JPH08265755A - Image processing apparatus and image processing method - Google Patents

Abstract

(57) [Abstract] [Purpose] It is possible to minimize the deterioration of the image due to the compression of the image data and to stably perform the image compression at a high compression rate. To aim. A distribution shape determination unit 20 that detects a distribution shape of spatial frequency coefficients in a spatial frequency domain, a scan pattern storage unit 21 that stores a plurality of types of array conversion patterns performed by a zigzag scanning unit 94, and the distribution described above. The array conversion pattern corresponding to the output of the shape determination unit 20 is
A table selection unit 22 that reads out from the scan pattern storage unit 21 and sets the zigzag scanning unit 94 is provided, and the array conversion pattern is changed according to the distribution shape of the image data detected by the distribution shape determination unit 20. By making it possible to make the 0 data in the array long regardless of the type of the image, it is possible to prevent the coding efficiency from being extremely deteriorated due to the different distribution state of the 0 coefficient. .

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and an image processing method, and is particularly suitable for use in an image processing apparatus for compressing image information in the spatial frequency domain.

[0002]

2. Description of the Related Art Conventionally, H.264 has been used as a technique for performing capacity compression of image information. 261, JPEG, MPEG and JB
Standards such as IG are known. In particular, the JPEG algorithm is widespread for still images, and the MPEG algorithm is widespread for moving images. The following description is a compression method for a multi-valued still image.

First, the JPEG system will be described. In the description, the techniques of the quantization unit and the zigzag scanning unit are the units particularly related to the image processing apparatus and the image processing method of the present invention. The JPEG method is further DCT method and SP.
It is divided into two, the ATIAL system, and the DCT system is the mainstream.

This is because the capacity compression of the DCT system is an irreversible coding accompanied by deterioration of the image, and the image data is compressed by the quantization in the spatial frequency domain and the entropy coding. In the following description, JPEG
The standard DCT method will be described.

FIG. 13 is a block diagram showing the structure of the JPEG system. In FIG. 13, 91 is RGB-YUV.
A transform unit, 92 is a DCT transform unit, 93 is a quantization unit, 94 is a zigzag scanning unit, 95 is a Huffman coding unit, 96 is a quantization table, and 97 is a Huffman coding table.

The operation of each block will be described below. The RGB-YUV conversion unit 91 is provided to convert the chromaticity coordinate system of the image. That is,
A normal image is recorded in RGB chromaticity coordinates, but the RGB-YUV conversion unit 91 sets the RGB chromaticity coordinates in YUV.
Convert to chromaticity coordinate system. Here, the Y signal is luminance, U, V
The signal is a color difference signal.

The DCT transforming unit 92 performs a discrete cosine transform (DCT) on the image to transform it into a spatial frequency distribution. In the above conversion, the image is divided into small blocks of 8 × 8 pixels, and DCT processing is performed for each block. An example of conversion at this time is shown in FIG.

In FIG. 3A, 60 is a unit block of 8 × 8 size, and the number written in the square in the block is the distribution coefficient. Explaining each coefficient, 1
The row 1 column shows the lowest frequency and the ratio of the DC component.

That is, the column direction represents the horizontal frequency component and the row direction represents the vertical frequency component. In each direction, as the row (column) number increases, the proportion of higher frequencies is shown. The range of the coefficient is, for example, -127.
Is an integer value in the range of to 127.

Next, returning to FIG. 13, an explanation will be given. The quantization unit 93 performs quantization by dividing the spatial frequency distribution coefficient by the quantization table 96. The quantization process is performed on the block basis. Quantization table 96
Is generally as shown in FIG. 15, and is set such that the table value is small at low frequencies and large at high frequencies.

This utilizes the fact that human visual characteristics are insensitive to high frequencies, and reduces the amount of data in the high frequencies. However, this quantization results in the loss of data and thus in the degradation of the image. For this reason, there is a system in which the user can arbitrarily specify the quantization table 96 according to the type of image.

