JPS62154875A - Multilevel image compressing system - Google Patents

Abstract

PURPOSE:To attain the efficient compression of an image by executing the gradation information reducing processing and picture element density reducing processing of respective image element data based on mode data determined in accordance with capacity difference between respective picture element data. CONSTITUTION:Plural picture element data out of respective picture element data in an image data having a halftone are collected as a block, each block is sorted into plural modes in accordance with the capacity difference between respective picture element data in the block and the gradation information reducing processing and picture element density reducing processing of the picture element data in each block are executed in accordance with parameters specified by the modes. Namely, a gradation information reducing circuit 6 changes the degree of gradation reduction at every block while referring the mode data M0-M3 in a mode selecting file, reduces the gradation reducing level of block data Db by detecting that a block with small values 0-3 of the mode data is a part with moderate density change or increases the gradation reducing level by deciding that a block with large values 0-3 is a part with sharp density change.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a gradation image compression method for compressing halftone image data by one bit.

Background of the Invention A predictive encoding method is known as a compression method for image data having halftones such as a film image. In this method, the pixel value at the current point is predicted from the pixel values that appeared in the past, and compression is performed by encoding the prediction error, which is the difference between the predicted value and the pixel value at the current point. To encode prediction errors, we take into account that the distribution of prediction errors follows a Laplace distribution, and assign short codes to prediction errors with a high occurrence order and long codes to prediction errors with low occurrence order. It is known to perform efficiency encoding.

Compression by this predictive coding is a continuous compression that can completely restore the original image as long as it is used in wide format.
Tl amphibiosis is not reduced. However, according to statistics regarding various images, the entropy of the pre-ΔIII error is about 4 bits in the case of an 8-bit image, and even if optimal encoding is performed, it can only be compressed to about the same level. This compression rate is hardly improved even if the prediction method is devised, and is considered to be the limit value of the pre-Δ+++ encoding method.

-・As another method of encoding, human visual characteristics and
There is block encoding that utilizes high correlation between adjacent pixels. The simplest block encoding method uses the average value of each block of nXn pixels as block information. Also, as something a little more elaborate,
Take the average value X a b of the pixel density of nXn blocks,
Each image Jc degree in the block is L from the average value X a b
Take the average value X, lII+ of the pixel group above the average value xab and the average value XABO of the pixel group below the average value xab, and calculate these average values Xd.
There is an encoding method that performs compression using bl and Xabo.

In this case, the required number of bits is 8n2 → n for an 8-bit image.
? +IB, and when n=4, 32/128=1/4
It can be compressed up to

However, in this block encoding method, since the block size is constant, the compression effect is insufficient in areas with small density changes, and conversely, in areas with large density changes, boundaries between blocks may occur due to coding errors. This results in a decrease in image quality. Regarding this point, there is a contradictory tendency that as the block size increases, the former improves and the latter worsens, and as the block size decreases, the latter improves and the former worsens.

If the block size is determined by balancing both, the compression ratio will be 9.
It becomes impossible to obtain satisfactory image quality.

A purpose of the present invention A purpose of the present invention is to make it possible to make image data with arbitrary density regions and density changes high in quality and at a high compression rate;
Furthermore, it is an object of the present invention to provide a compression method that can improve the encoding efficiency by centering the distribution of prediction errors in predictive encoding.

1. Structure of the Invention 1. The present invention collects a plurality of pieces of each pixel data of image data having halftones into a block, classifies the block into a plurality of modes according to the size of the difference between the pixel data in the block, In accordance with the parameters specified by the mode, gradation information reduction processing and pixel density reduction processing are performed on pixel data within each block - Example - Figure 0 is a block diagram showing an example of the present invention. The memory 1 stores halftone image data represented by multivalued gradation of each pixel read from an image (original image) of a film or the like. One block memory 2 stores one block of data D cut out from the image data in image memory 1. The block representative value memory 3 stores the representative value Dal of one block of data D transferred to the block memory 2 for one program.

The difference calculation circuit 4 sequentially reads data Db for each pixel from the block data D stored in the one block memory 2.
The difference S between the pixel and the adjacent pixel is calculated. The mode determining circuit 5 determines the mode FM from the maximum value of the difference Sa obtained by the difference needle 9 concave path 4. The 0 gradation information reduction circuit 6 changes the gradation of the block data Dh of the 1 block memory 2 according to the mode M. , this changed pronk data Dc is set to 1 again.
Insert into block memory 2. The difference calculation circuit 4 again calculates the difference Sa between adjacent pixels for the block data Dh written in +rf in the l block memory 2. The zero decimation circuit 7 uses the difference Sa from the difference calculation circuit 4 to set the mode M of the mode determination circuit 5. The encoder 8 performs variable-length encoding on the thinned-out difference data S0 according to the degree of thinning.

