JP2011077760A - Image compression device and image compression program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve coding efficiency while preventing degradation in image quality after decoding, in an image having a large density difference between adjacent pixels, such as a dot image. <P>SOLUTION: An image compression unit calculates a value A obtained by setting '0' or '1' to all lower bits without changing the bit digit of a pixel of interest, and a value B obtained by setting '0' or '1' to all the lower bits by changing the digit of the bit digit. The image compression unit calculates an absolute value (first error) of the difference between a value of the pixel of interest and the value A, and an absolute value (second error) of the difference between a value of the pixel of interest and the value B. The image compression unit does not change a value of the bit digit of the pixel of interest when the first error is not larger than the second error, and changes the value of the bit digit of the pixel of interest when the first error is larger than the second error. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image compression apparatus and an image compression program for compressing image data.

  In recent years, with the development of digital devices, various compression methods for image data and the like have been proposed, and JPEG (Joint Photographic Experts Group) 2000 is known as one of the compression methods. JPEG2000 defines an image compression / decompression method, and is an extension of JPEG. JPEG2000 divides image data into a plurality of tiles, and performs compression processing by wavelet transform, quantization for each tile, and arithmetic coding for each bit plane.

  Among various images, the halftone image has a large density difference between adjacent pixels, and the same value ('0' or '1') does not continue in the bit plane, so that the encoding efficiency tends to deteriorate. Therefore, a method of switching the quantization table by judging the type of image in order to improve the encoding efficiency (Patent Document 1), or replacing the least significant bit with '0' or '1' in the quantized data and fixing the value A method to do this has been proposed (Patent Document 2).

Japanese Patent Application Laid-Open No. 07-038761 JP-A-10-285405

  However, in the case of the method described in Patent Document 1, since the arrangement of values between adjacent pixels is not taken into consideration, an improvement in coding efficiency cannot be expected for halftone images. In addition, in the case of the method described in Patent Document 2, since the value of the least significant bit is forcibly changed, there is a problem that image quality degradation of an image after decoding becomes large. In particular, when the compression rate is controlled, the bit position to be encoded fluctuates, so the method of fixing the value of the least significant bit has no effect.

  The present invention has been made to solve the above problem, and is an image that improves coding efficiency while suppressing image quality degradation after decoding in an image having a large density difference between adjacent pixels such as a halftone image. An object of the present invention is to provide a compression device and an image compression program.

  An image compression apparatus according to a first aspect of the present invention is a conversion unit that converts input image data into a color system comprising a luminance component and a color difference component, and a division that divides the converted image data into a plurality of tiles. Means, wavelet transform means for performing subband decomposition by performing wavelet transform for each tile, quantization means for quantizing each subband, and decomposing the quantized subband into a plurality of blocks Decomposition means, expansion means for expanding the pixel values in the block for each block into a plurality of bit planes from the most significant bit to the least significant bit, the value of the pixel of interest that is one pixel in the block, The pair of the target pixel according to each value of the target bit digit in each pixel before and after the target pixel and a value of a lower bit than the target bit digit of the target pixel Bit rounding means for adjusting the value of the bit digit and changing all the lower bits to “0” or “1”, and encoding means for coding the bit plane subjected to bit rounding processing by the bit rounding means And comprising.

  According to a fifth aspect of the present invention, there is provided an image compression program for converting a computer to input color data into a color system composed of a luminance component and a color difference component, and converting the converted image data into a plurality of tiles. Dividing means for dividing the tiles, wavelet transform means for performing subband decomposition by performing wavelet transformation for each tile, quantizing means for quantizing each subband, and decomposing the quantized subbands into a plurality of blocks Disassembling means for expanding the pixel value in the block into a plurality of bit planes from the most significant bit to the least significant bit for each block, the value of the target pixel being one pixel in the block, the target According to each value of the target bit digit in each pixel before and after the pixel, and a value of a lower bit than the target bit digit of the target pixel A bit rounding unit that adjusts the value of the target bit digit of the eye pixel to change all the lower bits to “0” or “1”, and encodes a bit plane that has been subjected to bit rounding processing by the bit rounding unit It is made to function as an encoding means.

  According to these configurations, the value of the target pixel that is one pixel in the block, the value of the target bit digit in each pixel before and after the target pixel, and the value of the lower bit than the target bit digit of the target pixel In order to adjust the value of the target bit digit of the target pixel accordingly (specifically, whether or not to change it), bit rounding is performed in consideration of the continuity with each pixel before and after the target pixel and the value change of the target pixel. Processing can be performed. That is, encoding efficiency can be improved by considering continuity with each pixel before and after the target pixel. In addition, since the amount of change in the pixel value of the target pixel due to the bit rounding process can be suppressed as much as possible by considering the change in the value of the target pixel, image quality degradation after decoding can be reduced.

  An image compression apparatus according to a second aspect of the present invention is the image compression apparatus according to the first aspect, wherein the bit rounding means is different in each value of the target bit digit in each pixel before and after the target pixel. In this case, the value of the target pixel and the value of the target bit digit of the target pixel are left as they are, and the absolute value of the difference between the values when all the lower bits of the target bit digit are set to '0' or '1'. The difference between the first error and the value when the value of the target pixel and the value of the target bit digit of the target pixel are changed to set all the lower bits to “0” or “1”. When the second error, which is an absolute value, is calculated and the first error is equal to or smaller than the second error, the value of the target bit digit of the target pixel is left as it is, and the first error is larger than the second error Changes the value of the target bit digit of the pixel of interest Te, is to change the low-order bits to all '0' or '1'.

  According to this configuration, it is possible to improve the encoding efficiency while minimizing the change amount of the pixel value before and after the bit rounding process of the pixel of interest (suppressing image quality deterioration after decoding).