The zigzag scanning unit 94 rearranges the distribution coefficients in the block into a one-dimensional array. The rearrangement is performed in the order shown in FIG. FIG.
In (b), the coefficients are arranged in the order of the numbers written in the cells. A further schematic representation of this rearrangement is shown in FIG.
It becomes like (c). In FIG. 3C, the indication line 63
Indicates the order of sorting. Starting from the low-frequency component of row 1 and column 1, the high-frequency component of row 8 and column 8 is rearranged in a zigzag pattern.

The Huffman encoding unit 95 encodes the one-dimensional array output from the zigzag scanning unit 94. The content of the encoding process will be described with reference to the block diagram of FIG.

First, in the coefficient judging section 81, the distribution coefficient is 0.
Is determined. Then, the coefficient determination unit 81
If the determination result of is affirmative, the operation proceeds to block 82, and if not, the operation proceeds to block 83.

Reference numeral 82 is a run-length coding unit for outputting the number of consecutive coefficients having a value of 0.
A coefficient grouping unit 83 performs grouping according to the range of coefficient values and outputs a group number. A Huffman coding block 84 is for coding the outputs of the run length coding unit 82 and the coefficient grouping unit 83.

As is well known, Huffman coding assigns a short code length to a symbol having a high probability of occurrence. The redundant code is further compressed by this block.

Next, the storage structure of compressed data will be described with reference to FIG. In FIG. 16, 87 is data for one image, 85 is a header part, and 86 is an image frame part.

In the header section 85, SOI, DH
Codes such as T, DQT, DAC, DRI, COM, and APP are described. Among these codes, the one related to the above procedure is the Huffman table definition code DH.
T, the quantization table definition code DQT.

In the image frame section 86, compressed data is recorded following the frame start code SOF.
The compressed data is recorded separately for each of the Y signal, the U signal, and the V signal (first to third scans). further,
Within each scan, compressed data is stored for each block after the control code.

Next, the MPEG method, which is a moving image compression method, will be described. The MPEG system handles a plurality of image frames collectively as a GOP (group of pictures).

One GOP is an I picture, P
It is composed of three types of image frames, a picture and a B picture. The arrangement of the frames in the GOP is IBBPBBPBBPBB.

Only one I-picture exists in one GOP, and compression is performed in the same procedure as in the JPEG system. That is, compression is performed and recorded in the order of RGB → YUV conversion → DCT conversion → quantization → zigzag scan → Huffman coding.

There are three P pictures in the GOP, and they are created by temporally performing interframe prediction from I pictures and P pictures. B pictures are I that are temporally preceding and following.
It is created by performing inter-frame prediction from a picture and a P picture.

[0024]

However, in the above-mentioned conventional image processing apparatus, since the uniform processing is performed regardless of the content of the image data, there is a problem that the compression rate is lowered depending on the type of the image. In particular, in the run-length encoding unit 82, the distribution state of 0 coefficients differs depending on the type of image, so that the encoding efficiency may be extremely deteriorated.

In view of the above problems, it is an object of the present invention to minimize the deterioration of an image due to image compression and to enable stable image compression at a high compression rate. And

[0026]

The image processing apparatus of the present invention comprises a quantizing means for quantizing in the spatial frequency domain, and an array for rearranging the distribution coefficients quantized by the quantizing means into a one-dimensional array. An image processing apparatus having a conversion unit for compressing image data by encoding a one-dimensional array output from the array conversion unit, a distribution for determining a distribution shape of coefficients in the spatial frequency domain. Shape determining means, scan pattern storing means in which a plurality of types of array conversion patterns performed by the array converting means are stored, and array conversion patterns corresponding to the output of the distribution shape determination are read from the scan pattern storing means. And array conversion pattern setting means for setting the array conversion means.

Another feature of the image processing apparatus of the present invention is that it further comprises area dividing means for dividing an image into a plurality of areas, and the coefficient division is performed for each area divided by the area dividing means. The process of detecting the distribution shape is performed, and the optimum scan pattern is set for each area.

Another feature of the image processing apparatus of the present invention is that the above-mentioned encoding is entropy encoding.

