in the compression units 202 and 252 will be described with reference to FIG. In the present embodiment, this process is realized by the CPU 105 reading and executing a program stored in a storage unit such as the HDD storage unit 107. In the present embodiment, the input has, for example, 256 gradations of 8 bits for each of RGB, and the data will be described as a 24-bit (8 bits × 3 colors) image per pixel in dot order of 8 bits data.

  First, when processing is started, a 2 × 2 pixel tile is input from the input data (S601). The CPU 105 takes a 24-bit compare for all combinations of two pixels in the input tile (S602). As a result of this comparison, “1” is output when all bits match, and “0” is output when they do not match. As shown in FIG. 7, the tile is composed of 2 × 2 pixels. If the pixels in the block are coordinates 1, 2, 3, 4 in the order of upper left, upper right, lower left, and lower right, the set of 2 pixels is 1-2, 1-3, 1-4, 2-3, 2-4, 3-4 in total. Therefore, it is necessary to take 6 comparisons in one block, and as a result, a 6-bit value is output. For example, when all the pixels have the same color, all the comparison results are “1”, and conversely, when all the four pixels are disjoint pixel values, all the comparison results are “0”. As described above, since there are 15 patterns that can appear due to color matching with 4 pixels, the CPU 105 converts the output 6-bit comparison result into a 4-bit pattern flag (S603).

  After the conversion to the pattern flag, the CPU 105 extracts the number of colors and color data that appeared in the four pixels in the tile (S604). Here, from the pattern flag, when the value of the upper left pixel in the tile is the first color, it is possible to obtain the position where the second and subsequent colors are located. This will be described with reference to FIG. For example, if it is determined that all four pixels in the tile have the same value and are composed of one color (YES in S605), the second and subsequent colors do not exist. 4 bits and the first color value (24 bits) are output (S606). If it is determined that the four pixels in the tile are composed of two colors (YES in S607), CPU 105 calculates the coordinates of the pixel having the value of the second color from the pattern flag, and the pattern flag (4 bits) and pixel values (48 bits) for two colors are output (S608). The CPU 105 performs the same processing when the number of colors included in the tile is three colors or four colors (S609, S610, S611). At this time, the CPU 105 stores the pixel values of the colors that did not appear in the order of the coordinates of the pixels in the tile (in the order of 1, 2, 3, 4 in the order of upper left, upper right, lower left, and lower right). To remember.

  In this way, the input data of 4 colors (96 bits) in a tile composed of 2 × 2 pixels is output by a relatively simple process by outputting the pixel values for the number of colors existing in the tile with 4 bit pattern flags. The amount of data can be reduced. Specifically, the input data 96 bits can be compressed to 28 bits for one color, 52 bits for two colors, and 76 bits for three colors. Note that data of 100 bits is output only in the case of four colors. Further, by referring to the pattern flag, the number of colors and the color arrangement in the tile can be specified.

  By performing the above processing on all tiles included in the image data to be processed, the image data can be compressed. Further, the resolution conversion process shown in S303 and S307 of FIG. 3 is realized by this process. At this time, the interpolation data means data of the second color, the third color, and the fourth color.

[Storage method]
The pattern flag and color data obtained as described above are written into the memory 106 via the DMAC. At that time, the DMAC changes the writing position on the storage area for each of the pattern flag, the first color data, and the second, third, and fourth color data. The DMAC has 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 second, third, and fourth color data. Is specified. An example is shown in FIG. For example, a case will be described where 8-bit image data of RGB colors each having a size of M × N pixels is input to the compression unit. In this case, the data size of the pattern flag is M × N / 8 bytes, the data size of the first color is M × N × 3/4 bytes, and the data of the second, third, and fourth colors for the entire image data to be processed. The output is indefinite (different depending on the image) depending on the image.

  Here, with respect to the memory area after the first color writing position, the pixel data is stored without being quantized or encoded in units of pixels. That is, color processing that is completed with one pixel input and one pixel output, for example, color conversion using LUT, gamma correction processing, color space conversion processing using matrix operation, etc., do not need to refer to the pattern flag directly. Processing can be performed. The color processing unit 212 shown in FIG. 2 reads pixel data after the first color writing start address on the memory 106 via the DMAC, and after writing the data in units of pixels, writes it back onto the memory 106. At this time, if the number of pixel bits is not changed by arbitrary pixel unit processing, the memory can be saved by overwriting the same location on the memory 106.

