JP2008167309A - Image processing apparatus and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the coding information quantity while suppressing deterioration of image quality by integrating adjacent BTC-compressed blocks in accordance with an appropriate criterion. <P>SOLUTION: BTC compression is applied to an image, it is decided whether to integrate adjacent compressed blocks A and B on the basis of similarity of the blocks A and B, based on the block A and similarity based on the block B and when decided to integrate the blocks, these blocks are integrated into one block U. Furthermore, when the block size of the candidates is large, the decision criterion is relaxed. Moreover, when a difference S between maximum density and minimum density in the blocks A and B is below a specified value, the blocks are integrated. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing method for performing BTC compression on an image.

  One of the image compression (encoding) techniques is BTC (Block Truncation Coding) compression suitable for multi-gradation image compression. In BTC compression, an image is divided into blocks each having a predetermined number of pixels (for example, 4 pixels × 4 pixels), the density of each pixel in the block is changed to statistical information indicating the density distribution in the block, and the block statistical information. Is encoded so as to be expressed by relative information of the density of each pixel with respect to. For example, the statistical information is composed of the maximum / minimum difference and the maximum / minimum intermediate value, or the average value and the standard deviation, and the relative information for each pixel has a density range corresponding to the maximum / minimum difference or the standard deviation with a predetermined small gradation. It is composed of the gradation value of each pixel when represented by a number (usually, the gradation number represented by 1 bit or 2 bits).

  When such BTC compression is performed with a uniform block size regardless of image characteristics, it is necessary to reduce the block size in order to ensure restoration accuracy in non-flat portions such as edge portions and gradation portions. Therefore, the amount of encoded information cannot be made sufficiently small. Conversely, if the block size is increased in order to reduce the amount of encoded information (increase the compression rate), the image quality is sacrificed.

  As a technique for reducing the amount of encoded information by suppressing the deterioration of image quality, if the density of adjacent blocks is considered to be the same, these blocks are integrated, and the final block is obtained. A method of encoding each pixel as one pixel has been proposed (see, for example, Patent Document 1).

Japanese Patent Laid-Open No. 10-98621

  When the density of adjacent blocks can be regarded as the same, the amount of encoded information can be reduced by integrating these blocks. However, if the density of adjacent blocks cannot be considered to be the same, the image quality deteriorates due to the effect of a difference between the flat part and the non-flat part during restoration. .

  The present invention is intended to solve the above-described problem, and an image processing apparatus capable of reducing the amount of encoded information while suppressing deterioration in image quality by integrating adjacent blocks according to an appropriate criterion, and The object is to provide an image processing method.

  The gist of the present invention for achieving the object lies in the inventions of the following items.

[1] After the image is divided into a plurality of blocks, BTC compression is performed in units of blocks, and whether or not the first block and the second block adjacent to each other after the compression are integrated is determined based on the first block and the second block. , Based on both the similarity when the first block is used as a reference and the similarity when the second block is used as a reference. An image processing apparatus, wherein the image is encoded by integrating the blocks into one block.

  In the said invention, the similarity of the adjacent block after BTC compression is mutually judged from both blocks, and a block is integrated based on these judgment results. By determining from both sides, the similarity between blocks can be determined more accurately, and the amount of encoded information can be reduced by integrating the blocks so that the image quality is not affected.

[2] The image processing apparatus according to [1], wherein the similarity is a degree of overlap between the density distribution range of the first block and the density distribution range of the second block.

  In the above invention, the degree of similarity is determined based on how much the density distribution range of pixels belonging to each block overlaps between adjacent blocks.

[3] The reference value for determining the similarity based on the first block and the reference value for determining the similarity based on the second block are the same [1] or The image processing device according to [2].

  In the above invention, by determining the similarity based on the same criterion from both sides, it is possible to maintain the same image quality after integration in both blocks.

[4] After the image is divided into a plurality of blocks, BTC compression is performed in units of blocks, and whether or not the first block and the second block adjacent to each other after the compression are integrated is determined by the first block and the second block. Is determined on the basis of the difference between the maximum density and the minimum density, and when it is determined to be integrated, the first block and the second block are integrated into one block, whereby the image encoding process is performed. An image processing device characterized in that:

  In the above invention, if the difference between the maximum density and the minimum density in all the adjacent blocks of the integration candidate is small, the adjacent blocks of the integration candidate are in a flat portion and the image quality deterioration due to the integration is small. As it is estimated, integrate.