  An image compression apparatus according to a third aspect of the present invention is the image compression apparatus according to the first or second aspect, wherein the bit rounding means is configured to set each value of the target bit digit in each pixel before and after the target pixel. And when the same value continues for a predetermined number in the target bit digit of each pixel before the target pixel, the value of the target pixel and the value of the target bit digit of the target pixel are left as they are. A first error that is an absolute value of a difference from a value when all lower bits of the target bit digit are set to “0” or “1”, a value of the target pixel, and a value of the target bit digit of the target pixel Is changed to calculate a second error which is an absolute value of a difference from a value when all the lower bits of the target bit digit are set to “0” or “1”, and the first error and the second error are calculated. When the difference is within a predetermined setting range, If the difference between the first error and the second error is outside the set range and the first error is less than or equal to the second error, the target bit of the target pixel is left unchanged. If the digit value is left as it is, and the difference between the first error and the second error is outside the setting range and the first error is larger than the second error, the value of the target bit digit of the target pixel To change all the lower bits to '0' or '1'.

  According to this configuration, priority can be given to the continuity of the value of the target bit digit of the pixel before the pixel of interest, and the encoding efficiency can be improved. On the other hand, when the difference between the first error and the second error is outside the setting range, the value of the target bit digit of the target pixel is changed, thereby suppressing the change in the pixel value before and after the bit rounding process of the target pixel as much as possible (after decoding). Encoding efficiency can be improved while suppressing image quality degradation.

  A fourth aspect of the present invention is the image compression apparatus according to the first aspect, wherein the bit rounding means performs the attention when each value of the target bit digit in each pixel before and after the target pixel is different. When all the lower bits of the target bit digit of the pixel are set to “0”, when the value indicated by the lower bit of the target bit digit of the target pixel is equal to or greater than a predetermined threshold, the target bit of the target pixel When the digit is changed to '1' and the value indicated by the lower bit is less than the threshold value, the value of the target bit digit of the target pixel is left as it is, and all the lower bits from the target bit digit of the target pixel are all When “1” is set, the target bit digit of the target pixel is changed to “0” when the value indicated by the lower bit than the target bit digit of the target pixel is equal to or less than a predetermined threshold, and the lower order Bi When the value indicated by the bets is greater than the threshold value is intended to be as the value of the target bit digit of the pixel of interest.

  According to this configuration, it is possible to improve the encoding efficiency while minimizing the change in the pixel value before and after the bit rounding process of the target pixel (suppressing image quality deterioration after decoding).

  According to the present invention, it is possible to improve the encoding efficiency while minimizing the change in the pixel value before and after the bit rounding process of the pixel of interest (suppressing the image quality deterioration after decoding).

1 is a block diagram showing an electrical configuration of an image compression apparatus. The figure which shows the flow of the image compression by a JPEG2000 system. The figure for demonstrating tile division | segmentation. The schematic diagram shown about discrete wavelet transform. The figure which shows the flow of entropy encoding. The figure which showed the precinct division | segmentation typically. The figure which showed code block division | segmentation typically. The figure which showed bit-plane division | segmentation typically. The figure which showed the division | segmentation to a coding pass typically. The table | surface which showed the combination of the value of the object bit digit in the front pixel of the attention pixel, and a back pixel. The figure for demonstrating a 1st Example. The figure for demonstrating a 1st Example. The flowchart which showed the flow of the bit rounding process in a 1st Example. The figure for demonstrating a 2nd Example. The figure for demonstrating a 2nd Example. The figure for demonstrating a 2nd Example. The flowchart which showed the flow of the bit rounding process in a 2nd Example. The flowchart following FIG. The figure for demonstrating a 3rd Example. The figure for demonstrating a 3rd Example. The flowchart which showed the flow of the bit rounding process in a 3rd Example.

  Embodiments of an image compression apparatus and an image compression program according to the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an electrical configuration of an image compression apparatus 1 according to the present embodiment. The image compression apparatus 1 is realized by an information processing apparatus such as a personal computer (personal computer), a workstation, or a portable information terminal. The block diagram of FIG. 1 shows a case where the image compression apparatus 1 is realized by these information processing apparatuses. Is shown as an example. In addition, it may be used by being incorporated in an image reading apparatus such as a copying machine or a scanner, or an imaging apparatus such as a digital camera, a digital video camera, or a mobile phone equipped with a digital camera.

  As shown in FIG. 1, the image compression apparatus 1 includes a control unit 11, a storage unit 12, an input operation unit 13, a display unit 14, an I / F unit 15, a network I / F unit 16, an image compression / decoding unit 17, and the like. Configured. The control unit 11 is configured by a CPU (Central Processing Unit) or the like, reads a program stored in the storage unit 12 in accordance with an input instruction signal or the like, executes a process, and sends the processing to each functional unit. The image compression apparatus 1 is comprehensively controlled by outputting an instruction signal, transferring data, and the like.

  The storage unit 12 stores programs, data, and the like for realizing various functions provided in the image compression apparatus 1. In the present embodiment, the storage unit 12 stores an image compression program 121. The image compression program 121 is a program for compressing the image data input by the I / F unit 15 or the network I / F unit 16 using the JPEG2000 method to generate encoded data.

  The input operation unit 13 includes various operation buttons and a pointing device such as a mouse, and outputs an operation signal to the control unit 11 when operated by a user. The user also sets the compression rate via the input operation unit 13. The display unit 14 is a display screen such as a liquid crystal display, and displays according to the content input from the input operation unit 13 or displays the processing content and processing result by the control unit 11.

  The I / F unit 15 is an interface such as IEEE1394 or USB, and can directly transmit / receive data to / from an external device. The network I / F unit 16 includes a communication module such as a LAN board, and transmits / receives various data to / from an external device via a network (not shown) connected to the network I / F unit 16. The image compression / decoding unit 17 compresses the image data input via the I / F unit 15 or the network I / F unit 16 by the JPEG 2000 system by executing the image compression program 121 stored in the storage unit 12. To generate compressed data. The image compression / decoding unit 17 also decodes the compressed data.

  FIG. 2 is a diagram showing a flow of image compression according to the JPEG2000 system in the present embodiment. First, the image compression / decoding unit 17 performs component conversion of the color system of the input image data from, for example, the RGB system to the YCbCr system (color space conversion 21), and the image data obtained by converting the color system is shown in FIG. In this way, it is divided into a plurality of tiles (tile division 22).

