JP2006246374A - Image compressor, image compression method, and image compression program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image compression system that can compress image data mixedly containing text, photographs, figures, and the like with high compressibility and without damaging the features of an original image, and causes a small load to a reception terminal whose processing capability is low. <P>SOLUTION: A rectangular block extraction section 12 divides image data for extracting as a plurality of rectangular blocks. A compression mode determination section 13 determines the number of colors for indicating the number of colors contained in each rectangular block and an edge rate for indicating the rate of an edge image. The rectangular blocks are sorted into A type rectangular blocks, where the color number is 1, B type rectangular blocks, where the color number is less than a threshold th1, C type rectangular blocks, where the color number is a threshold th1 or higher and the edge rate is less than the threshold th2, and D type rectangular blocks, where the color number is the threshold th1 or higher and the edge rate is the threshold th2 or higher. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image compression method and an image compression apparatus that quantize and encode image data to reduce the amount of the data, and in particular, compresses an image that is displayed on a computer screen or the like and that includes photographs, figures, characters, and the like. The present invention relates to an image compressing apparatus, an image compressing method, and an image compressing program suitable for use in the above.

  In order to reduce the amount of data when storing and communicating image data, an image compression technique is generally used. By performing image compression, it is possible to store image data in a terminal having a limited storage capacity, such as a mobile phone terminal, or to transmit image data via a transmission path with a low transmission rate. Become. Various image compression methods have been proposed, each having advantages and disadvantages, and it is necessary to use different methods depending on the nature of the image. For example, the JPEG baseline uses an irreversible compression method in which high-frequency components are reduced using discrete cosine transform (DCT) or the like. This method is suitable for photographic images with a large number of colors and a gradual change in density, but because high-frequency components are reduced, images that are blurred when edges are applied to images with clear edges such as figures and characters. It is inappropriate.

A color reduction process is also used as an image compression method that does not lose the characteristics of an image including an edge (see, for example, Patent Document 1), but the color characteristics of an original image are often damaged. As described above, since various image compression methods have advantages and disadvantages, if a single image compression method is applied to image data in which text, photographs, graphics, etc. are mixed, for example, a certain part in an image is appropriately displayed. In addition, the image is compressed at a high compression rate, but the other portions are likely to have image characteristics deteriorated or an appropriate compression rate cannot be obtained.
JP-A-10-341347

  The present invention is an image compression apparatus capable of compressing image data in which text, photos, graphics, and the like are mixed, at a high compression rate and without damaging the characteristics of the original image, and reducing the load on the receiving terminal with low processing capability. An object of the present invention is to provide an image compression method.

  An image compression apparatus according to an aspect of the present invention includes a rectangular block extraction unit that divides image data based on colors and extracts the plurality of rectangular blocks, and a color indicating the number of colors included in each of the rectangular blocks And the edge ratio indicating the ratio of edge pixels is determined, the rectangular block having the single color number is the first type rectangular block, and the rectangular block having the color number less than the predetermined value is the second type rectangular block. The rectangular block whose number of colors is a predetermined value or more and the edge rate is less than a predetermined value is a third type rectangular block, and the number of colors is a predetermined value or more and the edge rate is a predetermined value or more. A determination unit that classifies each rectangular block as a fourth type rectangular block, a palette compression unit that applies palette compression to the image data of the second type rectangular block, and an image of the third type rectangular block An irreversible compression encoding unit that performs irreversible data compression on the data, predicts the number of important colors based on the number of colors of the fourth type rectangular block, and predicts the colors included in the fourth type rectangular block. And a color reduction processing unit that performs color reduction processing to reduce the number of important colors.

  An image compression method according to an aspect of the present invention includes a step of dividing image data based on colors to extract a plurality of rectangular blocks, a number of colors indicating the number of colors included in each of the rectangular blocks, and A step of determining an edge rate indicating a ratio of edge pixels; the rectangular block having a single color number is a first type rectangular block; the rectangular block having a color number less than a predetermined value is a second type rectangular block; The rectangular block in which the number of colors is equal to or greater than a predetermined value and the edge rate is less than a predetermined value is a third type rectangular block, and the rectangle in which the number of colors is equal to or greater than a predetermined value and the edge rate is equal to or greater than a predetermined value Classifying each block as a fourth type rectangular block, performing palette compression on the image data of the second type rectangular block, and image data of the third type rectangular block A step of performing irreversible data compression, and a color reduction for predicting the number of important colors based on the number of colors of the fourth type rectangular block and reducing the number of colors included in the fourth type rectangular block to the predicted number of important colors And a step of performing processing.