[5] After the image is divided into a plurality of blocks, BTC compression is performed on a block basis, and it is determined whether or not the first block and the second block adjacent to each other after the compression are integrated. The image is encoded by integrating the first block and the second block into one block, and
An image processing apparatus, wherein the reference value for the determination is changed according to a block size of either or both of the first block and the second block.

  In the above invention, when the block size is large, integration has been performed several times with neighboring blocks so far, and there is a high possibility that the neighborhood of the block is a flat part. Therefore, the criteria for integration will be relaxed to facilitate further integration.

[6] Character direction in the image is detected, and the priority of the arrangement direction of the first block and the second block to be determined is changed according to the detected character direction. The image processing apparatus according to any one of [1] to [5].

  In the above invention, the character often has a thin line in a specific direction. Therefore, in order to ensure the reproducibility of the thin line, the priority of the arrangement direction of adjacent blocks to be integrated candidates is changed according to the character direction in the image.

[7] The configuration data of each block after the encoding process is made into a one-dimensional or two-dimensional array of binary information,
A one-dimensional or two-dimensional array for each block is two-dimensionally arranged corresponding to the original block array in the image to form a binary image,
The image processing apparatus according to any one of [1] to [6], wherein the generated binary image is reversibly compressed.

  In the above invention, the BTC compression is performed, and the configuration data of each block obtained by integrating the adjacent blocks satisfying the integration condition is expanded into a two-dimensional image of binary information, and this is further reversibly compressed, thereby encoding. The amount of information can be reduced efficiently.

[8] After dividing the image into a plurality of blocks, BTC compression is performed in units of blocks, and whether or not the first block and the second block adjacent to each other after the compression are integrated is determined by the first block and the second block. , Based on both the similarity when the first block is used as a reference and the similarity when the second block is used as a reference. An image processing method, wherein the image is encoded by integrating the blocks into one block.

[9] The image processing method according to [8], wherein the similarity is a degree of overlap between the density distribution range of the first block and the density distribution range of the second block.

[10] The reference value for determining the similarity based on the first block is the same as the reference value for determining the similarity based on the second block [8] or The image processing method according to [9].

[11] After the image is divided into a plurality of blocks, BTC compression is performed on a block-by-block basis, and whether the first block and the second block adjacent to each other after the compression are integrated is determined by the first block and the second block. Is determined on the basis of the difference between the maximum density and the minimum density, and when it is determined to be integrated, the first block and the second block are integrated into one block, whereby the image encoding process is performed. An image processing method characterized in that:

[12] After the image is divided into a plurality of blocks, BTC compression is performed in units of blocks, and it is determined whether or not the first block and the second block adjacent to each other after the compression are integrated. The image is encoded by integrating the first block and the second block into one block, and
The image processing method according to claim 1, wherein the reference value for the determination is changed according to a block size of one or both of the first block and the second block.

[13] Characteristic direction in the image is detected, and priority of the arrangement direction of the first block and the second block to be determined is changed according to the detected character direction. The image processing method according to any one of [8] to [12].

[14] The configuration data of each block after the encoding process is made into a one-dimensional or two-dimensional array of binary information,
A one-dimensional or two-dimensional array for each block is two-dimensionally arranged corresponding to the original block array in the image to form a binary image,
The image processing method according to any one of [8] to [13], wherein the generated binary image is reversibly compressed.

  According to the image processing apparatus and the image processing method of the present invention, adjacent blocks after BTC compression are integrated according to an appropriate determination criterion, so that it is possible to reduce the amount of encoded information while suppressing deterioration in image quality.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 shows a system configuration of an image processing apparatus 10 according to an embodiment of the present invention. The image processing apparatus 10 is configured as, for example, a digital multifunction machine having a copy function, a facsimile function, a printer function, and the like. The image processing apparatus 10 includes a data control unit 11 that performs various operations on the overall operation of the apparatus and image data. The data control unit 11 includes an image display unit 12, an operation unit 13, an image input unit 14, and an image output unit. 15, a data holding unit 16 and a data communication I / F unit 17 are connected.

  The data control unit 11 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), and the like as main parts, and the CPU executes a program stored in the ROM to generate an image. Various functions as the processing apparatus 10 are realized. Further, the data control unit 11 functions as an image processing unit 20 that performs encoding processing of image data. The function of the image processing unit 20 may be realized by the CPU executing a program, or may be realized by dedicated hardware or a combination of dedicated hardware and software processing by the CPU.