  Then, the image compression decoding unit 17 performs discrete wavelet transform on each tile of one band and outputs coefficient data (discrete wavelet transform 23). FIG. 4 is a schematic diagram showing the discrete wavelet transform. Discrete wavelet transform is applied to the tile image (0LL) in FIG. 4A, and is decomposed into subbands 1LL, 1HL, 1LH, and 1HH as shown in FIG. 4B. Subsequently, a discrete wavelet transform is performed on the low frequency component 1LL, and it is decomposed into subbands 2LL, 2HL, 2LH and 2HH as shown in FIG. Although FIG. 4 illustrates two-level conversion, the number of discrete wavelet conversions is not particularly limited.

  Next, the image compression decoding unit 17 performs linear quantization on the coefficient data subjected to the discrete wavelet transform (quantization 24), and quantizes it with a compression rate (hereinafter referred to as “set compression rate”) set by the user or the like. Entropy coding is performed on the quantized data (entropy coding 25). Here, the processing procedure of the entropy encoding 25 will be described. FIG. 5 is a diagram showing a flow of entropy encoding. The image compression decoding unit 17 first divides each subband coefficient into areas called precincts (precinct division 251). FIG. 6 is a diagram schematically showing precinct division. In FIG. 6, a portion 51 is obtained by frequency-converting a portion of the same region in the original image, and these portions belong to the same precinct.

  Next, the image compression decoding unit 17 divides the precinct into smaller areas called code blocks (code block division 252). FIG. 7 is a diagram schematically showing code block division. This code block unit is a basic unit when entropy encoding is performed.

  Then, the image compression decoding unit 17 expands the transform coefficient of the discrete wavelet transform linearly quantized for each code block into a bit plane (bit plane expansion 253). FIG. 8 is a diagram schematically showing the bit plane development. In FIG. 8, 61 is an example of coefficient data of the linearly quantized discrete wavelet transform in a certain code block. 62 is a bit string representing the sign of the coefficient data 61, where value 0 means a positive value and value 1 means a negative value. 63 is a bit plane in which the absolute value of the coefficient data 61 is binary-developed from MSB (Most Significant Bit) to LSB (Least Significant Bit).

  For example, since the coefficient data 64 having a value of +12 is a positive value, the corresponding sign bit 65 is “0”. Also, since the binary representation of the absolute value of +12 is (1100), the values of the locations 66a, 66b, 66c and 66d corresponding to the bit plane are “1”, “1”, “0”, “0”, respectively. It becomes. Similarly, since the coefficient data 67 having a value of −6 is a negative value, the corresponding sign bit 68 is “1”. Since the binary representation of the absolute value of −6 is (0110), the values of the corresponding portions 69a, 69b, 69c, and 69d of the bit plane are “0”, “1”, “1”, “0”, respectively. 'Become. Bit planes that are all zero on the MSB side are called zero bit planes, and no data is stored, while the number of zero bit planes described later is counted for each code block.

  Subsequently, the image compression decoding unit 17 further divides the bit plane into a significance propagation pass, a magnitude refinement pass, and a cleanup pass (division 254 into coding passes). FIG. 9 is a diagram schematically showing division into coding passes. As shown in FIG. 9, the bit planes 71a to 71d (hereinafter collectively referred to as “bit plane 71”) are divided as follows by division into coding paths. That is, significance propagation paths 72b to 72d (hereinafter collectively referred to as “significance propagation path 72”), magnitude refinement paths 73b to 73d (hereinafter collectively referred to as “magnitude refinement path 73”), and cleanup path 74b. Divided into ~ 74d. However, the most significant bit (MSB side) bit plane 71a is made to correspond only to the cleanup path 74a. Hereinafter, the cleanup paths 74a to 74d are collectively referred to as “cleanup path 74”.

  Each bit plane 71 and each coding pass 72 to 74 are all equal in size according to the coordinate length in the vertical and horizontal directions. Each coding pass 72 to 74 has a position where a bit value is defined and a position where it is not defined. In FIG. 9, for example, 76 and 77, the positions where bit values are defined are shaded (hatched). The bit values defined in the shaded portions of the coding paths 72 to 74 (for example, the coding paths 72b to 74b) are the bit values at the corresponding positions on the bit plane 71 (for example, the bit plane 71b) before the division. equal. Since the processing method for dividing the bit plane into coding passes is well known in each document, description thereof is omitted.

  Finally, the image compression decoding unit 17 arithmetically encodes the data after the coding pass division (binary arithmetic encoding 255). After performing the entropy encoding 25 as described above, the image compression / decoding unit 17 adjusts the compression rate (compression rate adjustment 26), and the image compression / decoding unit 17 forms a code stream for writing data to the file. (Code stream 27). Since this code stream processing is well known from each document, the description thereof is omitted.

  Among various images, the halftone image has a large density difference between adjacent pixels, and the same value ('0' or '1') does not continue in the bit plane, so that the encoding efficiency tends to deteriorate. Therefore, a method for switching the quantization table by determining the type of image in order to increase the coding efficiency, or a method for fixing the value by replacing the least significant bit with '0' or '1' in the quantized data has been proposed. ing. However, in the method of switching the quantization table, since the arrangement of values between adjacent pixels is not taken into consideration, it is not possible to expect much improvement in encoding efficiency in a halftone image. In addition, in the method of fixing the value of the least significant bit, the value of the least significant bit is forcibly changed, so that there is a problem that image quality degradation of the image after decoding becomes large.

  Accordingly, first to third embodiments for improving the coding efficiency while suppressing image quality degradation after decoding even in an image having a large density difference between adjacent pixels such as a halftone image will be described below.

<First embodiment>
In this embodiment, when the values of the target bit digit in each pixel before and after the target pixel of the code block are different, all the lower-order bits from the target bit digit are maintained with the target pixel value and the target bit digit value of the target pixel unchanged. The first error, which is the absolute value of the difference from the value when it is set to “0” or “1”, and the value of the target pixel and the target bit digit of the target pixel are changed, and all the lower bits than the target bit digit are changed. The second error, which is the absolute value of the difference from the value when it is set to “0” or “1”, is calculated. When the first error is less than or equal to the second error, the value of the target bit digit of the target pixel remains as it is, A method for improving the coding efficiency while suppressing the deterioration of the image quality after decoding by changing the value of the target bit digit of the target pixel when the first error is larger than the second error will be described.