  According to another aspect of the present invention, there is provided an image compression program configured such that the image compression method can be executed by a computer.

  An image compression apparatus according to another aspect of the present invention includes a rectangular block extraction unit that divides image data based on colors and extracts a plurality of rectangular blocks, and the number of colors included in each of the rectangular blocks. A determination unit that at least classifies a rectangular block having a single number of colors as a first rectangular block, a rectangular block having a plurality of colors as a second rectangular block, and the second rectangular block; An image compression unit that selectively performs an image compression method related to the number of colors and the edge rate of the image.

  According to the present invention, image data in which text, photographs, graphics, and the like are mixed can be compressed with a high compression rate and without damaging the characteristics of the original image, and an image with low load on a receiving terminal with low processing capability. A compression device and an image compression method can be provided.

  Next, an image compression apparatus according to an embodiment of the present invention will be described in detail with reference to the drawings.

  First Embodiment The overall configuration of an image compression apparatus according to a first embodiment of the present invention will be described with reference to the block diagram of FIG. As shown in the block diagram of FIG. 1, the image compression apparatus includes a color image input device 11, a rectangular block extraction unit 12, a compression mode determination unit 13, an irreversible encoding unit 14, an important color number prediction unit 15, a color reduction process, and the like. 16, a palette compression unit 17, a block grouping unit 18, a lossless encoding unit 19, and a data shuffle unit 20, and these components can be realized by a computer and a computer program. The optimum compression format for each part in the color image is determined from a plurality of compression formats, and a different compression method is applied to each part. Hereinafter, each element will be described.

  The color image input device 11 is a device for inputting a color image, which is an original image before compression, for example, a 16.77 million full color image in which 3 colors of RGB are expressed by 8 bits each, and is, for example, a scanner. An image expressed in the LUV system or the HSV system instead of the RGB system may be used.

  The rectangular block extraction unit 12 has a function of extracting a partial image in the input color image by dividing it into rectangular blocks according to the properties thereof. For example, as shown in FIG. 2, the computer screen displays a graphic image such as a character, an icon, or a partial image such as a photograph, and a monochromatic area also exists in the gap. The rectangular block extracting unit 12 in this embodiment extracts these partial images by dividing them into a single-color rectangular block and a rectangular block (noted rectangular block) that is not, but is limited to this. However, the dividing method can be variously changed without departing from the gist of the invention.

  The compression mode determination unit 13 determines a compression method suitable for the plurality of extracted rectangular blocks based on the number Nc of colors included and the edge rate Eh indicating the ratio of edge images. In this embodiment, the monochromatic rectangular block is determined as a type A rectangular block. A type A block can be expressed only by a single color value and the size and position of a rectangular block, and does not require a compression operation.

  On the other hand, the title rectangular block is determined as type B, C, or D depending on whether the number of colors Nc is larger than a threshold th1 (here, th1 = 8) and whether the edge rate Eh is larger than a threshold th2. The That is, a rectangular block whose number of colors Nc is smaller than a threshold th1 (here, th1 = 8 is recommended) among the title rectangular blocks is determined as a type B rectangular block. Among the title rectangular blocks, rectangular blocks having the number of colors Nc equal to or greater than the threshold th1 and the edge rate Eh less than the threshold th2 are determined to be type C rectangular blocks. The type C rectangular block is typically a photographic image, but is not limited thereto. Among the title rectangular blocks, blocks whose color number Nc is equal to or greater than th1 and whose edge rate Eh is equal to or greater than the threshold th2 are classified as type D rectangular blocks. The type D block is typically an image such as an icon in which edges are clearly drawn in many colors, but is not limited thereto.

  The lossy encoding unit 14 is for compressing a type C rectangular block among the types A to D, and guarantees a high compression rate by a compression method with a partial loss of image quality. An example of the compression method here is a discrete cosine transform (DCT) that converts and encodes image data into a sum of various frequency components to reduce high-frequency components and encodes only low-frequency components with high correlation. However, the present invention is not limited to this.