  The image display unit 12 includes a liquid crystal display and displays various guidance screens and operation screens. The operation unit 13 includes a touch panel and switches laid on the liquid crystal display, and fulfills a function of receiving various setting operations and instruction operations from the user.

  The image input unit 14 performs a function of optically reading a document and capturing a corresponding image. Here, an original is read by a color image sensor, and multi-tone color digital image data including red (R) image data, green (G) image data, and blue (B) image data is output. It is like that.

  The image output unit 15 has a function of printing an image on a sheet and outputting the image. For example, a laser printer is used.

  The data holding unit 16 functions to hold image data and other data, and is configured by a storage device such as a semiconductor memory or a hard disk device. The data communication I / F unit 17 functions to connect the image processing apparatus 10 to a data transmission network 5 such as a LAN (Local Area Network) or the Internet and transmit / receive various data.

  FIG. 2 shows a functional configuration of the image processing unit 20. The image processing unit 20 includes a primary compression unit 21 including a block division unit 21a, a maximum / minimum value detection unit 21b, and a compression code calculation unit 21c, an adjacent block comparison unit 22, a block combination unit 23, and a character detection unit. 24 and a secondary compression unit 25.

  The block dividing unit 21a functions to divide an image to be processed into blocks of a predetermined size (for example, 4 pixels × 4 pixels). The maximum / minimum value detection unit 21b detects the maximum value and the minimum value of the pixel density in each block. The compression code calculation unit 21c functions to generate configuration data obtained by compressing and encoding image data in a block.

  The adjacent block comparison unit 22 has a function of comparing adjacent blocks and determining whether or not to integrate these blocks. The block combination unit 23 performs a function of integrating adjacent blocks determined to be integrated by the adjacent block comparison unit 22 into one.

  The character detection unit 24 detects the character direction in the image. The adjacent block comparison unit 22 changes the priority of the arrangement direction of adjacent blocks to be subjected to integration determination according to the character direction detected by the character detection unit 24. For example, the priority of which one of the block adjacent in the vertical direction and the block adjacent in the horizontal direction is preferentially compared is changed.

  The secondary compression unit 25 converts the configuration data of each block generated by the compression code calculation unit 21c into a one-dimensional or two-dimensional array of binary information, and converts the one-dimensional or two-dimensional array for each block into the current processing target. A binary image is constructed by two-dimensionally arranging it in correspondence with the original block arrangement in the image, and this binary image is reversibly compressed. The lossless compression method may be appropriate, for example, JBIG (Joint Bi-level Image experts Group) method.

  FIG. 3 shows a schematic flow of the entire processing performed by the image processing unit 20. The image processing unit 20 performs an encoding process based on BTC compression on the input image data to be processed (step S101), and the configuration data of each block after the encoding process is as described above. A two-dimensional binary image is developed (step S102), the binary image is losslessly compressed (step S103), and final encoded information is output.

  FIG. 4 shows the encoding process in step S101 of FIG. 3 in more detail. Here, the input image data is color image data composed of R color, G color, and B color image data. In each color, one pixel is represented by 8 bits and 256 gradations. The image processing unit 20 inputs the image data of the image to be processed (step S111), divides it into a default block size (here, 4 pixels × 4 pixels), and performs BTC compression (step S112).

  Next, the integration of adjacent blocks after BTC compression with the default block size is determined. Specifically, it is determined whether or not there is an integration candidate block (step S113). If there is an integration candidate block remaining (step S113; Y), a set of integration candidate blocks is selected from them (step S113; Y) (step S113; Y). Step S114). Selection of a set of integration candidate blocks is performed with a priority according to the character direction detected by the character detection unit 24.

  Here, when the character is in the horizontal direction as shown in FIG. 5A, priority is given to the integration of the blocks in the horizontal direction, and when the character is in the vertical direction as shown in FIG. Prioritize block integration. This is because the horizontal line is often a thin line in the character, and the priority in the same direction as the character arrangement is increased as the block integration direction.

  Next, the similarity between the selected set of integrated candidate blocks is evaluated based on the pixel density distribution (pixel density data distribution) and the like (FIG. 4: step S115). As a result of the evaluation, when these one set of integration candidate blocks satisfy a predetermined integration criterion (step S116; Y), the one set of integration candidate blocks is integrated (step S117), and the process returns to step S113 to be integrated. If the standard is not satisfied (step S116; N), the process returns to step S113 without integration. When there are no more integration candidate blocks (step S113; N), this process ends (encoding is completed).