  FIG. 10 is a table showing combinations of values of target bit digits in the front pixel and the rear pixel of the target pixel. First, when the values of the target bit digits of the front pixel, the target pixel, and the rear pixel are all “0” or “1”, the image compression decoding unit 17 does not change the target bit digit of the target pixel. On the other hand, when the value of the target bit digit of the front pixel and the rear pixel is the same and only the target bit digit of the target pixel is different, for example, the front pixel = '0', the target pixel = '1', and the rear pixel = '0' In this case, the image compression decoding unit 17 changes the value of the target bit digit of the target pixel to the value of the target bit digit of the previous pixel. That is, in the case of the above example, the target pixel = “1” is changed to “0”. Encoding efficiency is improved by providing value continuity in this way.

  The image compression decoding unit 17 adjusts the value of the target bit digit of the target pixel in this way, and then performs bit rounding processing with all the lower bits of the target bit digit set to ‘0’ or ‘1’.

  If the values of the target bit digits of the previous pixel and the subsequent pixel are different, for example, if the previous pixel = '0', the target pixel = '0', and the rear pixel = '1', the image compression decoding unit 17 It is determined whether to change the target bit digit by comparing the value of the target bit digit of the pixel with or without changing the value of the target bit digit of the pixel with the value of the target pixel before the change. Hereinafter, a method for determining whether or not to change the target bit digit of the target pixel by the image compression decoding unit 17 will be described in detail.

  First, it demonstrates using FIG. FIG. 11A is a diagram showing the original value of the pixel of interest and the values of the pixels before and after the pixel. The previous pixel = 110011111B, the pixel of interest = 1001100B, and the rear pixel = 11011000B. Referring to the value of the target bit digit of the previous pixel and the subsequent pixel and the table of FIG. 10, since the previous pixel = '0' and the subsequent pixel = '1', the image compression decoding unit 17 determines the target bit digit of the target pixel. It will be determined whether or not the value should be changed.

First, the image compression decoding unit 17
Value A in which the target bit digit value of the target pixel remains the same and all the lower bits are set to “0”
= 10000000B = 128D
Value B = 10010000B = 144D in which the target bit digit of the target pixel is changed from “0” to “1” and all the lower bits are set to “0”
Get each value of. The image compression decoding unit 17 calculates the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value (second error) of the difference between the value of the target pixel and the value B. That is, since the original target pixel is 10001100B = 140D,
First error = | 140−128 | = 12
2nd error = | 140−144 | = 4
It becomes.

Next, the image compression decoding unit 17 obtains the difference between the first error and the second error, and if the difference is 0 or less (that is, the first error is less than the second error), the value of the target bit digit of the target pixel remains as it is. If the difference is greater than 0 (that is, the second error is less than the first error), the value of the pixel of interest is changed. In the above case,
First error-second error = 12-4 = 8> 0
Therefore, the image compression decoding unit 17 changes the target bit digit of the target pixel. That is, the image compression / decoding unit 17 changes the value of the target bit digit of the target pixel as it is and sets all the lower bits to “0” and the value of the target bit digit changed to set all the lower bits to “0”. The absolute value of the difference from the original pixel of interest is calculated, and the smaller one (if the difference is the same, the target bit digit remains unchanged) is adopted.

  Next, another example will be described with reference to FIG. FIG. 12A shows the value of the original target pixel, where target pixel = 10000111B, front pixel = 110011111B, and rear pixel = 11011000B. Referring to the value of the target bit digit of the previous pixel and the subsequent pixel and the table of FIG. 10, since the previous pixel = '0' and the subsequent pixel = '1', the image compression decoding unit 17 determines the target bit digit of the target pixel. It will be determined whether or not the value should be changed.

First, the image compression decoding unit 17
Value A in which the target bit digit value of the target pixel remains the same and all the lower bits are set to “0”
= 10000000B = 128D
Value B = 10010000B = 144D in which the target bit digit of the target pixel is changed from “0” to “1” and all the lower bits are set to “0”
Get each value of. The image compression decoding unit 17 calculates the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value (second error) of the difference between the value of the target pixel and the value B. That is, since the value of the original target pixel is 10000111B = 135D,
First error = | 135−128 | = 7
2nd error = | 135-144 | = 9
It becomes.

Next, the image compression decoding unit 17 obtains the difference between the first error and the second error, and if the difference is 0 or less, the value of the target bit digit of the target pixel is left as it is, and if the difference is a value greater than 0, Change the value of the target pixel. In the above case,
First error−second error = 7−9 = −2 ≦ 0
Therefore, the image compression decoding unit 17 leaves the target bit digit of the target pixel as it is.

  FIG. 13 is a flowchart showing the flow of bit rounding processing in the present embodiment. Hereinafter, “bit rounding processing” refers to adjustment of the target bit digit of the target pixel (processing executed by determining whether to change or leave it as is), and all lower bits from the target bit digit are “0” or “1”. The process of replacing with '. A program related to bit rounding processing is included in the image compression program 121, and the image compression decoding unit 17 performs bit rounding processing while executing processing according to the image compression program 121.

  First, the image compression decoding unit 17 compares the target pixel of the code block, the values of the target bit digits of the previous pixel and the subsequent pixel of the target pixel. When the values of the target bit digits of the front pixel, the target pixel, and the rear pixel are all the same (step S11; YES), the image compression decoding unit 17 moves the process to step S21.

  When the values of the target bit digits of the front pixel, the target pixel, and the rear pixel are not all the same (step S11; NO), and the values of the target bit digits of the front pixel and the rear pixel are the same, that is, When only the value of the target bit digit is different (step S12; YES), if the value of the target bit digit of the previous pixel is “0” (step S13; “0”), the image compression decoding unit 17 sets the target pixel target. The bit digit value is changed from '1' to '0' (step S14). If the value of the target bit digit of the previous pixel is “1” (step S13; “1”), the image compression decoding unit 17 changes the value of the target bit digit of the target pixel from “0” to “1” ( Step S14). Then, the image compression decoding unit 17 shifts the process to step S21.