The predictive encoding circuit 9 preliminarily encodes the representative value data D, 2 stored in the block representative value memory 3. The coded block representative value data C3 obtained by the predictive coding circuit 9, the coded difference data CO obtained by the encoder 8, and the mode M obtained by the mode determination circuit 5 are combined and extracted as compressed image data 10. be done.

Processing operations for actual image data in such a configuration will be described in detail below.

The image data collected in the image memory 1 is divided into one block memory 2, and one block of the data is transferred. Here, the bronc minute 11.1] fi will be explained. When dividing an image into N×N minute blocks, if N is too small or too large, the properties of the blocks cannot be accurately grasped. However, it is very difficult to accurately determine the size of N because it largely depends on the individual characteristics of the target image. Therefore, in this embodiment, N=8 is used as a number with good separation and a number that can be easily converted into hardware. If N=8, the block data can be further divided into 4×4.2×2 blocks, which has the advantage of making it easier to move to more detailed processing.

Now, with N=8 set, 1 block memory 2 has 64
pixel data is transferred. At the same time, one pixel data is stored in the block representative value memory 3 as a representative value.

This representative value may be any of the 64 pixel data, but
In this embodiment, pixel data at the left corner of the block data (8×8) is set as a block representative value, and the pixel data is stored in the block representative value memory 3. This set of representative values is stored in the memory 3 as one file, and the size of this file is 1/64 of the original image data.

Additionally, this file can be made even smaller by lossless compression, and can ultimately be reduced to less than 1%.

Regarding the block data Dh consisting of 64 pixel data transferred to l block memory 2, if each pixel is named as follows, Xll X12 X13-
--XINX21 X22 .XNI

(a) Ignore the difference dX11 with respect to Xll.

(b) dXi 1=Xi l -X (i-1) l
to However, 2<=i<=8 (c) dXi j=Xi j−Xi (J−1). However, j<>1 The mode determining circuit 5 that takes in the set Sa of the differences dX obtained in this way selects the one with the largest absolute Sa, compares it with three preset thresholds, and determines the block data. are classified and stored in four modes. The correspondence between this threshold value and mode M is set as follows, for example.

Table 1: In this mode classification, a classification number t) of O to 3 is also filed for each block, and this file is hereinafter referred to as a mode selection file. Since this mode selection file requires 2 focus points for each block, the file size is (1/64) x (2/8) = 1/25B, and together with the representative value file, it is approximately the size of the image data. 1.5
%become.

Next, the gradation information reduction circuit 6 performs gradation reduction of the block data DIl in the mode M according to the mode data MO-M3 of the mode M from the mode determining circuit 5 as described above.

Care must be taken when reducing gradation to avoid creating false contours. Pseudo contours tend to occur in areas where density changes are gentle. Therefore, in this embodiment, the degree of gradation reduction is changed for each block while referring to the mode data MO-M3 of the mode selection file. That is, for a small block with a mode data value of O to 3, it is determined that the density change is gradual and the proac data D b
By reducing the amount of gradation reduction of block data Db and increasing the amount of gradation reduction of block data Db for the large proac with a large value O~3 as a part where the density change is evident, pressure can be reduced while suppressing the occurrence of false contours. Increase fi rate.

Specifically, the gradation reduction process by the gradation information reduction circuit 6 is performed so that the difference between adjacent pixels as the processed data DC becomes a multiple of the mode data MO-M3 numerical value 0 to 3 (gradation reduction variable). and then set the difference value with the adjacent pixel to a value between 0 and 3.
The divided value is stored in one block memory 2 as gradation-reduced data Dc.

For example, suppose that the numerical values of the block data Db of 4 zero 4 are arranged as shown below.

For this data Db, the difference from the block representative value at the upper left corner is as follows. Here, if the gradation reduction variable is 3 (M3), this difference is changed to . Therefore, the image data is as follows. If the difference is calculated from this using the same procedure as when creating the mode selection file, the upper left interval will be O, and dividing this by the tone reduction variable 3 will yield data ooi. As is clear from this data, the distribution of prediction errors (differences with adjacent pixels) is centered (0
), it is possible to improve the subsequent encoding efficiency and obtain a large compression ratio.