A feature of the image processing method of the present invention is that quantization in the spatial frequency domain is performed, and array conversion processing for rearranging the quantized distribution coefficients into a one-dimensional array is performed. In an image processing method for compressing image data by encoding a rearranged one-dimensional array, a distribution shape determination processing for determining a distribution shape of coefficients in the spatial frequency domain and an array conversion processing are performed. A plurality of types of array conversion patterns used for this purpose are stored in the scan pattern storage unit, and an array conversion pattern corresponding to the result of the distribution shape determination process is read from the scan pattern storage unit and the array An array conversion pattern setting process that is set in the array conversion means for performing the conversion process is performed.

Another feature of the image processing method of the present invention is that the image is further divided into a plurality of areas, and area distribution processing is performed to detect the coefficient distribution shape for each of the divided areas. By doing so, an optimum scan pattern is set for each area.

[0031]

Since the present invention has the above technical means, the array conversion pattern can be changed according to the distribution shape of the image data detected by the distribution shape determination, and the 0 data in the array is irrespective of the type of the image. Will be able to take longer. This minimizes image deterioration and enables stable image compression at a high compression rate.

Further, according to another feature of the present invention, the image compression utilizing the characteristics that the blocks in the same region have similar hues and textures and the distribution shapes of the coefficients are substantially the same, to the maximum extent. It becomes possible to do.

[0033]

Embodiments of the image processing apparatus and the image processing method of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram for explaining the detailed configuration of the image compression / decompression unit 16 used in the image processing apparatus of this embodiment.

In FIG. 1, 20 is a distribution shape determination unit, 21 is a scan pattern storage unit, and 22 is a scan table selection unit. Further, 91 is an RGB-YUV conversion unit, and 92 is D
CT conversion unit, 93 quantization unit, 94 zigzag scanning unit, 95 Huffman coding unit, 96 quantization table,
Reference numeral 97 is a Huffman coding table. RGB-
The YUV conversion unit 91 to the Huffman coding table 97 have the same configurations as the configurations described in the related art, and thus detailed description thereof will be omitted.

First, the operation of the entire apparatus will be described. When the image is compressed, the image data is input from the RGB-YUV converter 91, the image is compressed, and the compressed image data is output from the Huffman encoder 95. I am trying to do it. Further, when the image is expanded, the compressed image data is input from the Huffman encoding unit 95 to expand the image, and the expanded image data is output from the RGB-YUV conversion unit 91.

<Description of Processing of Distribution Shape Determining Section (FIGS. 4 and 8)> Next, the operation of the distribution shape determining section 20 will be described. The distribution shape determination unit 20 determines the shape by comparing the shape of the distribution coefficient block output from the DCT conversion unit 92 with a preset pattern.

FIGS. 4A to 4C show examples of distribution patterns. In each pattern, an 8 × 8 size block is divided into two regions. For example, in FIG. 4A, the first areas (1) and (73) indicated by halftone dots,
It is divided into the second areas (2) and (76) of the white background. Similarly in FIGS. 4B and 4C, 74, 75
Is the first area (1), and 77 and 78 are divided as the second area (2).

The correspondence between these distribution patterns and actual images will be described. First, when the image contains many distinct horizontal stripes, the coefficient block exhibits an A distribution. Further, when many clear vertical stripes are included, the coefficient block shows a B distribution. In other cases, landscape etc. shows C distribution. Further, the coefficients are concentrated in the first area (1) of the corresponding distribution pattern, and the coefficient is almost 0 in the second area (2).

In the following description, the shape determination processing of the distribution coefficient block using the distribution pattern of FIG. 4 will be described. First, the absolute values of the coefficient values included in the first area (1) of each distribution pattern are integrated for the block. And
The integrated values of the respective distribution patterns are compared, and the pattern having the maximum integrated value is determined as the block distribution shape. That is,
The coefficient distribution shape of the block is determined by the magnitude of the integrated value in the first area (1).

FIG. 8 is a flow chart showing the flow of processing of the above-mentioned determination procedure, which will be described with reference to FIG. First, in step S100, for the first distribution pattern A, the coefficients included in the first region (1) are integrated.
Assuming that the coefficient value is C (n) and the integrated sum is SumA,

[0041]

[Equation 1]

It is calculated by

In the next step S101, similarly, the second
The cumulative sum SumB of the coefficients is obtained for the distribution pattern B of. Further, in the next step S102, the integrated sum SumC is obtained for the third distribution pattern C.