  On the other hand, the values of the target bit digits of the previous pixel, the target pixel, and the rear pixel are not all the same (step S11; NO), and the values of the target bit digits of the front pixel and the rear pixel are different (step S12; NO). The image compression decoding unit 17 changes the value A of the target bit digit of the target pixel by changing the value A in which all the lower bits are set to “0” or “1” and the value of the target bit digit of the target pixel without changing the value of the target bit digit of the target pixel. A value B that is all set to “0” or “1” is generated (step S16). Subsequently, the image compression decoding unit 17 calculates a first error that is an absolute value of the difference between the target pixel and the value A, and a second error that is an absolute value of the difference between the target pixel and the value B (step S17).

  Then, the image compression decoding unit 17 calculates the difference between the first error and the second error (step S18). When the difference is 0 or less (step S18; YES), the image compression decoding unit 17 keeps the value of the target bit digit of the target pixel as it is (step S19). When the difference is larger than 0 (step S18; NO), the image compression decoding unit 17 changes the value of the target bit digit of the target pixel (step S20).

  Next, the image compression decoding unit 17 changes all the bits lower than the target bit digit of the target pixel to “0” or “1” (step S21). When all the lower bits are set to “0”, the image compression decoding unit 17 generates a value A and a value B when all the lower bits are set to “0” in step S16. When all the lower bits are set to ‘1’, the image compression decoding unit 17 generates a value A and a value B when all the lower bits are set to ‘1’ in step S <b> 16.

  If the bit rounding process is not performed on all the pixels of the code block (step S22; NO), the image compression decoding unit 17 moves the target pixel (step S23) and shifts the process to step S11. When the bit rounding process for all the pixels of the code block is completed (step S22; NO), the image compression decoding unit 17 ends the process. Then, the image compression decoding unit 17 performs arithmetic coding after the bit rounding process is completed.

  Next, the flow of the compressed data decoding process will be described. First, the image compression decoding unit 17 performs inverse entropy coding on the compressed data to obtain quantized data. Subsequently, the image compression decoding unit 17 performs inverse quantization on the quantized data, and obtains coefficient data subjected to discrete wavelet transform. Then, the image compression decoding unit 17 performs inverse discrete wavelet transform on the coefficient data, restores the tile image, performs colorimetric component conversion of the image obtained by combining the tile images, acquires the decoded data, and performs the decoding process. finish.

  As described above, the image compression decoding unit 17 changes the target bit digit of the target pixel without changing the target bit digit as it is and sets the lower bit to “0” or “1” and the lower bit by changing the target bit digit. A value B that is all set to “0” or “1” is calculated, and the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value of the difference between the value of the target pixel and the value B (second) When the first error is less than or equal to the second error, the value of the target bit digit of the target pixel is left as it is, and when the first error is greater than the second error, the value of the target bit digit of the target pixel In an image with a large density difference between adjacent pixels, such as a halftone image, while minimizing the value change before and after the bit change of the pixel of interest (while suppressing image quality degradation after decoding) Encoding efficiency can be improved.

<Second embodiment>
In the present embodiment, a case will be described in which each value of the target bit digit in each pixel before and after the target pixel of the code block is different, and the same value continues for a predetermined number or more in the target bit digit of each pixel before the target pixel. .

  First, a description will be given with reference to FIG. FIG. 14A is a diagram illustrating the values of the target pixel and its surrounding pixels, where the previous pixel = 110011111B, the target pixel = 100000111B, and the rear pixel = 11011000B. Referring to the value of the target bit digit of the previous pixel and the subsequent pixel and the table of FIG. 10, since the previous pixel = '0' and the subsequent pixel = '1', the image compression decoding unit 17 determines the target bit digit of the target pixel. It will be determined whether or not the value should be changed.

Next, the image compression decoding unit 17
Value A in which the target bit digit value of the target pixel remains the same and all the lower bits are set to “0”
= 10000000B = 128D
Value B = 10010000B = 144D in which the target bit digit of the target pixel is changed from “0” to “1” and all the lower bits are set to “0”
Get each value of. The image compression decoding unit 17 calculates the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value (second error) of the difference between the value of the target pixel and the value B. That is, since the original value of the target pixel is 10000111B = 135D,
First error = | 135−128 | = 7
2nd error = | 135-144 | = 9
It becomes.

Subsequently, the image compression decoding unit 17 determines whether or not the same value continues in a predetermined number (for example, five or more) in the target bit digit of the pixel before the target pixel. The continuous number of the same value can be set as appropriate. In the case of FIG. 14A, five “0” s are continued in the target bit digit of the pixel before the target pixel (shaded portion in FIG. 14A). Accordingly, the image compression decoding unit 17 obtains the difference between the first error and the second error. Then, the image compression decoding unit 17 determines whether or not the difference between the first error and the second error is within the set range, and when the difference between the first error and the second error is within the set range, the target pixel The value of the target bit digit is kept as it is, and when the difference between the first error and the second error is outside the setting range, the value of the target bit digit of the target pixel is adjusted according to the method described in the first embodiment. For example, when the setting range is set to −6 ≦ x ≦ + 6 (x is the difference between the first error and the second error),
First error-second error = 7-9 = -2
Thus, the difference between the first error and the second error is within the set range. Therefore, the image compression decoding unit 17 keeps the value of the target bit digit of the target pixel as it is. FIG. 14B is a diagram illustrating values of the pixel of interest and the surrounding pixels after the bit rounding process.

  Next, another example will be described with reference to FIG. FIG. 15A is a diagram showing the values of the original target pixel and its peripheral pixels, where front pixel = 1100111B, target pixel = 1001100B, and rear pixel = 11011000B.

First, the image compression decoding unit 17
Value A in which the target bit digit value of the target pixel remains the same and all the lower bits are set to “0”
= 10000000B = 128D
Value B = 10010000B = 144D in which the target bit digit of the target pixel is changed from “0” to “1” and all the lower bits are set to “0”
Get each value of. The image compression decoding unit 17 calculates the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value (second error) of the difference between the value of the target pixel and the value B. That is, since the target pixel = 1001100B = 140D,
First error = | 140−128 | = 12
2nd error = | 140−144 | = 4
It becomes.