  By directly using the compressed data in this way, the transfer efficiency on the memory bus is improved, and data with a small number of pixels in the original image is processed, so that high-speed processing is possible.

  By storing the image data discretely in the memory space as shown in FIG. 9, when considering only the first color storage unit, the pixel of the upper left coordinate is stored for each tile of 2 × 2 pixel units. The sampled image results are continuously present in the memory space. The MFP described in this embodiment also has functions such as preview display of accumulated PDL image data and scanned image data, and network transmission described above. In this case, even if the print resolution is 600 dpi, a resolution up to that is usually not required at the time of preview or transmission, and 300 dpi or less is often sufficient. When it is necessary to obtain such reduced data, a half-size raster image can be easily obtained by using only the first color image data without using the pattern flag or the second, third, and fourth color data. I can do it.

  For example, in the above-described example, a reduced transmission time when image data is stored at 600 dpi will be described. When a resolution larger than 300 dpi being sampled is specified, such as 400 dpi as the resolution at the time of transmission, the compressed data including the pattern flag is once expanded and transmitted using a known scaling method. To do. On the other hand, when transmission resolution less than 300 dpi is specified, only the data in the first color storage unit is used, and data is read while switching according to the required image size, such as performing scaling processing to the specified resolution. Do.

[Development part]
Next, the expansion units 222 and 232 that are paired with the compression unit will be described. The development units 222 and 232 indicate processing for returning to the original raster image data using the pattern flag and pixel data as described above. The expansion units 222 and 232 have three patterns, the pattern flag write start address, the first color write start address, and the second, third and fourth color write start addresses of the compressed data arranged in the memory space as shown in FIG. Specify the address to the DMAC. The DMAC reads the compressed data from the three addresses and transmits it to the decompressing units 222 and 232. The development units 222 and 232 first interpret the 4-bit pattern flag and calculate the number of colors in the tile. In addition to the first color data, the second, third and fourth color data are read according to the number of colors. Then, the development units 222 and 232 rearrange the pixel data in accordance with the color arrangement defined by the pattern flag. As a result, a tile composed of 2 × 2 pixels is expanded and decoded.

  Further, 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 write start address is stored in the DMAC. Specify only. Then, only the first color data is read from the memory 106 to form an image. As a result, the bandwidth of the image memory bus can be saved.

  In the present embodiment, a tile composed of 2 × 2 pixels is described as the minimum unit, but the present invention is not limited to this. For example, the number of pixels may be changed in accordance with image characteristics. In the description of the compression, 8 bits for each color of RGB has been described as an example of image data. However, the image data may have a CMYK color space, GRAY scale data, or a pixel value other than 8 bits.

[Color reduction processing]
Details of the color reduction processing method will be described below. An image that is expected to be highly compressed by the resolution conversion processing method is an image in which there are many areas in which the adjacent pixel value is not different from one level in the image data. The level here indicates the gradation of the color value of each pixel. For example, when each color is expressed with 256 gradations, it means a value of 0 to 256. This is often the case with raster image data rendered as high-resolution PDL data. However, in image data read by a scanner, noise components from the scanner are superimposed on the image, and several levels of normal pixel values are used. Differences (color value differences) can occur. In such a case, as compressed data, it is often determined that pixel values of four colors are included in a tile composed of 2 × 2 pixels, and not only the amount of data is not reduced after compression but also the pattern The amount of data will increase as flags are added. However, depending on the image data, it is often impossible to visually recognize the difference in several levels within a tile composed of 2 × 2 pixels. In such a case, the difference in several levels can be said to be redundant data.

  Therefore, when two or more colors are included in the block for each pixel having several levels of redundant differences in the tile, the number of colors is reduced by performing a color reduction process. After this, a pattern flag is created to improve the compression rate. As described above, a flow for performing image compression based on possible combinations of 2 × 2 pixels will be described with reference to FIG. The input will be described as image data having 256 gradations of 8 bits for each RGB color as before. In the present embodiment, this process is realized by the CPU 105 reading and executing a program stored in a storage unit such as the HDD storage unit 107.