  The important color number prediction unit 15 is for compressing the type D rectangular block among the types A to D, and performs preprocessing of the color reduction processing by the color reduction processing unit 16. In other words, the important color number prediction unit 15 predicts the number of colors that are considered to be important (predicted important color number) Nic among the color numbers Nc of the colors actually included in the type D rectangular block. The color reduction processing unit 16 has a function of executing color reduction processing of a type D rectangular block based on the predicted important color number Nic predicted by the important color number prediction unit 15. Specifically, important colors are selected from the colors included in the type D rectangular block by the number of predicted important colors Nic based on a predetermined criterion, and are selected for pixels that are not important colors. Then, a process of replacing the pixel with the closest color among the Nic important colors is performed.

  The palette compression unit 17 has a function of performing palette compression on the compression of the type B rectangular block among the types A to D. That is, the palette compressing unit 17 prepares a palette corresponding to the number of colors included in the type B rectangular block, and stores a number for specifying each palette and a color value indicating each color in association with each other. is there. When the number of colors Nc is as small as less than 8, for example, a large compression rate can be obtained without reducing information. In this embodiment, when creating palette data, the palettes are arranged in descending order of the number of pixels. By performing this alignment, the relevance of a plurality of rectangular blocks can be increased, and the compression rate can be further improved in the block grouping process to be performed later.

  The block grouping unit 18 groups the type D rectangular block after the color reduction processing by the color reduction processing unit 16 and the type B rectangular block after the palette compression by the palette compression unit 17 for each block having the same number of colors. It has a function. The lossless encoding unit 19 has a function of further increasing the compression rate by applying a lossless encoding method to the grouped rectangular blocks. The lossless encoding method here is a method for reducing the amount of information by suppressing the redundancy of the image. For example, run-length encoding, Huffman encoding, LZW, JBIG, MMR, etc. can be considered. It is not limited.

  The data shuffle unit 20 has a function of converting data indicating a type A rectangular block and compressed type B to D rectangular blocks into a data structure suitable for communication in a communication circuit and data storage in a storage medium. . As described above, the type A rectangular block is directly input to the data shuffle unit 20 without being subjected to a compression operation.

  As described above, in the present embodiment, among the type A to D rectangular blocks, the color reduction processing based on the predicted number of important colors is performed for the type D rectangular block. In an image compression method such as discrete cosine transform, similar calculation processing is required in a transmission terminal that performs image compression and a reception terminal that decodes an image. For this reason, if the compression encoding process is complicated in order to increase the compression rate at the transmission side terminal that performs compression encoding, the reception terminal that performs decompression decoding accordingly requires complicated data processing for decoding. Become. In a case where image data is transmitted to a terminal having a small processing capability such as a mobile phone terminal, it is difficult to decode the image data subjected to a complicated compression process because the processing capability is limited.

  In the present embodiment, the color reduction processing executed for the type D rectangular block is asymmetric processing, and processing on the receiving terminal side that performs data decompression (such as inverse quantization) is unnecessary. Accordingly, the load on the receiving terminal on the data decompression side is reduced.

  [Rectangular Block Extraction] Next, the operation of each unit shown in FIG. 1 will be described. First, the operation of the rectangular block extraction unit 12 will be described with reference to the flowchart of FIG. The rectangular block extraction by the rectangular block extraction unit 12 first extracts a single-color rectangular block (which will be identified later as a type A rectangular block) to the maximum (S11 to S14), and the remaining portion is not a single color. This is performed by extracting as a “marked rectangular block” (later specified as a type B, C, or D rectangular block) in which some figure, character, pattern, photograph or the like is marked.

  A procedure for extracting a monochrome rectangular block will be described. A zigzag scan is started from the upper left to the lower right with respect to the computer screen of the original image (S11), and a minimum single color rectangular block of 16 × 16 dots is searched (S12). When a 16 × 16 dot minimum single-color rectangular block is found (YES in S12), it is expanded to the maximum by one row / column in four directions, up, down, left, and right (S13). Here, “maximum” is intended to increase the data compression rate by extending the monochrome rectangular block to the vicinity of the boundary of the non-monochromatic region. There are three possible methods for expansion. In the first method, when there is a monochromatic area as shown in FIG. 4A, as shown in FIG. 4B, the minimum monochromatic rectangular block is first expanded to the maximum in the left-right direction, and then the maximum up and down. It is a method of extending to the limit. As shown in FIG. 4C, the second method is a method in which the image is expanded up to the maximum in the vertical direction and then expanded in the horizontal direction to the maximum. As shown in FIG. 4D, the third method is a method in which the horizontal direction and the vertical direction are alternately expanded to the maximum.