FIG. 6 shows the processing content of one block encoding processing in BTC compression. Image data (original image information) corresponding to a rectangular area 41 of 4 pixels × 4 pixels forming one block (area with hatching in the figure) is R color image data (R ₁₁ ) for 4 pixels × 4 pixels. To R ₄₄ ), 4 pixels × 4 pixels of G color image data (G _{11 to} G ₄₄ ), and 4 pixels × 4 pixels of B color image data (B _{11 to} B ₄₄ ). Each pixel of each color is represented by 8 bits and 256 gradations, and the total amount of information of one block before encoding is 384 bits.

The BTC compression is performed individually for each of the R color image data, the G color image data, and the B color image data. For example, when RTC image data for one block is subjected to BTC compression, first, the maximum value (max _R ) and the minimum value (min _R ) of the pixel density in the block for the R color are obtained. Then, the maximum value (max _R) minimum and maximum minimum gap LD _R which is a difference between (min _R), the maximum value (max _R) and the minimum value (min _R) and maximum and minimum intermediate value is an intermediate value of and LA _R, and block statistics indicating the tendency of the entire block of data distribution (density distribution). Maximum Minimum gap LD _R, respectively maximum and minimum intermediate value LA _R is represented by 8 bits is the same number of bits as the original pixel.

The pixel density and the minimum value (min _R) the integer part of those four times divided by the maximum and minimum disparity LD _R the difference between the relative information of the individual pixels, finding it for each pixel in the block. Note that 4 times is used when the relative information of the individual pixels is expressed by 2 bits, and how many times the relative information is expressed depends on how many bits the relative information of the individual pixels is expressed. For example, when it is represented by 1 bit, it is doubled and when it is represented by 3 bits, it is 8 times.

The BTC-compressed configuration data relating to the R color image data includes the block statistical information (maximum / minimum difference LD _R , maximum / minimum intermediate value LA _R ) and relative information (φ _{R11 to} φ) for each pixel included in the block. _R44 ). Maximum minimum disparity LD _R, respectively 8 bits maximum minimum intermediate value LA _R, expressed relative information of the individual pixels in each pixel two bits, the information amount per one color becomes 48 bits, R color, G color The amount of encoded information for one block of a color image combining the B colors is 144 bits.

  FIG. 7 is a compression rate comparison table 50 illustrating the correspondence between block size, encoded information amount, and compression rate. The maximum / minimum difference LD and the maximum / minimum intermediate value LA are each represented by 8 bits, and the relative information of the individual pixels is represented by 2 bits for each pixel. When the block size is 4 pixels × 4 pixels, as described above, the information amount of the original data (A) is 384 bits, the encoded information amount (B) after BTC compression is 144 bits, and the compression rate (B / A) is 37.5%. The compression rate is 31.3% when the block size is 8 pixels.times.4 pixels, the compression rate is 29.2% when the block size is 12 pixels.times.4 pixels, and the compression rate is when the block size is 8 pixels.times.8 pixels. The rate is 28.1%. Thus, in BTC compression, the compression rate increases as the block size increases.

  FIG. 8 shows a process of integrating adjacent blocks A and B of 4 pixels × 4 pixels into one block U. The configuration data of block A is composed of block statistical information As and relative information Aφ of individual pixels for 16 pixels, and the configuration data of block B is composed of block statistical information Bs and relative information Bφ of individual pixels for 16 pixels. Become. The configuration data of the block U obtained by integrating these into one (hereinafter referred to as an integrated block) includes block statistical information Us for one block and relative information Uφ for individual pixels for 32 pixels. Therefore, the block statistical information As and Bs in each of the two blocks A and B are shared by the block statistical information Us by the block integration, the amount of encoded information is reduced, and the compression rate is improved.

  For example, when the block statistical information of the constituent data of each color of R, G, and B of each block A, B is composed of an 8-bit maximum / minimum difference LD and an 8-bit maximum / minimum intermediate value LA, as shown in FIG. The integration of blocks reduces the amount of encoded information by 48 bits (16 bits each for R, G, and B colors).

  FIG. 9 shows the flow of “one set of block integration processing” shown in step S117 of FIG. It is checked whether or not the data distribution range (concentration distribution range) of the integration target blocks is in a conjugated relationship (step S121). Note that the conjugation relationship refers to a case where the data distribution range of one block (the range from the maximum value to the minimum value of the pixel density in the block) includes the entire data distribution range of the other block.