  First, when processing is started, a tile composed of 2 × 2 pixels is input (S1001). For the input tile, the CPU 105 performs color reduction processing to one color for the data of the four pixels included in the tile (S1002). Here, for example, the color reduction to one color is realized by calculating the average pixel value of four pixels. Subsequently, the CPU 105 calculates the difference between the pixel value obtained by the color reduction process and the input pixel value of the four pixels (S1003). Then, the CPU 105 determines the magnitude of the difference (S1004). For example, the sum of absolute values of the difference between the input value and the RGB value after color reduction is taken, and when the value is equal to or less than a predetermined threshold, it can be determined that the difference is small. If it is determined that the difference is small (YES in S1004), the CPU 105 assigns a pattern flag to the tile to be processed (S1005). Then, the CPU 105 outputs the pixel value of one color and the pattern flag that have been reduced in color (S1006).

  If it is determined that the error is large (NO in S1004), the CPU 105 performs a color reduction process to two colors (S1007). Here, for example, two pixels A and B having the largest RGB value difference are extracted from the four pixels included in the tile, and the remaining two pixels are clustered according to which one is closer to A or B, and the average within the cluster is extracted. By obtaining the value, the two colors are obtained. Then, the CPU 105 calculates the difference between the two-color pixel value and the pixel value of the four pixels included in the input tile (S1008). Then, the CPU 105 determines the magnitude of the difference (S1009). This determination can be made by comparing with a predefined threshold value in the same manner as in S1004. If it is determined that the difference is small (YES in S1009), CPU 105 detects a two-color pattern based on FIG. 4 and assigns a pattern flag (S1010). Then, the CPU 105 outputs the pixel values of two colors and the pattern flag that have been reduced in color (S1011).

  If it is determined again that the difference is large (NO in S1009), the CPU 105 performs a color reduction process to three colors (S1012). Here, for example, two pixels having the smallest difference in RGB values are extracted from the four pixels, the values of the two pixels are averaged, and three colors of four pixels are obtained using the average value and the values of the other two pixels. Reduce to. Then, the CPU 105 calculates a difference between the pixel value of the three-colored tile and the pixel value of the four pixels included in the input tile (S1013). Then, the CPU 105 determines whether the difference is large or small (S1014). This determination can be made by comparing with a predefined threshold value in the same manner as in S1004. If it is determined that the difference is small (YES in S1014), CPU 105 detects a pattern of three colors based on FIG. 4 and assigns a pattern flag (S1015). Then, the CPU 105 outputs the pixel values of three colors and the pattern flag that have been reduced in color (S1016).

  If it is determined that the difference is large again (NO in S1014), this means that the tile is visually problematic due to color reduction, so the CPU 105 determines the pixel values of all four colors and the corresponding pattern flags. Is given (S1017). Then, the CPU 105 outputs all four pixels included in the input tile (S1018). Thus, the present processing flow ends.

  By adopting such a configuration and applying the processing to all the tiles included in the image data to be processed, the compression ratio is drastically deteriorated due to the difference in the number of pixels as represented by the scanned image. Therefore, it is possible to improve the compression rate even for such an image. Further, the compression rate and the image quality can be adjusted depending on the difference determination threshold value. For example, by completely eliminating the difference between allowed pixels (set to 0), the image is completely reversible. On the contrary, by making the allowable difference infinite, all tiles are expressed by one color, which is equivalent to halving the resolution, and it is possible to guarantee that the data amount is ¼.

  Here, the description is given in the compression unit using subtractive color. However, if compression is performed using this method, it is not necessary to be aware of the compression method of whether or not the decompression process is accompanied by subtractive color, and the resolution described above. The configuration may be exactly the same as the expansion processing described in the detailed description of the conversion processing.

  Note that the color reduction processing used in the description here is an example, and other known color reduction processing may be used. Further, although it is a means for calculating the difference between the input pixel and the pixel after the color reduction processing, it may be replaced by another means for obtaining a known color difference.

[Encoding method selection]
A method for selecting an encoding method, which is a characteristic part of the present invention, will be described with reference to FIG. In the present embodiment, this process is realized by the CPU 105 reading and executing a program stored in a storage unit such as the HDD storage unit 107.

  The process flow of FIG. 11 explains part of the process (S309 to S312) shown in FIG. 3 in more detail. First, the CPU 105 acquires the reduced color 1200 dpi interpolation data amount generated in S306 (S1100). The acquisition method here is acquired from the amount of data stored in the memory 106. Next, the CPU 105 acquires the 600 dpi interpolation data amount generated in S308 (S1101). The acquisition method here is acquired from the amount of data stored in the memory 106.