  The portion of the single color rectangular block that has been extracted is set as a processed area of rectangular block extraction (S14). When all the single color rectangular blocks are cut out and no new 16 × 16 dot minimum single color rectangular block is found (NO in S12), the extraction step of the single color rectangular block ends, and the remaining area divided by the single color rectangular block is Each divided area is extracted as a title rectangular block (S15).

  In the above example, the size of the minimum single-color rectangular block has been described as 16 × 16 dots. However, the present invention is not limited to this. However, if the size of the minimum single-color rectangular block is smaller than the inter-character distance and inter-character distance of the characters included in the original image, the screen is divided too much, and conversely, the total compression rate may be lowered. Therefore, the size of the minimum single-color rectangular block is preferably set with the text character size as a guide. Further, the expansion of the monochromatic rectangular block may be performed for each row and column, but may be performed for a plurality of rows / columns in order to increase the processing speed.

  Subsequently, the remaining rectangular block other than the monochrome rectangular block is zigzag scanned to search for the pixel of the rectangular block (S16). If a pixel is found, unless it reaches the end of the image (NO in S17), As in the case of the expansion of the monochrome rectangular block, the region including the pixels is expanded to the maximum by one row / one column (or by a plurality of rows / columns) in four directions, up, down, left, and right (S18). When the expansion is completed, it is assumed that the expanded area has been processed as a title rectangular block (S19). If the end of the image has been reached (YES at S17), the extraction process is terminated (S20), assuming that all areas have been processed as single-color rectangular blocks or marked rectangular blocks.

  [Compression Mode Determination] Next, the operation of the compression mode determination unit 13 will be described with reference to the flowchart of FIG. When the rectangular block extraction is completed by the rectangular block extraction unit 12 (S21), the compression mode determination unit 13 counts the number of colors Nc of each rectangular block (S22). When the number of colors Nc is 1, that is, when the rectangular block is a monochromatic rectangular block, the rectangular block is determined as a type A rectangular block (S23, S24). When the number of colors Nc is larger than 1 and smaller than the threshold th1 (YES in S25), the rectangular block is determined as a type B block (S26). When the number of colors Nc is equal to or greater than the threshold th1 (NO in S25), the edge ratio Eh in the horizontal and vertical directions of the rectangular block is calculated (S27). When the edge rate Eh in either the horizontal direction or the vertical direction is smaller than the threshold th2 (YES in S28), the rectangular block is determined as a type C block (S29). When the edge rate Eh is equal to or greater than the threshold th2 (NO in S28), the rectangular block is determined as a type D block (S30).

  An example of the calculation of the edge rate in S27 will be described with reference to FIG. The image in the rectangular block is divided into 3 × 3 pixel groups, and it is determined whether each of the nine pixels in the 3 × 3 pixel group belongs to an edge or not. The determination can be made based on the difference between the color value of each pixel and the color value of an adjacent pixel. If the number of pixels belonging to the edge is larger than the set number (for example, 3), the entire 3 × 3 pixel group is determined as an edge region, and if it is less than the predetermined number, the entire 3 × 3 pixel group is determined as a non-edge. It is determined as an area. In this way, the above determination is performed for all the 3 × 3 pixel groups in the rectangular block, and the ratio of the 3 × 3 pixel groups determined to be edge regions is calculated as the edge rate Eh.

  [Compression encoding processing] When classification into type A to D rectangular blocks is performed in this way, compression encoding processing suitable for the type is performed on each block. The lossy encoding unit 14 performs discrete cosine transform (DCT) used in, for example, the JPEG baseline method, quantizes the DCT coefficient, performs entropy encoding, and performs lossy encoding. The image of the C rectangular block is compression encoded. In the type C rectangular block, the number of colors Nc is greater than or equal to the threshold th1, but the edge rate Eh is less than the threshold th2. Therefore, even if the high frequency component is reduced by DCT or the like, the degradation in image quality is small.

  The palette compressing unit 17 compresses and encodes the image data of the type B rectangular block using a known palette compression. Since the type B rectangular block is, for example, a block having a color number Nc of less than 8, a high compression rate can be obtained.

