JP2513654B2 - Image coding device - Google Patents

Description

The present invention relates to an image encoding device that encodes multi-valued image data for each block.

[Prior Art] Conventionally, as a compression method of image data, there is a vector quantization method. This is to convert two-dimensional image data into one code data in m × m pixel block unit and compress it. A high compression rate can be obtained, decoding can be performed at high speed, and the image address is saved. However, since it is divided into m × m for encoding, m × m
There is a defect that a texture structure (moire) of m units is likely to appear.

[Problems to be Solved by the Invention] The present invention has been made in view of the above-mentioned conventional example, and utilizes the correlation between blocks to suppress image deterioration due to image data encoding and image data. It is an object of the present invention to provide an image encoding device that encodes by increasing the compression rate of.

[Means for Solving Problems] In order to achieve the above object, the image coding apparatus of the present invention has the following configuration. That is, input means for inputting multi-valued image data for each block, average value calculation means for calculating an average value of the multi-valued image data input by the input means for each block, and calculation by the average value calculation means. A binarizing means for binarizing the multi-valued image data for each block using the average value for each block and outputting pattern data for each block; and a pattern corresponding to the pattern data for each block. Selecting means for selecting a code for each block; storage means for storing a pattern code for each block selected by the selecting means; and encoding for each block based on the pattern code selected by the selecting means And encoding means for performing.

[Operation] In the above configuration, the average value of the input multi-valued image data is calculated for each block, and the calculated multi-valued image data for each block is binarized. Thus, the pattern data for each block is output, and the selecting means stores the pattern code according to the pattern data for each block for each block to store the pattern code of the target block, which is stored in the storage means. The distance between the pattern data extracted based on the pattern data corresponding to the pattern code of the adjacent block of the target block and the pattern data of the target block is calculated, and the pattern code corresponding to the pattern data of the minimum distance is set to the target. Select as the block pattern code. Coding is performed for each block based on the pattern code selected in this way.

Embodiments Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[Description of Encoding of Gradation Data (FIGS. 1 to 3)] FIG. 1 is a functional block diagram of this embodiment.

An input unit 100 inputs image data and decomposes it into a 4 × 4 pixel matrix shown in FIG. Here, each pixel a (0,0) to a (3,3) in FIG. 2 is an 8-bit monochromatic image.

FIG. 3 shows 16-bit data encoded by this embodiment, and the compression ratio in this case is 16 bits / (8 bits × 16 (pixels)) = 1/8. 30 of 16 bit data encoded
Is the 8-bit gradation data (GREY) and 31 is the 8-bit resolution data (SHAPE).

 How to obtain the gradation data 30 will be described below.

When the pixel data divided into the 4 × 4 matrix is input by the input unit 100, the average calculation unit 101 obtains the average value of 4 × 4 pixels according to the following equation.

In the variance value calculation unit 102, The dispersion value is obtained by The GREY coding unit 103 quantizes the average value A into 6 bits and the variance value into 2 bits to perform coding.

Note that the calculation of the gradation data 30 is not limited to the above-described method. For example, a combination having a high appearance frequency among combinations of the maximum value and the minimum value of the pixel matrix is coded and stored. Is also good.

[Description of Encoding of Resolution Data (FIGS. 1 and 4)] Hereinafter, a method of obtaining the resolution data 31 of FIG. 3 will be described.

The binarization unit 104 in FIG.
The pixel matrix is binarized based on For example, when the pixel matrix of FIG. 4A is input, the average value of the pixels calculated by the average value calculation unit 101 is 103.8.
The binarization pattern of FIG. 4B is obtained by binarizing based on this average value. This pattern is 16 bits (6553
6 values) can be taken, but this value is further reduced and encoded into 8 bits (256 values).

FIG. 5 shows a part of the appearance states (that is, 256 ways) of the 8-bit pattern stored in the storage unit 105. The pattern having a high appearance frequency in consideration of the statistical properties of the image. Shown to choose from.

A pattern similar to the binary pattern of 16-bit image data binarized as shown in FIG.
However, in the past, the pattern with the minimum distance, which will be described later, was selected, which causes discontinuity with the resolution data of the front, rear, left, and right blocks, which causes the deterioration of image quality and the texture structure. It was causing the outbreak.

Therefore, the block 61 on the preceding line and the preceding block 62 on the same line of the block 60 to be encoded are read from the storage unit 105, and the block 60 is encoded in consideration of the continuity of resolution data. This is shown in FIG.

FIGS. 7A and 7B are diagrams showing patterns read from the storage unit 105 according to an instruction from the control unit 106.

The pattern 62 in FIG. 7 (A) is a front block on the same line, and this pattern corresponds to the pattern 10 in the pattern classification of FIG. 70 shows a part of a pattern group which is continuous with the pattern of the block 62.
In the figure, the pattern classification number is shown.

Similarly, in FIG. 7 (B), the block 61 on the previous line corresponds to the pattern classification 2 in FIG. 5, and 71 pattern groups are shown as a part of the pattern group continuous to the pattern of the block 61. Has been done.

 As a result, continuous patterns are obtained for block 62, pattern classification (code) = 10,18,25,46, ... For block 61, pattern classification (code) = 2,18,40,46, ... Becomes

The above-mentioned candidate pattern codes are prepared in advance as a list for each pattern and stored in the storage unit 105. For example, in the case of the block 61 of the previous line, it is a continuous pattern of pattern 1; , 33,98 ... As a continuous pattern of pattern 2; 14,23,36,72 ... ・ ・ ・ ・ ・ In the case of front block 62, as a continuous pattern of pattern 1, 1,4,5, 12,13 ... As a continuous pattern of pattern 2, it is assumed that a list such as 1,4,5,12,13 ...