  Subsequently, the image compression decoding unit 17 determines whether five or more of the same values are consecutive in the target bit digit of the pixel before the target pixel. In the case of FIG. 15A, five “0” s are continued in the target bit digit of the pixel before the target pixel (the shaded portion in FIG. 15A). Accordingly, the image compression decoding unit 17 obtains the difference between the first error and the second error.

First error-second error = 12-4 = 8
Since the difference between the first error and the second error is outside the setting range, the image compression decoding unit 17 determines whether or not to change the value of the target bit digit of the target pixel according to the method described in the first embodiment. to decide. In the above case, since the difference between the first error and the second error is +8, the second error is smaller than the first error. Therefore, the image compression decoding unit 17 changes the target bit digit. FIG. 15B is a diagram illustrating values of the pixel of interest and the surrounding pixels after the bit rounding process.

  Furthermore, another example will be described with reference to FIG. FIG. 16A is a diagram showing the values of the original target pixel and its surrounding pixels, where front pixel = 1100111B, target pixel = 10000011B, and rear pixel = 11011000B.

First, the image compression decoding unit 17
Value A in which the target bit digit value of the target pixel remains the same and all the lower bits are set to “0”
= 10000000B = 128D
Value B = 10010000B = 144D in which the target bit digit of the target pixel is changed from “0” to “1” and all the lower bits are set to “0”
Get each value of. The image compression decoding unit 17 calculates the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value (second error) of the difference between the value of the target pixel and the value B. That is, since the target pixel = 10000011B = 131D,
First error = | 131−128 | = 3
2nd error = | 131-144 | = 13
It becomes.

  Next, the image compression decoding unit 17 determines whether five or more of the same values are consecutive in the target bit digit of the pixel before the target pixel. In the case of FIG. 16A, five “0” s are continued in the target bit digit of the pixel before the target pixel (shaded portion in FIG. 16A). Accordingly, the image compression decoding unit 17 obtains the difference between the first error and the second error.

First error-second error = 3-13 = -10
Since the difference between the first error and the second error is outside the setting range, the image compression decoding unit 17 determines whether or not to change the value of the target bit digit of the target pixel according to the method described in the first embodiment. to decide. In the above case, since the difference between the first error and the second error is −10, the first error is smaller than the second error. Therefore, the image compression decoding unit 17 leaves the target bit digit as it is. FIG. 16B is a diagram illustrating values of the pixel of interest and the surrounding pixels after the bit rounding process.

  In this way, in this embodiment, the value of the target bit digit is adjusted by giving priority to the continuity of the value of the target bit digit of the pixel before the target pixel, so even if the second error is smaller than the first error. In other words, if the difference between the first error and the second error is within the set range, the value of the target bit digit of the pixel of interest As it is. If the difference between the first error and the second error is outside the setting range, the value of the target bit digit of the target pixel is adjusted according to the first embodiment in order to prioritize the image quality retention after decoding.

  17 and 18 are flowcharts showing the flow of the bit rounding process for the pixel of interest in this embodiment. First, the image compression decoding unit 17 compares the target pixel of the code block, the values of the target bit digits of the previous pixel and the subsequent pixel of the target pixel. If the values of the target bit digits of the target pixel, the previous pixel and the rear pixel of the target pixel are all the same (step S11; YES), the image compression decoding unit 17 proceeds to step S21.

  When the values of the target bit digits of the front pixel, the target pixel, and the rear pixel are not all the same (step S11; NO), and the values of the target bit digits of the front pixel and the rear pixel are the same, that is, When only the value of the target bit digit is different (step S12; YES), if the value of the target bit digit of the previous pixel is “0” (step S13; “0”), the image compression decoding unit 17 sets the target pixel target. The bit digit value is changed from '1' to '0' (step S14). If the value of the target bit digit of the previous pixel is “1” (step S13; “1”), the image compression decoding unit 17 changes the value of the target bit digit of the target pixel from “0” to “1” ( Step S14). Then, the image compression decoding unit 17 shifts the process to step S21.

  On the other hand, the values of the target bit digits of the previous pixel, the target pixel, and the rear pixel are not all the same (step S11; NO), and the values of the target bit digits of the front pixel and the rear pixel are different (step S12; NO). The image compression decoding unit 17 changes the value A of the target bit digit of the target pixel by changing the value A in which all the lower bits are set to “0” or “1” and the value of the target bit digit of the target pixel without changing the value of the target bit digit of the target pixel. A value B that is all set to “0” or “1” is generated (step S16). Subsequently, the image compression decoding unit 17 calculates a first error that is an absolute value of the difference between the target pixel and the value A, and a second error that is an absolute value of the difference between the target pixel and the value B (step S17).

  Then, the image compression decoding unit 17 determines whether or not the same value continues for a predetermined number or more in the target bit digit of the pixel before the target pixel (step S31). If the same value continues for a predetermined number or more (step S31; YES), the image compression decoding unit 17 calculates the difference between the first error and the second error, and determines whether or not the difference is within the set range ( Step S32). When the difference between the first error and the second error is within the set range (step S32; YES), the image compression decoding unit 17 keeps the value of the target bit digit of the target pixel as it is (step S19). When the difference between the first error and the second error is outside the setting range (step S32; NO) and is equal to or less than 0 (step S18; YES), the image compression decoding unit 17 sets the value of the target bit digit of the target pixel. Leave as it is (step S19). If the difference between the first error and the second error is outside the setting range (step S32; NO) and greater than 0 (step S18; NO), the image compression decoding unit 17 changes the value of the target bit digit of the target pixel. (Step S20).

  If the same value does not continue in the target bit digit of the pixel before the target pixel for a predetermined number or more (step S31; NO), the image compression decoding unit 17 shifts the processing to step S18.

  Subsequently, the image compression decoding unit 17 changes all the bits lower than the target bit digit of the target pixel to “0” or “1” (step S21). When all the lower bits are set to “0”, the image compression decoding unit 17 generates a value A and a value B when all the lower bits are set to “0” in step S16. When all the lower bits are set to ‘1’, the image compression decoding unit 17 generates a value A and a value B when all the lower bits are set to ‘1’ in step S <b> 16.