  [Important Color Number Prediction] Refer to FIG. 7 for an example of the prediction important color number prediction procedure executed by the important color number prediction unit 15 as a pre-stage of the color reduction processing by the color reduction processing unit 16 for the type D rectangular block. To explain. The important color number prediction unit 15 in this example uses a look-up table (not shown) indicating the correspondence (FIG. 7) between the color number Nc of the colors actually included in the rectangular block and the predicted important color number Nic. The number of predicted important colors Nic is determined with reference to this. The basic concept of predicting the number of important colors is that when the number of colors Nc actually included in the rectangular block is large, the ratio of color reduction (the number of colors after color reduction processing is actually included before color reduction processing) The ratio of the number of colors to the number of colors) can be increased. Therefore, in this example, as shown in FIG. 7, when the number of colors Nc is greater than or equal to the threshold th1 and less than N2, the graph slope k1 is increased to reduce the color reduction ratio. On the other hand, when the number of colors Nc is greater than or equal to N2 and less than N3, the slope k2 is made smaller than k1 to slightly increase the rate of color reduction. This is because, since the number of colors Nc is large, it is considered that the characteristics of the original image are hardly lost even if the color reduction ratio is increased. If the number of colors Nc is equal to or greater than N3, the gradient is set to zero and an upper limit is set for the predicted important number of colors Nic. When the number of colors Nc is less than the threshold th1, it is determined as a type A or type B rectangular block, and thus is not a target for predicting the number of important colors. In FIG. 7, the graph is shown as a polygonal line, but the graph is not limited to this, and may be a stepped graph or a graph in which the slope changes to a natural curve.

  [Color Reduction Processing] Next, the procedure of color reduction processing by the color reduction processing unit 16 will be described with reference to the flowchart of FIG. When the predicted important color number Nic is determined (S31), the color reduction processing unit 16 first creates a histogram (FIG. 9) that shows the number of pixels of each color actually included in the type D rectangular block on the vertical axis. Then, this histogram is divided into a number of color interval cells equal to the number of predicted important color numbers Nic output (S32). Here, it is assumed that the number of predicted important colors Nic output is 3, and the cells are divided into three color interval cells A, C, and D. First, based on the characteristics of the shape of the histogram, by dividing into A cell and B cell, and further dividing B cell into two, C cell and D cell, a number equal to the predicted important color number Nic (three ) Color interval cells A, C, and D.

  Next, the color interval cells A, C, and D are arranged in descending order of the number of pixels, and cell numbers are assigned (S33). Here, since the number of pixels increases in the order of A, C, and D, {A, C, D} = (0, 1, 2) and a cell number are assigned (FIG. 10). By performing this alignment, the relevance of a plurality of rectangular blocks can be increased, and the compression rate can be further improved in the block grouping process to be performed later.

  Subsequently, the average value of the colors in the color section cells A, C, and D is calculated (S34). Pallet data that associates the cell number with the calculated average color values of the color section cells A, C, and D is created, and the color value of each pixel is converted based on this palette data (S35). Then, a color reduction process is performed. Since the format-reduced type D rectangular block data and the palette-compressed type B rectangular block data have the same format, they are grouped by the block grouping unit 18 in the next stage. In the histogram of FIG. 9, the color section is described in one dimension, but the same algorithm can be applied to the case where the color section is expressed by a multidimensional color section such as an RGB section.

  [Block Grouping] Next, block grouping executed by the block grouping unit 18 will be described with reference to FIGS. 11 and 12. The block grouping referred to here is a rectangle having substantially the same characteristics such as the number of colors in the type B or D rectangular blocks subjected to the color reduction processing by the color reduction processing unit 16 or the palette compression by the palette compression unit 17. Data is compressed by bundling block data. For example, as shown in FIG. 11, two-color rectangular blocks 1 to 3 can be bundled into one group, and four-color rectangular blocks 1 and 2 can be bundled into one group.

  FIG. 12 shows an example of dividing a rectangular block into four groups (G1 to G4) by block grouping. Since the two-color rectangular block can be displayed at 1 bit / pixel, it is divided into one group G1. Rectangular blocks with 3 to N1 colors are divided into group G2, rectangular blocks with N1 + 1 to N2 colors are divided into group G3, and rectangular blocks with N2 + 1 or more colors are divided into group G4. For example, if N1 = 4 and N2 = 16, the group G2 can display data at 2 bits / pixel and the group G3 can display data at 4 bits / pixel.