Next, in step S103, the integrated sum Su
The largest one among mA, SumB, and SumC is determined. In the following step S104, the determined distribution shape is output, and the process ends. Step S1 described above
The coefficient distribution shape of the block can be determined by the processing of 00 to step S104.

<Description of Processing by Scan Table Selection Unit 22 (FIGS. 5, 6, 9)> Next, the operation of the scan table selection unit 22 will be described. Here, the scan pattern corresponding to the output of the distribution shape determination unit 20 is read from the scan pattern storage unit 21 and set in the zigzag scan unit 94.

FIGS. 5A to 5C are views for explaining the contents of the scan pattern stored in the scan pattern storage unit 21. In each scan pattern, they are arranged in the order shown by the instruction lines 66, 67, 63.
The pattern of FIG. 5A is selected when the distribution shape determination unit 20 has the A distribution as a result.

The pattern of FIG. 5B is selected when the pattern has B distribution, and the pattern of FIG. 5C is selected when it has C distribution. In FIG. 5A, scanning is performed in the row direction starting from the coefficient in the first row and the first column. In the case of the A distribution, the coefficients are concentrated in the first column, so 0 will continue from the second column onward.

That is, considering the one-dimensional array after scanning, 0s are continuous from the 9th element onward. On the other hand, FIG.
The pattern (b) is scanned in the column direction. Since the coefficients in the case of B distribution are concentrated in the first row, the 9th and subsequent array elements can be made continuous with 0s by performing the scan in FIG. 5B. The pattern of FIG. 5C is similar to the conventional one, and is performed in the diagonal direction.

FIG. 9 is a flow chart showing the flow of processing for selecting a scan pattern, which will be described below with reference to FIG. First, in step S110, the distribution shape is input from the distribution shape determination unit 20. Next, step S
At 111, it is determined whether the block has an A distribution,
If affirmative, the process proceeds to step S112, and if negative, the process proceeds to step S113.

If the process proceeds to step S112, the pattern shown in FIG. 5A is set in the zigzag scanning unit 94, and the process ends. If the process proceeds to step S113, it is determined whether the block has a B distribution. If the result is affirmative, the process proceeds to step S114, and if the result is negative, the process proceeds to step S114.
Proceed to 115.

If the process proceeds to step S114, the pattern shown in FIG. 5B is set in the zigzag scanning unit 94 and the process ends. If the process proceeds to step S115, the pattern shown in FIG. 5C is set in the zigzag scanning unit 94 and the process ends.

Next, how the one-dimensional array is stored after scanning will be described with reference to FIGS. Figure 6
In (a), 50 is the result of performing the scan of FIG. 5 (a). It can be seen that the block matrix is stored in the row direction in the array element SD (n).

In FIG. 6B, 51 is the result of the scan of FIG. 5B. It can be seen that the array elements SD (n) are stored in the column direction. FIG. 6 (c)
In the figure, 52 is the result of the scan of FIG. Through the above processing, the processing of selecting the scan pattern according to the distribution shape can be performed.

<Description of Storage Form (FIGS. 7 and 16)> The data storage structure after Huffman coding will be described below. In this embodiment, the entire storage structure is similar to that shown in FIG. In the figure, the storage structure in the unit block is performed in the order shown in FIG.

In FIG. 7, 40 is compressed data for one block. 41 is the scan pattern number,
Huffman codes corresponding to the A distribution, the B distribution, and the C distribution are recorded. 42 to 48 are compressed data obtained by Huffman coding the array data shown in FIG. Figure 6
By adopting the storage form of, the scan pattern can be selected even at the time of decompression.

As described above, according to this embodiment,
The pattern of zigzag scanning can be changed by determining the distribution shape of image data. As a result, the 0 data in the array can be taken long regardless of the image type. This enhances the efficiency of run-length encoding, so that it is possible to perform image compression at a high compression rate regardless of the type of image.

<Explanation of Apparatus Configuration (FIG. 2)> Next, an example in which an image processing apparatus is constructed by combining the image compression means 16 of this embodiment with a personal computer will be explained.