  Subsequently, the CPU 105 calculates an overall interpolation data amount that is the sum of the subtractive color 1200 dpi interpolation data amount and the 600 dpi interpolation data amount (S1102). Thereby, an interpolation data amount calculation means is realized. Then, the CPU 105 compares the calculated total interpolation data amount with the interpolation data amount threshold value (first threshold value) (S1103). If the total interpolation data amount is larger (YES in S1103), the process proceeds to S1104. If the total interpolation data amount is smaller (NO in S1103), the process proceeds to S1106. The interpolation data amount threshold value can be set by register setting from the CPU 105. In addition, the value can be changed via the CPU 105 depending on the type of image or input from the external input device 108.

  Subsequently, the CPU 105 calculates a difference between the 1200 dpi interpolation data amount generated in S304 and the subtractive color 1200 dpi interpolation data (S1104). This value is the amount of interpolation data reduction. Thereby, a reduction amount calculation means is realized. Then, the CPU 105 compares the interpolation data reduction amount with a reduction amount threshold (second threshold) (S1105). If the interpolation data reduction amount is large (NO in S1105), the process proceeds to S1106. If the interpolation data reduction amount is small (YES in S1105), the process proceeds to S1107. Similar to the interpolation data amount threshold, the reduction amount threshold can be set by register setting from the CPU 105. In addition, the value can be changed via the CPU 105 depending on the type of image or input from the external input device 108. In step S1106, the CPU 105 performs lossless encoding on the 300 dpi packet data. In step S1107, the CPU 105 performs lossy encoding on the 600 dpi packet data. Then, after the process of S1106 or S1107, the process proceeds to the process of S313 in FIG.

  The first threshold value and the second threshold value used in S1103 and S1105 may be determined based on the total data amount before and after compression, image characteristics, and the like.

  With the above processing, it is possible to select the encoding process according to the total interpolation data amount (corresponding to the second, third, and fourth color data in this embodiment) that is a variable part of the encoding data amount. . That is, when the interpolation data is small, lossless coding with a relatively low coding rate is selected to prevent image quality deterioration. Even when the amount of interpolation data is large, the encoding process can be selected according to the amount of interpolation data reduced by the color reduction process. In other words, when the amount of data reduction is large due to the color reduction processing, image quality degradation is suppressed by selecting lossless encoding. On the other hand, when the data reduction amount due to the color reduction process is small, the image quality deterioration due to the color reduction process is relatively small, so lossy encoding is selected to reduce the overall data amount.

  What is important when selecting an encoding method is to optimize the amount of data reduction due to encoding and the degree of image quality degradation. Through the above processing, it is possible to perform encoding that suppresses data amount and image quality deterioration according to an input image.

<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 (9)