  As described above, the palette data obtained by palette compression of the type B rectangular block is arranged in descending order of the number of pixels corresponding to the color, and the type D rectangular block is the same. For this reason, since the type B rectangular block and the type D rectangular block take data of the same format, they can be classified into the same group as long as the color values are common. By performing this alignment, colors with a large number of pixels can be bundled in the same palette, and the compression rate can be increased. The type A rectangular block and the type C rectangular block are not subject to grouping by the block grouping unit 20, but are classified into groups G0 and G5, respectively (see FIG. 12). In FIG. 12, it has been described that the rectangular blocks are divided into four groups in the block grouping. However, for example, it is possible to divide the groups into smaller groups, or to divide the groups into more groups.

  [Data Shuffle] Next, the data shuffling process by the data shuffle unit 20 will be described with reference to FIG. Here, a data structure is used in which the data of the groups G0 to G5 grouped by the block grouping unit 20 and the like are sequentially arranged. The groups G0 to G5 include the size of each rectangular block, the position in the image, the palette row, the compressed data, and the like in addition to the sorting mac and the number of rectangular blocks (Num).

  The group G0 ((b) in FIG. 13) is a group of type A rectangular blocks which are monochromatic rectangular blocks. The “color column” is a data column indicating color information of each type A rectangular block. The number of data in the color column is equal to the number of rectangular blocks Num. The group G1 ((c) in FIG. 13) is a group of two-color rectangular blocks as a result of color reduction processing or palette compression of a type B or D rectangular block. The compressed size and compressed data in (c) of FIG. The palletized data of all rectangular blocks in the group G1 are bundled and compressed. The groups G2 to G4 are groups of 3 to N1 color rectangular blocks, N1 + 1 to N2 color rectangular blocks, and N2 + 1 or more color rectangular blocks as a result of color reduction processing or palette compression of the type B or D rectangular blocks. The data structure is as shown in (d) of FIG. G2 to G4 have color number data for each rectangular block in order to specify the number of colors for each rectangular block.

  G5 is a group of rectangular blocks as a result of lossy encoding performed on a type C rectangular block. Since the number of colors varies from block to block, the rectangular block data is not grouped and compressed, but compressed for each rectangular block data.

  In addition, although the example which arranges data in the order of groups G0 to G5 was demonstrated in FIG. 13, it is not restricted to this, It is possible to arrange data in various orders. In addition, the arrangement order of the data items in each group G0 to G5 can be variously changed.

  Second Embodiment Next, an image compression apparatus according to a second embodiment of the present invention will be described with reference to FIG. Constituent elements similar to those of the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted. In this embodiment, in order to more accurately discriminate the type C or D rectangular block, the operation of the compression mode determination unit 13 is different and the compression mode addition determination unit 13A is provided. This is different from the embodiment.

  In the first embodiment, the type C rectangular block and the type D rectangular block are classified by determining whether or not the edge rate Eh is larger than the threshold th2. However, how large the threshold th2 is to be set is a difficult problem. In the case of an edge rate in the vicinity of the threshold th2, it is difficult to determine whether the color reduction process is appropriate or the lossy encoding process is appropriate. It can be difficult.

  Therefore, in the second embodiment, in the compression mode determination unit 13, for the rectangular block whose threshold is th2 ± α near th2, the processing by the lossy encoding unit 14, the important color number prediction unit 15, and Both the processing by the color reduction processing 16 are performed. Which of the two types of processing provides the higher compression rate is determined by the compression mode addition determination unit 13A, and it is determined whether it is appropriate to determine the rectangular block as type C or D. That is, the compression rate of the processing result (1) by the lossy encoding unit 14 and the processing result (2) encoded by the lossless encoding unit 19B after the processing by the color reduction processing unit 16 are compared. If it is a rate, the processing result (1) is output as it is. When the latter is a higher compression rate, the processing result (3) is the result of the block grouping unit 18 grouping the processing results by the color reduction processing unit 16 and lossless encoding by the lossless encoding unit 19B. Is output.

  [Third Embodiment] Next, an image compression apparatus according to a third embodiment of the present invention will be described with reference to FIG. Constituent elements similar to those of the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted. In this embodiment, from the rectangular blocks subjected to the color reduction processing by the color reduction processing unit 16, a finer single color rectangular block and a small number of rectangular blocks are extracted by the rectangular block extraction unit 12A, and the type of the extracted rectangular block This is different from the first embodiment in that (single color, two or more colors) is determined and classified by the compression mode determination unit 13B. When a color reduction process is performed on a type D rectangular block, it is expected that a single color area and a small color area in the rectangular block will increase as the number of colors in the rectangular block decreases. Further improvement in compression rate can be expected. Since the operation of the rectangular block extraction unit 12A is substantially the same as that of the rectangular block extraction unit 12, a detailed description thereof will be omitted.