  If they are in a conjugation relationship (step S121; Y), the block statistical information for conjugation is directly used as block statistical information for the integrated block (step S122). In addition, the relative information of the individual pixels for all the pixels of the integrated block is used as it is before the integration for the relative information of the individual pixels of the conjugated block, and the relative information of the individual pixels of the conjugated block is used. The information is calculated and obtained based on the integrated block statistical information (step S123), and the process ends (end).

  If not in the conjugation relationship (step S121; N), the data distribution range (maximum value and minimum value of the range) of the integrated block is determined. That is, the smaller value among the minimum values of both blocks to be integrated is set as the minimum value of the integrated block density, and the larger value among the maximum values of both blocks is set as the maximum value of the integrated block density ( Step S124).

  Based on these maximum and minimum values, block statistical information of the integrated block is calculated. For example, the difference between the maximum value and the minimum value of the integrated block density is set as the maximum / minimum difference, and the intermediate value between the maximum value and the minimum value of the integrated block density is set as the maximum / minimum intermediate value (step S125). Further, based on the block statistical information of the integrated block obtained above, the relative information of the individual pixels is calculated for all the pixels in the integrated block (step S126), and the process is ended (END).

  By repeating such block integration, a large block is formed in a flat portion where the density change is small.

  FIG. 10 illustrates a state in which the block size is expanded by repeating block integration. In this example, the block A and the block B adjacent in the horizontal direction are first integrated into the integrated block U1, and the integrated block U1 and the block C adjacent in the vertical direction are further integrated into the integrated block U2, and further integrated. Block D adjacent to block U2 is integrated into integrated block U3.

  In the present embodiment, as a limit value when integrating as one block, the maximum integration number in the horizontal direction, the maximum integration number in the vertical direction, or the maximum integration number regardless of vertical and horizontal are set in advance. Integration is repeated within a range not exceeding this limit value. FIG. 11 shows an example of a block arrangement state in an image after integration. For example, the integration block U4 at the left corner is repeatedly integrated up to the limit value of 4 blocks, and the right adjacent block is newly integrated from one block, and as a result of the determination, it is not integrated. Yes.

Next, the criteria for determining the integration will be described.
FIG. 12 shows the relationship between the distribution state of the pixel density and the determination result of whether to integrate or not. 12 (a) it is, when a reference to the data distribution range D _B of data distribution range D _A and the block B is a density distribution range of the pixel belonging to the block A, the block A (when viewed from the block A) When the block B is used as a reference (when viewed from the block B), the block A and the block B are integrated into one integrated block U.

  FIG. 12B also shows a case where the overlapping degree when the block A is used as a reference and the overlapping degree when the block B is used as a reference are both large. Set to block U. In this example, block A (conjugate side) and block B (conjugate side) are in a conjugated relationship.

FIG. 12 (c), the data distribution range D _B of data distribution range D _A and the block B of the block A, and the degree of overlap when relative to the overlapping degree and the block B in the case relative to the block A This is a case where both are small and the difference S between the maximum value and the minimum value of the data (density) in the block A and the block B is large, and the block A and the block B are not integrated.

FIG. 12 (d) of the data distribution range D _B of data distribution range D _A and the block B of the block A, the degree of overlap when relative to the block A is small, the degree of overlap when relative to the block B Is large, and block A and block B are not integrated.

FIG. 12 (e) of the data distribution range D _B of data distribution range D _A and the block B of the block A, the degree of overlap when relative to the overlapping degree and the block B in the case relative to the block A Although both are small, the difference S between the maximum value and the minimum value of the data (density) in the block A and the block B, that is, the data distribution range including both the block A and the block B is small. And block B are integrated into one integrated block U.

  FIG. 13 shows the flow of the integrated determination process. This process corresponds to steps S115, S116, and S117 in FIG. First, a data distribution range (density distribution range) is specified or estimated for each of block A and block B that are integration candidates (step S131). Specifically, the data distribution ranges (maximum value and minimum value) of blocks A and B are specified or estimated from the block statistical information.

  For example, as shown in FIG. 14, when the block statistical information is represented by a maximum value and a minimum value, or a maximum minimum intermediate value and a maximum / minimum difference, or a maximum value and a maximum / minimum difference, block statistics The maximum and minimum values are identified from the information. On the other hand, the block statistical information is represented by an average value and a standard deviation value, or “average value−standard deviation” and “average value + standard deviation”, or a lower quartile and an upper quartile. If so, the maximum value and the minimum value are estimated from the block statistical information.