FIG. 2 is a block diagram showing a schematic configuration of the image processing apparatus in this embodiment. In the figure, reference numeral 1 is a CPU that controls the entire apparatus, and executes processing such as calculation and I / O control according to a program stored in the memory unit 3.

The peripheral device controller 2 is used as a set with the CPU 1, and is an I / O required to control the peripheral devices.
(Serial communication, parallel communication, real-time clock, timer, interrupt control, DMA control, etc.) are controlled.

The memory unit 3 serves as a main memory, and DR
It includes a memory such as an AM, a cache RAM, and a ROM. Four
Is a hard disk drive that stores user data, device settings, and image data.

Reference numeral 7 is a keyboard control section, and 8 is a keyboard. Reference numeral 5 is an FDD control unit, and 6 is a floppy disk drive. Reference numeral 9 is a display unit, and in the display unit 9, 10 is a display control unit,
This is for performing a process of sequentially reading display data from the VRAM 11 and transferring the data to the liquid crystal panel 12 while performing gradation conversion and the like.

Further, the display control unit 10 controls the CPU 1 to V
Access to RAM 11 and display unit 6 from VRAM 11
Path control is also performed to prevent collision of access to data transfer.

Furthermore, AND, OR, EXO with a pattern preset for the contents of the VRAM 11
Performs logical operations such as R. In the present embodiment, the liquid crystal panel 12 displays information of various files and also image data.

Reference numeral 13 is a communication unit. In the communication unit 13, 15 is a communication interface,
14 is a communication control unit. The communication standard is R
It has interfaces for serial communication such as S-232C and Ethernet, and parallel communication such as Centronics and SCSI. Then, in addition to various data such as text, image data is input / output. Next, 16 is an image compression / decompression unit that compresses and decompresses still images.

<Explanation of Image Compression / Expansion Unit (FIG. 1)> The image compression / expansion unit 16 of this embodiment compresses image data using RGB-YUV conversion, DCT conversion, quantization and Huffman coding. I do. Here, the distribution shape of the coefficients is detected for each block after the DCT processing, and the operation of switching the zigzag scan pattern according to the distribution shape is performed.

The present invention is not limited to the above-mentioned embodiments, but can be widely applied. For example, although the distribution shape determination unit 20 of the embodiment inputs from the DCT conversion unit 92, the distribution shape determination unit 20 may input from the quantization unit 93.

Further, in order to perform the distribution shape determination in the distribution shape determination unit 20, the integrated value of the pixel values is used in the above-mentioned embodiment, but the number of pixels whose value is not 0 may be used. Furthermore, although the pattern shown in FIG. 5 has been described as an example of the scan pattern, any scan pattern can be used.

For example, instead of the reciprocal scanning pattern as shown in FIG. 5, scanning may be performed in one direction for each row and each column. In addition, in the above embodiment, the process of switching the scan table according to the distribution shape has been described, but the quantization table may be further switched.

In the quantization table at this time, values to be quantized are arranged according to the distribution shape. For example, in the “first area (1)” in FIG. 4, the quantized value is reduced and the distribution coefficient is saved.

In the other portion, that is, in the "second area (2)", the quantization value is increased so that unnecessary coefficients are eliminated. By such processing, it is possible to improve the compression efficiency while suppressing the deterioration of the image. further,
The present invention can also be applied to video compression.

That is, in the compression method such as MPEG, in the compression of the I picture, the processing of DCT conversion-quantization-zigzag scan is commonly performed. Therefore,
The processing of this embodiment can be used as it is for the compression of the I picture, and the compression efficiency of the moving image can be improved.

Next, a second embodiment of the present invention will be described.
In the above-described embodiment, the process of detecting the coefficient distribution shape for each block and changing the zigzag scan pattern has been described. In the second embodiment, first, a process of dividing an image into regions and detecting a coefficient distribution shape for each region will be described. That is, the blocks in the same region have similar hues and textures, and the distribution shapes of the coefficients are almost the same.

Therefore, if the image is first divided into a plurality of areas and an appropriate zigzag scan pattern is set for each area, the same effect as that of the above-described embodiment can be obtained. Furthermore, since the scan pattern code is written only once per region in the compressed data, the compression rate of the entire image can be further improved.