  The compression mode determination unit 13B counts the number of colors Nc in the extracted rectangular block (S131). When Nc = 1 (YES in S132), the compression mode determination unit 13B determines that the rectangular block is a type A rectangular block, and the data Output to the shuffle unit 20. If Nc> 1 (NO in S132), the rectangular block is determined to be a type B rectangular block, and the palette compression unit 17 performs palette compression together with the rectangular block determined to be the type B rectangular block by the compression mode determination unit 13. set to target.

  Although the embodiments of the invention have been described above, the present invention is not limited to these embodiments, and various modifications and additions can be made without departing from the spirit of the invention. For example, in the third embodiment (FIG. 15), instead of separately providing the rectangular block extraction unit 12A compression mode determination unit 13B, the output of the color reduction processing unit 16 is fed back to the rectangular block extraction unit 12, and the rectangular block extraction unit 12 and 13 may also serve as the roles of 12A and 13B.

It is a block diagram which shows the structure of the image compression apparatus which concerns on 1st embodiment of this invention. 2 shows an example of a computer screen that is a processing target of the image compression apparatus in FIG. 1. 5 is a flowchart for explaining the operation of a rectangular block extraction unit 12. A method for expanding a monochrome rectangular block will be described. A method for expanding a monochrome rectangular block will be described. A method for expanding a monochrome rectangular block will be described. A method for expanding a monochrome rectangular block will be described. 5 is a flowchart for explaining the operation of a compression mode determination unit 13. An example of the edge rate determination method in the flowchart of FIG. 5 will be described. A method for determining the number of predicted important colors will be described. It is a flowchart which shows the procedure of the color reduction process by the color reduction process part 16 of FIG. A specific example of S32 in the flowchart of FIG. 8 will be described. A specific example of S33 in the flowchart of FIG. 8 will be described. FIG. 5 is a conceptual diagram showing the operation of the block grouping unit 18. An example of a grouping method in the block grouping unit 18 will be described. An example of a data shuffling process in the data shuffle unit 20 will be described. It is a block diagram which shows the structure of the image compression apparatus which concerns on 2nd embodiment of this invention. It is a block diagram which shows the structure of the image compression apparatus which concerns on 3rd embodiment of this invention. The operation of the compression mode determination unit 13B in FIG. 15 will be described.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Color image input device, 12 ... Rectangular block extraction part, 13 ... Compression mode determination part, 14 ... Lossy encoding part, 15 ... Important color number prediction part, 16 ... -Color reduction processing unit, 17 ... palette compression unit, 18 ... block grouping unit, 19 ... lossless encoding unit, 20 ... data shuffle unit.

Claims (14)