  Next, from the data distribution range (maximum value and minimum value) of each block A and B specified or estimated from the block statistical information, the maximum value and minimum value of the data (density) in both blocks A and B It is determined whether the difference, that is, the size of the total minimum data distribution range including both blocks A and B is equal to or smaller than a predetermined range size (step S132). Y), block integration is performed (step S137), and this process is terminated (END). When the above difference is small, the data distribution of both the integration candidate blocks A and B is in a narrow range, and thus the influence of the integration is difficult to occur. Therefore, they are integrated regardless of the degree of overlap.

  When the above difference exceeds the specified range size (step S132; N), the degree of overlap of the integration candidate blocks is obtained (step S133). That is, the overlapping degree OA (when viewed from the block A) with the block A as a reference and the overlapping degree OB with respect to the block B (when viewed from the block B) are calculated (step S133). Details of the calculation are shown in FIG.

  Next, it is determined whether or not the block size of at least one of the integration candidate blocks A and B is equal to or greater than a specified value (step S134). If the block size is equal to or greater than the specified value (step S134; Y), a reference value for integration determination described later. Is adjusted to be loose (step S135). That is, a block having a large block size is a block integrated with a neighboring block several times, and the integrated block has a high probability of being a flat portion of density. Therefore, if the integration candidate block size is large, the reference value is relaxed to facilitate integration with the next block.

  In this case, when the block serving as a reference for integration is block A, and the block integrated therein is block B, it is determined whether or not the block size on the block A side is equal to or larger than a specified value. In addition, the configuration may be such that the total or average block size of the block A and the block B is compared with a specified value.

  In the adjustment, for example, the reference value compared with the larger one of the overlapping degree OA when the block A is used as a reference and the overlapping degree OB when the block B is used as a reference is set as a reference 1, and When the reference value compared with the smaller one is set as the reference 2, both the reference 1 and the reference 2 are lowered from the standard value, or only one of them is lowered. When the standard value is 80%, both the reference 1 and the reference 2 are reduced to 60%, or the reference 2 is set to zero%.

  Next, based on the adjusted reference value, or when the block size of the integration candidate block is smaller than the specified value (step S134; N), the integration candidate blocks are integrated into one based on the reference value that is not adjusted. (Step S136; Y), block integration (step S137) ends this processing (end), and if it is determined not to integrate (step S136; Y) Step S136; N), the process is terminated without integration (end).

  For example, when the overlap degree OA when the block A is used as a reference and the overlap degree OB when the block B is used as a reference is larger than the reference 1, and the smaller OA and OB are more than the reference 2 It is determined to be integrated into Note that the reference value may be set arbitrarily. For example, both may be set to the same value, or may be different from each other, such as 50% for reference 1 and 75% for reference 2.

  Note that by making the reference values the same, it is possible to suppress degradation in image quality due to integration to the same level for both the integration side and the integration side blocks. Further, since it is comprehensively determined whether or not to integrate based on both the degree of overlap based on block A and the degree of overlap based on block B, image quality deterioration due to integration is suppressed for any block. As compared with the case where it is determined whether or not to integrate only one block, it is possible to maintain good image quality.

  FIG. 16 illustrates the state of the two-dimensional development performed in step S102 of FIG. The configuration data P shown in the upper part of the figure is configuration data for one block of a color image made up of R, G, and B colors, and the numerical values in [] are the bit positions of each data (for example, [1] Indicates bit 1), and [N: M] indicates data from the Nth bit to the Mth bit. In the following description, for n-bit data, LSB (Least Significant Bit) is defined as the 0th bit and MSB (Most Significant Bit) is defined as the (n-1) th bit.

In the first stage of the two-dimensional development process, the configuration data P is arranged so as to be a one-dimensional array J or a two-dimensional array K. In the two-dimensional array K, R color data is arranged in the upper stage Ka. Specifically, φ _R11 [0] ... φ _R44 [0] is _{placed in} the 4 × 4 area on the upper left, φ _R11 [1]… φ _R44 [1] is _placed in the upper 4 × 4 area, and the upper right LD _R [7]… LD _R [4], LD _R [3]… LD _R [0], LA _R [7]… LA _R [4], LA _R [3]… LA _R [0] It is arranged. Similarly, the G color data is arranged in the middle stage Kb, and the B color data is arranged in the lower stage Kc.

  In the second stage of the two-dimensional development process, as shown in the lower part of FIG. 16, the one-dimensional array J or the two-dimensional array K for each block obtained in the first stage is replaced with the original block array in the original image. The two-dimensional binary image V1 or V2 is configured in two dimensions. The secondary compression unit 25 compresses the binary image V1 or V2 by a lossless compression method such as JBIG, and outputs final encoded information.