Therefore, patterns 18, 46 ... Are obtained as common patterns for the blocks 61, 62.

Next, the control unit 106 and the calculation unit 107 calculate the distance between the plurality of selected common patterns and the pattern binarized by the binarization unit 104, and select the pattern with the smallest distance. That is, the pattern of the block 60 and the candidate patterns 18, 46 ... Are compared for each pixel according to the following equation.

Here, is the exclusive OR, a _A (i, j) is the pixel data of the block 61, and a _X (i, j) is the pixel data of the block 62. As a result, L = 0 in comparison with the pattern 18 and L = 1 in comparison with the pattern 46. Therefore, the pattern 18 is selected as the pattern with the smallest distance.

Fig. 8 shows the state where pattern 18 is selected. As you can see from the figure, the resolution data becomes continuous, and
It can be seen that the continuity of the line drawing is improved and the image quality is improved.

In this way, when the resolution data is encoded into 8 bits and output to the encoded code output unit 108, the GREY encoding unit 1
Encoded code of gradation data from 03 (8 bits)
Both are configured to 16 bits and output.

[Explanation of pattern comparison and coding procedure (see FIG. 9, FIG.
(FIG. 10)] FIG. 9 is a configuration diagram of the control unit 106 and the storage unit 105.

In the figure, 90 is the ROM 91 shown in the flow chart of FIG.
CPU that performs various operations described below according to the control program of
Is. 92 is a RAM as a work area of the CPU 90.
Reference numeral 93 is a table memory for storing the pattern data shown in FIG. 5, and 94 is a memory for inputting the pattern data for each line from the binarizing unit 104 and storing it as code information.

FIG. 10 is a flow chart showing the encoding process of the resolution data of the CPU 90 of this embodiment.

In step S1, the code of the binarization pattern is input from the binarization unit 104. In step S2, the codes of the block 61 and the front block 62 on the previous line are read from the memory 94 as shown in FIG. In step S3, the code of the continuous pattern of the block 61 is read from the table memory 93, and similarly, in step S4, the code of the continuous pattern of the block 62 is read. At step S5, the common code of the code obtained at step S3 and the code obtained at step S4 is extracted, and at step S6 the calculation unit 107 obtains the binary matrix pattern from the binarization unit 104 and at step S5. The patterns corresponding to the generated codes are given, and the distance between the patterns is calculated.

In step S7, a pattern having the smallest distance is selected, and in step S8, coding is performed according to the code of the pattern data. Then, in step S9, the 8-bit code obtained by encoding the resolution data is output to the encoded code output unit 108 via 97, and the process is terminated.

As described above, according to the present embodiment, it is possible to provide an image data encoding method with a small texture structure (a block structure is hard to appear).

[Effects of the Invention] As described above, according to the present invention, the correlation between blocks is utilized to suppress the deterioration of the image due to the encoding of the image data, and to increase the compression rate of the image data to perform the encoding. The effect is that you can do it.

[Brief description of drawings]

FIG. 1 is a functional block diagram of this embodiment, FIG. 2 is a diagram showing the structure of a 4 × 4 pixel matrix, FIG. 3 is a diagram showing a 16-bit code encoded by this embodiment, and FIG. A) is a diagram showing an example of a pixel matrix, FIG. 4B is a diagram showing an example of binarizing the pixel matrix of FIG. 4A, and FIG. 5 is a part of the appearance state of an 8-bit pattern. FIG. 6, FIG. 6 is a diagram in which continuity of resolution data is considered, FIGS. 7 (A) and (B) are diagrams illustrating an example of continuous pattern extraction, and FIG. 8 is a pattern selection result in consideration of continuity. FIG. 9 is a block diagram of the control unit and the storage unit, and FIG. 10 is a flow chart showing the encoding process of the resolution data in the control unit. In the figure, 30 …… gradation data, 31 …… resolution data, 60 ……
Pixel matrix of interest, 61 …… Pixel matrix of previous line,
62 …… Prefixed pixel matrix, 90 …… CPU, 91 …… ROM, 92
... RAM, 93 ... Table memory, 94 ... Memory, 100 ...
... Input unit, 101 ... Average calculation unit, 102 ... Dispersion value calculation unit,
103 ... GREY encoding unit, 104 ... Binarization unit, 105 ... Storage unit, 106 ... Control unit, 107 ... Calculation unit, 108 ... Encoding code output unit.

Claims (1)

(57) [Claims] 1. Input means for inputting multi-valued image data for each block; average value calculation means for calculating an average value of the multi-valued image data input by the input means for each block; and said average value calculation means. Binarizing means for outputting the pattern data for each block by binarizing the multivalued image data for each block using the average value for each block calculated by Selecting means for selecting a corresponding pattern code for each block, storing means for storing the pattern code for each block selected by the selecting means, and for each block based on the pattern code selected by the selecting means And a coding unit that performs coding of, the selecting unit stores in the storage unit when selecting a pattern code of a target block. The distance between the pattern data extracted based on the stored pattern data corresponding to the pattern code of the adjacent block of the target block and the pattern data of the target block is calculated, and the pattern code corresponding to the pattern data of the minimum distance is calculated. An image coding apparatus, wherein the image coding apparatus is selected as a pattern code of the target block.