An image processing apparatus that divides image data into blocks each having a predetermined number of pixels, and sequentially encodes the divided blocks.
In the image data to be processed, among the pixels included in the block, the value of a pixel at a predetermined position is a representative color, the value of a pixel other than the representative color is an interpolation color, and the representative color and the interpolation color in the block Compression means for outputting the arrangement of as arrangement information;
Color reduction means for performing color reduction processing on each block in the image data;
The amount of interpolated color data obtained by performing the processing by the compression unit on the image data reduced by the color reduction unit, and the representative color obtained by performing the processing by the compression unit on the image data Interpolation data amount calculation means for calculating a total interpolation data amount that is a sum of the interpolation color data amount obtained by further performing processing by the compression means on the reduced image data comprising:
The amount of interpolation color data obtained by performing the processing by the compression unit on the image data reduced by the color reduction unit, and the interpolation color obtained by performing the processing by the compression unit on the image data A reduction amount calculation means for calculating a reduction amount that is a difference from the data amount of
Selection means for selecting whether to perform lossless encoding or lossy encoding according to the total interpolation data amount and the reduction amount;
A lossless encoding means for losslessly encoding the reduced image data composed of the representative color output by the compression means when the selection means is selected to perform lossless encoding;
And irreversible encoding means for irreversibly encoding the reduced image data composed of the representative color output by the compression means when the selection means selects to perform irreversible encoding. Image processing device. The compression means includes
A dividing means for dividing the image data to be processed into blocks of 2 × 2 pixels;
Means for determining the number of colors contained in the block;
Means for obtaining the arrangement information for the colors included in the block;
Means for obtaining the representative color from the colors included in the block;
Means for obtaining 0 to 3 interpolation colors other than the representative color from among the colors included in the block;
The image processing apparatus according to claim 1, wherein the arrangement information, the representative color, and the interpolation color are stored in different storage areas. The color reduction means is
When the number of colors of 2 or more is included in the block composed of 2 × 2 pixels, a determination is made as to whether or not the number of colors can be reduced according to the difference in color value included in the block Means,
3. The image processing apparatus according to claim 1, further comprising a reduction unit configured to reduce the number of colors included in the block when it is determined that the reduction can be performed by the determination unit.   The selection unit compares the total interpolation data amount with a first threshold value, and selects the lossless encoding by the lossless encoding unit when the total interpolation data amount is smaller than the first threshold value. The image processing apparatus according to claim 1, wherein the image processing apparatus is an image processing apparatus.   The selection unit compares the reduction amount with a second threshold, and when the reduction amount is smaller than the second threshold, selects the lossy encoding by the lossy encoding unit, and The image processing apparatus according to any one of claims 1 to 4, wherein when the reduction amount is larger than a second threshold value, lossless encoding by the lossless encoding unit is selected.   When the selection unit selects that lossless encoding by the lossless encoding unit is performed, the image data including the representative colors obtained by performing the processing by the compression unit on the processing target image data is further added. 6. The lossless encoding by the lossless encoding means is applied to reduced image data composed of representative colors obtained by performing the processing by the compression means. Image processing apparatus.   When the selection means selects to perform lossy encoding by the lossy encoding means, a reduced image composed of representative colors obtained by performing processing by the compression means on the image data to be processed The image processing apparatus according to claim 1, wherein the processing by the lossy encoding unit is applied to data. An image processing method in which image data is divided into blocks each having a predetermined number of pixels, and the divided blocks are sequentially encoded.
In the image data to be processed, the compression means uses the value of a pixel at a predetermined position among the pixels included in the block as a representative color, the value of a pixel other than the representative color as an interpolation color, and the representative color in the block And a compression process for outputting the arrangement of interpolation colors as arrangement information;
A color reduction process in which the color reduction means performs a color reduction process on each block in the image data;
Interpolation data amount calculation means performs processing in the compression step on the interpolation color data amount obtained by performing the processing in the compression step on the image data reduced in the color reduction step, and the image data. An interpolation data amount calculation step of calculating an overall interpolation data amount that is the sum of the interpolation color data amount obtained by further performing the processing in the compression step on the reduced image data composed of the representative colors obtained When,
The reduction amount calculation means performs processing in the compression step on the interpolation color data amount obtained by performing the processing in the compression step on the image data reduced in the color reduction step, and the image data. A reduction amount calculation step of calculating a reduction amount that is a difference from the data amount of the interpolation color obtained by
A selection step in which the selection means selects whether to perform lossless encoding or lossy encoding according to the total interpolation data amount and the reduction amount;
When the lossless encoding means is selected to perform lossless encoding in the selection step, the lossless encoding step of losslessly encoding the reduced image data composed of the representative color output in the compression step;
An irreversible encoding step of irreversibly encoding the reduced image data composed of the representative colors output in the compression step when the irreversible encoding means is selected to perform irreversible encoding in the selection step; An image processing method comprising: Computer
In the image data to be processed, among the pixels included in the block, the value of a pixel at a predetermined position is a representative color, the value of a pixel other than the representative color is an interpolation color, and the representative color and the interpolation color in the block Compression means for outputting the arrangement as arrangement information;
A color reduction means for performing a color reduction process on each block in the image data;
The amount of interpolated color data obtained by performing the processing by the compression unit on the image data reduced by the color reduction unit, and the representative color obtained by performing the processing by the compression unit on the image data Interpolation data amount calculation means for calculating the total interpolation data amount that is the sum of the interpolation color data amount obtained by further performing processing by the compression means on the reduced image data consisting of:
The amount of interpolation color data obtained by performing the processing by the compression unit on the image data reduced by the color reduction unit, and the interpolation color obtained by performing the processing by the compression unit on the image data Reduction amount calculating means for calculating a reduction amount that is a difference from the data amount of
Selection means for selecting either lossless encoding or lossy encoding according to the total interpolation data amount and the reduction amount;
A lossless encoding means for losslessly encoding the reduced image data composed of the representative color output by the compression means when the selection means is selected to perform lossless encoding;
A program for functioning as an irreversible encoding unit for irreversibly encoding reduced image data composed of the representative color output by the compression unit when the selection unit selects to perform irreversible encoding.

Images