A rectangular block extraction unit that divides image data based on colors and extracts a plurality of rectangular blocks;
The number of colors indicating the number of colors included in each of the rectangular blocks and the edge rate indicating the ratio of edge pixels are determined, the rectangular block having the single number of colors is designated as a first type rectangular block, and the number of colors is The rectangular block that is less than a predetermined value is a second type rectangular block, the number of colors is equal to or greater than a predetermined value, and the rectangular block that is less than a predetermined value is a third type rectangular block, and the number of colors is a predetermined value A determination unit that classifies the rectangular blocks having the edge ratio equal to or higher than a predetermined value as fourth type rectangular blocks,
A palette compression unit that applies palette compression to the image data of the second type rectangular block;
An irreversible compression encoding unit that performs irreversible data compression on the image data of the third type rectangular block;
A color reduction processing unit that predicts the number of important colors based on the number of colors of the fourth type rectangular block and performs a color reduction process to reduce the number of colors included in the fourth type rectangular block to the predicted number of important colors. An image compression apparatus characterized by the above.   The rectangular block extraction unit searches the image data for a single-color rectangular block having a predetermined size, and expands the single-color rectangular block when adjacent rows or adjacent columns have the same color. The image compression apparatus according to claim 1, wherein the first type rectangular block is maximized by doing so.   The image compression apparatus according to claim 1, further comprising a grouping unit that groups the compressed images compressed by the palette compression unit or the color reduction processing unit according to the number of colors included.   The image compression apparatus according to claim 1, wherein the lossy compression coding unit performs data compression by transform coding that transforms image data into a spatial frequency domain using orthogonal transform.   The determination unit compares the compression rate by the lossy compression encoding unit and the compression rate by the color reduction processing unit when the edge rate of a rectangular block approximates the predetermined value, and the former is the latter 2. The image compression apparatus according to claim 1, wherein when it is larger, the rectangular block is classified as a third type rectangular block, while when the latter is larger than the former, the rectangular block is classified as a fourth type rectangular block.   The rectangular block extraction unit further classifies the rectangular blocks after the color reduction processing by the color reduction processing unit into fine rectangular blocks, and the compression mode determination unit converts the fine rectangular blocks into the first type to the fourth type. The image compression apparatus according to claim 1, wherein the image compression apparatus is classified into rectangular blocks. Dividing the image data based on color to extract a plurality of rectangular blocks;
Determining the number of colors indicating the number of colors included in each of the rectangular blocks, and the edge rate indicating the ratio of edge pixels;
The rectangular block having the singular color is a first type rectangular block, the rectangular block having the number of colors less than a predetermined value is the second type rectangular block, the number of colors is equal to or greater than a predetermined value, and the edge rate is Classifying the rectangular block having a value less than a predetermined value as a third type rectangular block, and classifying the rectangular block having the number of colors equal to or greater than a predetermined value and the edge rate as a fourth value as a fourth type rectangular block;
Applying palette compression to the image data of the second type rectangular block;
Performing irreversible data compression on the image data of the third type rectangular block;
Applying a color reduction process for predicting the number of important colors based on the number of colors of the fourth type rectangular block and reducing the number of colors included in the fourth type rectangular block to the predicted number of important colors. Image compression method.   In the step of dividing the image data and extracting it as a plurality of rectangular blocks, a single-color rectangular block having a predetermined size is searched in the image data, and the color of the adjacent row or adjacent column of the searched single-color rectangular block is the same. 8. The image compression method according to claim 7, wherein the first-type rectangular block is maximized by expanding the single-color rectangular block in some cases.   The image compression method according to claim 7, further comprising a step of grouping the compressed images compressed by the palette compression or the color reduction processing according to the number of colors included.   The classifying step compares the compression rate by the irreversible data compression and the compression rate by the color reduction processing when the edge rate of a rectangular block is a value approximate to the predetermined value, and the former is the latter 8. The image compression method according to claim 7, wherein when the size is larger, the rectangular block is classified as a third type rectangular block, while when the latter is larger than the former, the rectangular block is classified as a fourth type rectangular block.   8. The method according to claim 7, further comprising the step of classifying the rectangular blocks after the color reduction processing by the color reduction processing into fine rectangular blocks, and classifying the small rectangular blocks into the first type to fourth type rectangular blocks. Image compression method. Dividing the image data based on color to extract a plurality of rectangular blocks;
Determining the number of colors indicating the number of colors included in each of the rectangular blocks, and the edge rate indicating the ratio of edge pixels;
The rectangular block having the singular color is a first type rectangular block, the rectangular block having the number of colors less than a predetermined value is the second type rectangular block, the number of colors is equal to or greater than a predetermined value, and the edge rate is Classifying the rectangular block having a value less than a predetermined value as a third type rectangular block, and classifying the rectangular block having the number of colors equal to or greater than a predetermined value and the edge rate as a fourth value as a fourth type rectangular block;
Applying palette compression to the image data of the second type rectangular block;
Performing irreversible data compression on the image data of the third type rectangular block;
Causing the computer to execute a color reduction process for predicting the number of important colors based on the number of colors of the fourth type rectangular block and reducing the number of colors included in the fourth type rectangular block to the predicted number of important colors. An image compression program configured in A rectangular block extraction unit that divides image data based on colors and extracts a plurality of rectangular blocks;
The number of colors indicating the number of colors included in each of the rectangular blocks is obtained, and the rectangular block having a single number of colors is defined as a first rectangular block, and the rectangular block having a plurality of colors is defined as a second rectangular block. A determination unit for classification;
For the second rectangular block, an image compression unit that selectively performs an image compression method related to the number of colors and the edge rate of the image;
An image compression apparatus comprising: The said image compression part has a means to obtain | require the compression rate by a several image compression method regarding a 2nd rectangular block, and to select the image compression method with the highest compression rate in the said rectangular block, The said compression unit is characterized by the above-mentioned. Image compression device.

Images