<Explanation of Example of Area Division (FIG. 10)> In the present embodiment, the area division is first performed in block units.
For this area division, known techniques such as a method using a density histogram and a method using clustering can be applied.

FIG. 10 is a diagram showing how the image is divided into areas, which will be described with reference to FIG. 20 in the figure
Reference numerals 0 and 201 are sample images, and 202 is a unit block composed of 8 × 8 pixels.

In FIG. 10A, reference numerals 203 to 205 are objects included in the image. On the other hand, FIG.
(B) is a result of performing the area division in block units. In FIG. 10, the image is divided into four regions 206-209. The blocks in which the same numbers are entered belong to the same area.

<Explanation of Processing Operation (FIG. 11)> FIG.
12 is a flowchart showing a flow of image compression processing of the present embodiment, which will be described with reference to FIG. 11. In this embodiment, the image area is first divided. Next, the distribution shape is detected for each area. Then, a scan pattern setting process is performed according to the detection result.

In this embodiment, software such as RGB-YUV conversion, DCT conversion, quantization, zigzag scanning, and Huffman coding, which are realized by hardware in the first embodiment, are realized by software. . In the figure, in step S150, area division is performed in block units.

A known method such as a density histogram method or a clustering method can be used as the area dividing method.
The description here is omitted. In step S151, RGB-YUV conversion and DCT conversion processing is performed on the entire screen. In subsequent steps S152 to S155, processing is performed for each of the divided areas.

In step S152, the distribution shape is detected using the same method as in the first embodiment. The block in the center of the area is used as the block for calculation. The block may be selected at the corners or end faces of the area.

Further, the entire area may be calculated and the average may be obtained. In the next step S153,
The scan pattern is set according to the distribution shape. This setting is similar to the procedure described in the first embodiment.

In step S154, quantization processing, zigzag scanning processing, and Huffman coding are performed on the blocks in the processing target area. In step S155, the compressed data is output. This data is output to the hard disk drive 4 and the communication unit 13. The output format will be described later.

In step S156, it is determined whether or not the conversion of all areas of the image is completed. If the determination is affirmative, the process is terminated, and if the determination is negative, the process proceeds to step S152. With the above processing, it is possible to perform the processing of detecting the distribution shape for each area of the image and switching the scan pattern.

<Description of Storage Structure of Compressed Data (FIG. 12)
> FIG. 12 is a diagram showing a storage structure of compressed data. FIG. 12A corresponds to the structure of the nth scan in FIG. Within each scan, the blocks are recorded in the order of arrangement. Reference numeral 210 represents the storage state of the line indicated by BB in the image of FIG. The same line is 211-213
It is composed of three areas. The storage format of each area is
It is as shown in FIG.

In the figure, 221 is an area identifier, and 22
2 is a scan pattern number used for zigzag scanning within the area. 223 to 226 are compressed data for each block. The storage format of each block is shown in the 12th (c). 42 to 48 are coefficient data after Huffman coding.

By storing as described above, the number of times the scan pattern number is recorded can be reduced. In addition, depending on the configuration of the device, the number of divided areas and the scan pattern to be used can be stored in the header section. At this time, the number of times the scan pattern number is recorded is only once per area, and the compression rate can be further improved.

As described above, according to this embodiment,
By detecting the distribution shape of the coefficient for each area and selecting the scan pattern, the same effect as that of the above-described embodiment can be obtained.

Furthermore, in the present embodiment, the number of times the scan pattern code is written can be reduced, so that the compression rate can be further improved. At the same time, the distribution shape determination process needs to be performed for the number of regions in the image, so that the processing speed can be increased.

[0089]

As described above, according to the present invention, the array conversion pattern can be changed according to the distribution shape of the image data detected by the distribution shape determination. It becomes possible to take 0 data for a long time. As a result, it is possible to perform image compression at a high compression rate regardless of the type of image, and it is possible to minimize deterioration of the image.

Further, according to another feature of the present invention, the image compression which makes the maximum use of the characteristic that the blocks in the same region have similar hues and textures and the distribution shapes of the coefficients are also substantially the same. Can be performed, and the compression rate of the entire image can be further improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image compression / decompression unit according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a schematic configuration of an image processing apparatus in the first embodiment.