  If the bit rounding process is not performed on all the pixels of the code block (step S22; NO), the image compression decoding unit 17 moves the target pixel (step S23) and shifts the process to step S11. When the bit rounding process for all the pixels of the code block is completed (step S22; NO), the image compression decoding unit 17 ends the process. Then, the image compression decoding unit 17 performs arithmetic coding after the bit rounding process is completed.

  As described above, the image compression decoding unit 17 changes the target bit digit of the target pixel without changing the target bit digit as it is and sets the lower bit to “0” or “1” and the lower bit by changing the target bit digit. A value B that is all set to “0” or “1” is calculated, and the absolute value (first error) of the difference between the value of the target pixel and the value A and the absolute value of the difference between the value of the target pixel and the value B (second) Error) and the same value continues in the target bit digit of the pixel before the target pixel, and the difference between the first error and the second error is within the setting range, the target bit digit of the target pixel By leaving the value as it is, it is possible to prioritize the continuity of values and improve the encoding efficiency. On the other hand, when the difference between the first error and the second error is outside the setting range, the value of the target bit digit of the target pixel is changed to minimize the value change before and after the bit change of the target pixel (after decoding). Encoding efficiency can be improved even in an image having a large density difference between adjacent pixels, such as a halftone image.

<Third embodiment>
In this embodiment, when the values of the target bit digit in each pixel before and after the target pixel of the code block are different and all the lower bits are set to “0” in the bit rounding process, the lower bit than the target bit digit of the target pixel. When the value is greater than or equal to the threshold value, the value of the target bit digit of the target pixel is set to “1”. When the value is smaller than the threshold value, the target bit digit value of the target pixel is left as it is and all the lower bits are set to “1”. The case where the value of the target bit digit of the target pixel is set to “0” when the value of the lower bit than the target bit digit of the pixel is equal to or smaller than the threshold value, and the value of the target bit digit of the target pixel is left as it is when the value is larger than the threshold value. To do.

  First, the image compression decoding unit 17 compares each value of the target bit digit of the target pixel, the previous pixel, and the subsequent pixel, and adjusts the value of the target bit digit of the target pixel according to the table shown in FIG. Among these, when the value of the target bit digit of the previous pixel and the subsequent pixel is different, the image compression decoding unit 17 adjusts the value of the target bit digit of the target pixel according to the value of the lower bit than the target bit digit of the target pixel. .

  This will be described with reference to FIG. FIG. 19A is a diagram illustrating the values of the target pixel and its surrounding pixels, where front pixel = 110011111B, target pixel = 10000111B, and rear pixel = 11011000B. First, the image compression decoding unit 17 acquires the value of the lower bit of the target bit digit of the target pixel. In the case of FIG. 19A, lower bits = 01111B = 7D. For example, when threshold = 6, since the value of the lower bit is equal to or greater than the threshold, the image compression decoding unit 17 sets the value of the target bit digit of the target pixel to “1”.

  Another example will be described with reference to FIG. In the case of FIG. 20A, lower bits = 0100B = 4D. Since the value of the lower bit is smaller than the threshold value = 6, the image compression decoding unit 17 keeps the value of the target bit digit of the target pixel as it is (“0”).

  That is, when all the lower bits are set to “0” in the bit rounding process, if the value of the lower bits is equal to or greater than the threshold value, the value of the target bit digit is set to “1”, so that the original value of the target bit digit is “0”. If it is', the value of the lower bit is rounded up. If it is less than the threshold, the value of the target bit digit is left as it is and all the lower bits become '0', so the lower bit value is rounded down. It will be.

  Conversely, when all the lower bits are set to “1” in the bit rounding process, if the value of the lower bit is equal to or less than the threshold, the value of the target bit digit is set to “0”, so that the original value of the target bit digit is “0”. In the case of “1”, the value of the target bit digit is rounded down from “1” to “0” instead of all the lower bits being “1”. On the other hand, when the value of the lower bit is larger than the threshold, all the lower bits are set to “1” with the target bit digit as it is.

  FIG. 21 is a flowchart showing the flow of the bit rounding process for the pixel of interest in this embodiment. First, the image compression decoding unit 17 compares the target pixel of the code block, the values of the target bit digits of the previous pixel and the subsequent pixel of the target pixel. If the values of the target bit digits of the target pixel, the previous pixel and the rear pixel of the target pixel are all the same (step S11; YES), the image compression decoding unit 17 proceeds to step S21.

  When the values of the target bit digits of the front pixel, the target pixel, and the rear pixel are not all the same (step S11; NO), and the values of the target bit digits of the front pixel and the rear pixel are the same, that is, When only the value of the target bit digit is different (step S12; YES), if the value of the target bit digit of the previous pixel is “0” (step S13; “0”), the image compression decoding unit 17 sets the target pixel target. The bit digit value is changed from '1' to '0' (step S14). If the value of the target bit digit of the previous pixel is “1” (step S13; “1”), the image compression decoding unit 17 changes the value of the target bit digit of the target pixel from “0” to “1” ( Step S14). Then, the image compression decoding unit 17 shifts the process to step S21.

  On the other hand, the values of the target bit digits of the previous pixel, the target pixel, and the rear pixel are not all the same (step S11; NO), and the values of the target bit digits of the front pixel and the rear pixel are different (step S12; NO). The image compression decoding unit 17 acquires a value of lower bits than the target bit digit of the target pixel (step S41). If all the lower bits are set to “0” in step S21 (step S42; YES), and if the value of the lower bits is equal to or greater than the threshold value (step S43; YES), the image compression decoding unit 17 determines the pixel of interest. The value of the target bit digit is set to “1” (step S44). On the other hand, in the case where all the lower bits are set to “0” in step S21 (step S42; YES), and the value of the lower bits is smaller than the threshold value (step S43; NO), the image compression decoding unit 17 determines the target pixel. The value of the target bit digit is left as it is (step S44).