  As described above, according to the encoding processing by the image processing unit 20, the average block size is increased by selectively integrating blocks in a flat portion of image data, and the amount of block statistical information is reduced. Therefore, the amount of encoded information can be suppressed without affecting the restoration accuracy of the non-flat portion. In addition, since the relative information of the individual pixels in the block is ensured even in the flat portion, the gentle gradation pattern can be restored naturally without unnaturally uniforming the data. Furthermore, since it is determined whether or not to integrate based on the degree of overlap calculated from the block statistical information, the calculation load related to the determination is small.

  In addition, since it is determined whether or not to integrate based on both the degree of overlap when one block of integration candidates is used as a reference and the degree of overlap when the other is used as a reference, image quality deterioration due to integration is reduced. be able to.

  In addition, when the data distribution range of two adjacent blocks is within a narrow range, by integrating these blocks, even if the data distribution range does not overlap or the overlap is small, the flat portion Can be extracted appropriately and blocks can be integrated.

  Furthermore, in the case where the reference value for determining the integration is changed according to the block size of the integration candidate, the block integration is promoted in the region where the probability of the flat portion is high due to the integration state of the neighboring regions.

  In the case of changing the priority of the integration direction according to the character direction, it is possible to efficiently determine the direction that is highly correlated with neighboring pixels using the character direction as a clue in character-based documents, etc., and the reproducibility of the thin line of characters Decline can be prevented. In addition, more efficient block integration can be expected by preferentially determining the direction of high correlation. Furthermore, even when arithmetic coding is combined as post-processing, correlation with adjacent blocks is easily maintained even after integration, and the amount of encoded information can be suppressed.

  Further, unnecessary information is reduced in the flat portion by the block integration process, so that information with low correlation is selectively excluded, which contributes to suppression of the amount of information at the time of encoding in the secondary compression unit 25.

  The embodiment of the present invention has been described with reference to the drawings. However, the specific configuration is not limited to that shown in the embodiment, and there are changes and additions within the scope of the present invention. Are also included in the present invention.

  For example, when the character direction is diagonal, the blocks adjacent to each other in the diagonal direction may be preferentially determined for integration. In addition, a typeface identification function for characters may be provided so that priority is applied only when the typeface is predetermined. Further, when a character cannot be detected, priority may be applied when it is determined that the characteristic has anisotropy by a spatial frequency analysis process or the like.

The configuration information developed in the two-dimensional array may include, in addition to the configuration data of each block, region attribute information of the image acquired separately (any plane, local region, pixel, etc.). The lossless compression method may be arithmetic coding other than JBIG.
In addition, the determination of similarity is not limited to that based on the degree of overlap of the data distribution range (density distribution range) by the maximum value and the minimum value. For example, when the average value and the standard deviation are used as block statistical information Alternatively, the determination may be made based on the overlapping degree of the density range represented by the average value and the standard deviation.

It is a block diagram which shows the structure of the image processing apparatus which concerns on embodiment of this invention. It is a block diagram which shows the image processing part of the image processing apparatus which concerns on embodiment of this invention. It is a flowchart which shows the outline of the whole process which the image processing part of the image processing apparatus which concerns on embodiment of this invention performs. 4 is a flowchart showing in more detail the encoding process in step S101 of FIG. 3. It is explanatory drawing which shows the relationship between the priority of a character direction and an integrated direction. It is explanatory drawing which shows the processing content of the encoding process of 1 block in BTC compression. It is explanatory drawing which illustrated the correspondence of block size, encoding information amount, and compression rate. 4 is an explanatory diagram showing a process of integrating 4 × 4 adjacent blocks A and B into one block U. FIG. 6 is a flowchart showing in more detail “one set of block integration processing” shown in step S117 of FIG. 4. It is explanatory drawing which illustrated a mode that block size expanded by repeating block integration. It is explanatory drawing which shows an example of the integration state of the block within an image. It is explanatory drawing which shows the relationship between the distribution state of pixel density | concentration, and the determination result of integrating / not performing. It is a flowchart which shows the integrated determination process which the image processing part of the image processing apparatus which concerns on embodiment of this invention performs. It is explanatory drawing which illustrated the combination of the block statistical information for specifying or estimating the maximum value and minimum value of a density | concentration. It is explanatory drawing which shows the calculation formula which calculates the overlapping degree OA on the basis of the block A and the overlapping degree OB on the basis of the block B. It is explanatory drawing which illustrated the mode of the two-dimensional expansion | deployment performed by step S102 of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 5 ... Data transmission network 10 ... Image processing apparatus 11 ... Data control part 12 ... Image display part 13 ... Operation part 14 ... Image input part 15 ... Image output part 16 ... Data holding part 17 ... Data communication I / F part 20 ... Image Processing unit 21 ... Primary compression unit 21a ... Block division unit 21b ... Maximum / minimum value detection unit 21c ... Compression code calculation unit 22 ... Adjacent block comparison unit 23 ... Block combination unit 24 ... Character detection unit 25 ... Secondary compression unit 41 ... 4 pixels x 4 pixels rectangular area forming 1 block 50 ... Compression ratio comparison table