FIG. 3A is a diagram for explaining the distribution coefficient after DCT processing in the conventional technique, and FIGS. 3B and 3C are diagrams for explaining the scan order of the zigzag scan in the conventional technique.

FIG. 4 is a diagram illustrating a setting pattern of a distribution shape determination unit according to the first embodiment.

FIG. 5 is a diagram illustrating a setting pattern of a zigzag scanning unit in the first embodiment.

FIG. 6 is a diagram illustrating the arrangement of array data after zigzag scanning in the first embodiment.

FIG. 7 is a diagram illustrating a storage structure of block data after Huffman coding according to the first embodiment.

FIG. 8 is a flowchart showing a processing flow of a distribution shape determination unit in the first embodiment.

FIG. 9 is a flowchart showing a processing flow of a scan pattern selection unit in the first embodiment.

FIG. 10 is a diagram showing how an image is divided into areas according to the second embodiment.

FIG. 11 is a flowchart showing the flow of image compression / decompression processing in the second embodiment.

FIG. 12 is a diagram illustrating a storage structure of compressed data according to a second embodiment.

FIG. 13 is a block diagram showing a configuration of an image compression / decompression unit in the related art.

FIG. 14 is a block diagram showing a configuration of a Huffman encoding unit in a conventional technique.

FIG. 15 is a diagram illustrating a state of a quantization table in the related art.

FIG. 16 is a diagram illustrating a storage structure of compressed data according to a conventional technique.

[Explanation of symbols]

 1 CPU 2 Peripheral controller 3 Memory part 4 Hard disk drive 5 FDD control part 6 FDD drive 7 Keyboard control part 8 Keyboard 9 Display unit 10 Display control part 11 VRAM 12 Liquid crystal display 13 Communication unit 14 Communication control part 15 Communication interface 20 Distribution shape judgment 21 scan pattern storing section 22 scan table selecting section 81 0 coefficient determining section 82 run length coding section 83 coefficient grouping 84 Huffman coding block 85 header section 86 frame section 16 image compression / decompression section 91 RGB-YUV conversion section 92 DCT Transformation unit 93 Quantization unit 94 Zigzag scanning unit 95 Huffman coding unit 96 Quantization table 97 Huffman coding table

Claims (5)

[Claims] 1. A quantization means for performing quantization in a spatial frequency domain, and an array conversion means for rearranging the distribution coefficients quantized by the quantization means into a one-dimensional array. In an image processing apparatus that encodes an output one-dimensional array to compress image data, a distribution shape determination unit that determines a distribution shape of coefficients in the spatial frequency domain and an array conversion unit perform the above. A scan pattern storage unit that stores a plurality of types of array conversion patterns, and an array conversion pattern setting that reads the array conversion pattern corresponding to the output of the distribution shape determination from the scan pattern storage unit and sets it in the array conversion unit. An image processing apparatus comprising: 2. An area dividing means for dividing an image into a plurality of areas is further provided, and a process of detecting a coefficient distribution shape is performed for each of the areas divided by the area dividing means, and the optimum shape is obtained for each area. The image processing apparatus according to claim 1, wherein a scan pattern is set. 3. The image processing apparatus according to claim 1, wherein the encoding is entropy encoding. 4. Quantization in a spatial frequency domain, array conversion processing for rearranging the quantized distribution coefficients into a one-dimensional array, and further encoding the rearranged one-dimensional array. In an image processing method for compressing image data, a distribution pattern determination process for determining a distribution pattern of coefficients in the spatial frequency domain, and a plurality of types of array conversion patterns used for performing the array conversion process are stored in a scan pattern. Array for storing scan pattern storage process and array conversion pattern corresponding to the result of the distribution shape determination process from the scan pattern storage unit and set in the array conversion means for performing the array conversion process. An image processing method, characterized in that conversion pattern setting processing is performed. 5. An area dividing process for dividing an image into a plurality of regions is further performed, and a process for detecting a coefficient distribution shape is performed for each of the divided regions to set an optimum scan pattern for each region. 4. The method according to claim 3, wherein
The image processing method described in.