  Conversely, if all the lower bits are set to “1” in step S21 (step S42; NO), and the value of the lower bits is equal to or less than the threshold value (step S46; YES), the image compression decoding unit 17 pays attention. The value of the target bit digit of the pixel is set to “0” (step S47). On the other hand, if all the lower bits are set to “0” in step S21 (step S42; NO) and the value of the lower bits is larger than the threshold value (step S46; NO), the image compression decoding unit 17 determines that the pixel of interest The value of the target bit digit is left as it is (step S48).

  Next, the image compression decoding unit 17 changes all the bits lower than the target bit digit of the target pixel to “0” or “1” (step S21). If the bit rounding process is not performed on all the pixels of the code block (step S22; NO), the image compression decoding unit 17 moves the target pixel (step S23) and shifts the process to step S11. When the bit rounding process for all the pixels of the code block is completed (step S22; NO), the image compression decoding unit 17 ends the process. Note that the image compression decoding unit 17 performs arithmetic coding after performing the processing of the flowchart shown in FIG. 13 for all code blocks of all tiles.

  As described above, the image compression decoding unit 17 sets the target bit when the lower bits are all “0” in the bit rounding process and the value of the lower bit than the target bit digit of the target pixel is equal to or greater than the threshold value. If the value of '1' is smaller than the threshold, the value of the target bit digit is left as it is, and all the lower bits are set to '1' in the bit rounding process, and the value of the lower bit than the target bit digit of the target pixel When the value of the target pixel is less than the threshold, the value of the target bit is set to “0”. When the value of the target bit is larger than the threshold, the value of the target bit digit is left as it is. Encoding efficiency can be improved even in an image having a large density difference between adjacent pixels, such as a halftone image, while suppressing deterioration in image quality.

DESCRIPTION OF SYMBOLS 1 Image compression apparatus 11 Control part 12 Storage part 121 Image compression program 13 Input operation part 14 Display part 15 I / F part 16 Network I / F part 17 Image compression decoding part (Conversion means, division | segmentation means, wavelet transformation means, quantization) Means, decomposition means, expansion means, bit rounding means, encoding means)

Claims (5)

Conversion means for converting the input image data into a color system comprising a luminance component and a color difference component;
Dividing means for dividing the converted image data into a plurality of tiles;
Wavelet transform means for performing subband decomposition by performing wavelet transform for each tile;
Quantization means for quantizing each subband;
Decomposition means for decomposing the quantized subbands into a plurality of blocks;
Expanding means for expanding the pixel value in the block into a plurality of bit planes from the most significant bit to the least significant bit for each block;
The pixel of interest according to the value of the pixel of interest that is one pixel in the block, the value of the target bit digit in each pixel before and after the pixel of interest, and the value of the lower bits of the pixel of interest than the target bit digit A bit rounding unit that adjusts the value of the target bit digit and changes all the lower bits to '0' or '1';
Encoding means for encoding the bit plane subjected to bit rounding processing by the bit rounding means;
An image compression apparatus comprising:   The bit rounding means determines that the value of the target pixel and the value of the target bit digit of the target pixel remain unchanged from the target bit digit when the values of the target bit digit in the pixels before and after the target pixel are different. The first error, which is the absolute value of the difference when the lower bits are all set to “0” or “1”, the value of the target pixel and the value of the target bit digit of the target pixel are changed, and A second error that is an absolute value of a difference from a value when all lower bits of the target bit digit are set to “0” or “1” is calculated, and when the first error is less than or equal to the second error, the attention is paid The value of the target bit digit of the pixel is left as it is, and when the first error is larger than the second error, the value of the target bit digit of the target pixel is changed, and all the lower bits are set to '0' or ' It is what changes to 1 '. Image compression device. In the bit rounding means, each value of the target bit digit in each pixel before and after the target pixel is different, and the same value is continuously greater than a predetermined number in the target bit digit of each pixel before the target pixel. If
A first absolute value of a difference between the value of the target pixel and the value of the target bit digit of the target pixel as it is with all the lower bits of the target bit digit set to '0' or '1' The absolute value of the difference between the error and the value obtained when the value of the target pixel and the value of the target bit digit of the target pixel are changed and all the lower bits of the target bit digit are set to '0' or '1' When the difference between the first error and the second error is within a predetermined setting range, the target bit digit of the target pixel is left as it is, and the first error and the second error are calculated. When the difference between the two errors is outside the set range and the first error is less than or equal to the second error, the value of the target bit digit of the target pixel is left as it is, and the first error and the second error If the difference is outside the set range and the first error is greater than the second error Wherein by changing the target bit digit value of the pixel of interest, the image compression apparatus according to claim 1 or 2 is to change the low-order bits to all '0' or '1'. The bit rounding means, when each value of the target bit digit in each pixel before and after the target pixel is different,
When all the lower bits of the target bit digit of the target pixel are set to “0”, the value indicated by the lower bit of the target bit digit of the target pixel is equal to or greater than a predetermined threshold value. Change the target bit digit to '1', and when the value indicated by the lower bit is less than the threshold, leave the value of the target bit digit of the target pixel as it is,
When all the lower bits of the target bit digit of the target pixel are set to “1”, the value indicated by the lower bit of the target bit digit of the target pixel is equal to or less than a predetermined threshold value. 2. The image compression apparatus according to claim 1, wherein the target bit digit is changed to “0”, and the value of the target bit digit of the target pixel is left as it is when the value indicated by the lower bit is larger than the threshold value. . Computer
Conversion means for converting the input image data into a color system comprising a luminance component and a color difference component;
A dividing unit for dividing the converted image data into a plurality of tiles;
Wavelet transform means for performing subband decomposition by performing wavelet transform for each tile;
Quantization means for quantizing each subband,
Decomposition means for decomposing the quantized subbands into a plurality of blocks;
Expanding means for expanding the pixel value in the block into a plurality of bit planes from the most significant bit to the least significant bit for each block;
The pixel of interest according to the value of the pixel of interest that is one pixel in the block, the value of the target bit digit in each pixel before and after the pixel of interest, and the value of the lower bits of the pixel of interest than the target bit digit Bit rounding means for adjusting the value of the target bit digit and changing all the lower bits to '0' or '1';
Encoding means for encoding the bit plane subjected to bit rounding by the bit rounding means;
Image compression program that functions as

Images