Claims (14)

After dividing an image into a plurality of blocks, whether to perform BTC compression on a block basis and integrate the first block and the second block adjacent to each other after compression, whether the first block and the second block, Judgment is made based on both the similarity when the first block is used as a reference and the similarity when the second block is used as a reference. When it is determined that the first block and the second block are integrated, the first block and the second block are determined. An image processing apparatus, wherein the image is encoded by being integrated into one block. The image processing apparatus according to claim 1, wherein the similarity is a degree of overlap between the density distribution range of the first block and the density distribution range of the second block. The reference value for determining the similarity based on the first block and the reference value for determining the similarity based on the second block are the same. 3. Image processing apparatus. After the image is divided into a plurality of blocks, BTC compression is performed on a block-by-block basis, and whether or not the first block and the second block adjacent to each other after the compression are integrated is determined between the first block and the second block. Judgment based on the difference between the maximum density and the minimum density of the image, and if it is determined to integrate, the first block and the second block are integrated into one block, thereby performing the image encoding process. A featured image processing apparatus. After the image is divided into a plurality of blocks, BTC compression is performed on a block-by-block basis, and it is determined whether or not the compressed first and second blocks adjacent to each other are integrated. By integrating the block and the second block into one block, the image is encoded,
An image processing apparatus, wherein the reference value for the determination is changed according to a block size of either or both of the first block and the second block. The character direction in the image is detected, and the priority of the arrangement direction of the first block and the second block to be determined is changed according to the detected character direction. The image processing apparatus according to any one of 1 to 5. The configuration data of each block after the encoding process is made into a one-dimensional or two-dimensional array of binary information,
A one-dimensional or two-dimensional array for each block is two-dimensionally arranged corresponding to the original block array in the image to form a binary image,
The image processing apparatus according to claim 1, wherein the generated binary image is reversibly compressed. After dividing an image into a plurality of blocks, whether to perform BTC compression on a block basis and integrate the first block and the second block adjacent to each other after compression, whether the first block and the second block, Judgment is made based on both the similarity when the first block is used as a reference and the similarity when the second block is used as a reference. When it is determined that the first block and the second block are integrated, the first block and the second block are determined. An image processing method, wherein the image is encoded by being integrated into one block. The image processing method according to claim 8, wherein the similarity is a degree of overlap between the density distribution range of the first block and the density distribution range of the second block. 10. The reference value for determining similarity based on the first block and the reference value for determining similarity based on the second block are the same. 10. Image processing method. After the image is divided into a plurality of blocks, BTC compression is performed on a block-by-block basis, and whether or not the first block and the second block adjacent to each other after the compression are integrated is determined between the first block and the second block. Judgment based on the difference between the maximum density and the minimum density of the image, and if it is determined to integrate, the first block and the second block are integrated into one block, thereby performing the image encoding process. A featured image processing method. After the image is divided into a plurality of blocks, BTC compression is performed on a block-by-block basis, and it is determined whether or not the compressed first and second blocks adjacent to each other are integrated. By integrating the block and the second block into one block, the image is encoded,
The image processing method according to claim 1, wherein the reference value for the determination is changed according to a block size of one or both of the first block and the second block. The character direction in the image is detected, and the priority of the arrangement direction of the first block and the second block to be determined is changed according to the detected character direction. The image processing method according to any one of 8 to 12. The configuration data of each block after the encoding process is made into a one-dimensional or two-dimensional array of binary information,
A one-dimensional or two-dimensional array for each block is two-dimensionally arranged corresponding to the original block array in the image to form a binary image,
The image processing method according to any one of claims 8 to 13, wherein the generated binary image is reversibly compressed.

Images