Next, with respect to the data Sd for which the gradation has been reduced and the difference has been calculated, the compression ratio is further improved by thinning out the data by the thinning circuit 7, that is, by reducing the pixel density. If this data is thinned out using a simple method such as simply thinning out every other pixel, the reproduced image will become blurred, and the image quality will deteriorate, although the false contours will not be affected. In this embodiment, the thinning ζ! takes into consideration the characteristics of one image! Then, the thinning circuit 7 performs thinning processing by changing the degree of thinning according to the mode data MO to M3 from the mode determining circuit 5. In other words, when one pixel is thinned out, the effect is greatest in areas where the image changes minutely, and in this case, the data difference between pixels within the block is considered to be large. When the numerical value of the mode data MO-M3 obtained from the mode selection file is large, the thinning is reduced, and when the numerical value is small, the thinning is increased, thereby reducing the occurrence of blurring and improving the compression ratio.

In addition, for thinning processing, the processing target block is 2X2.4X
The thinning effect can be further ensured by dividing the pixel data into four small blocks and starting from mode data with a representative pixel data value such as the upper left corner of the pixel data. Furthermore, pixels thinned out by one thinning process are interpolated during reproduction from compressed image data.

The thinning process and interpolation method from 8×8 blocks to 4×4 blocks will be specifically described below.

First, when 8×8 block data has the following coordinate array 4 Xll X12 X13 X14 X15
− − −XI5X21−−−−−−−−−X25−
--X28X31... From @1166+1 35X4
1-...No.@@@116X4j5X51・φ・
・ φ ・ Life ・ Meeting ・X5!1X81---
------X85----X8B In this block,
Divide into four 4×4 blocks and select the pixel X in the upper left corner of the block.
ll, X51, X15. Using the value of X55 as a representative value, thinning processing is performed from the mode data of each block.Next, an interpolation method from compressed image data that has been thinned out by one will be explained. The representative values of the 8×8 blocks above are each Yl
l, Y21. Y12. When extracted as compressed image data as Y22, the following margin Yll---Y12- - ・ ・ φ AIJ ・ ・ ・ e ・ ・Y21-
- - - Y22- - - Thinned out pixel A
To interpolate IJ, the representative values Yll, Y12. Y2
The weight of +, Y22 is the reciprocal of the distance from pixel AIJ, and the distance is the sum of the difference between pixel AIJ and each representative value on the horizontal axis and the difference on the vertical axis. That is, the coordinates of the representative value Yll are (Nl, N2), and the coordinates of the pixel AIJ are (N2, N2).
2), then the distance between the two @Dll is D11= (N 1
-N2) + (M-N2), and the weight of the representative value Yll for the pixel AIJ is 1/Dll. Similarly, the representative value Y12°Y21. Y22 (7) Distance #D12.
D21. Using all these distances @DIJ, the interpolated value of pixel AIJ is (Yll/DI 1.Y121012+Y211021
+Y22/D22) AIJ=□ (11011, 1/D12+1/D21+11022)
It is required as.

Note that supplementary rules for 2×2 blocks are similarly performed, but if 4×4 blocks are adjacent, no reference data exists.

In order to prevent this, interpolation is first performed for 4x4 blocks, and then interpolation is performed for 2x2 blocks.

Next, the encoder 8 uses the data S thinned out by the thinning circuit 7.
b is variable-length coded according to the coding pattern. The encoding pattern at this time is shown in the table below as an example of variable length code bits for data Sb thinned out based on the difference value from the representative value after gradation reduction processing.

Table 2 with blank space below: In this table, the part F(N-2) divides N-2 into the following ranges and encodes them.

Table 3 The data Cb encoded in this manner is used as a compressed data file, the data cd from the pre-all encoding circuit 9 is used as a representative value near aisle, and the mode data M from the mode determination circuit 5 is used for mode selection. These files become compressed image data 10 as three independent files or one combined file. For example, if the difference data Sb of the 4-tree 4 is oot and the mode M is 1, the compressed image data 10 becomes a linear data group.

Table 4 continues on next page Based on the examples described above, compression of test images and evaluation of reproduced images were performed.

Note that in the blocks that have been thinned out by one, focusing only on the extracted data, and assuming that each block is a 2×2 or 4×4 block, the same gradation reduction process as in the gradation information reduction circuit 6 is performed.
The resulting coded difference between adjacent pixels was used as compressed data. At this time, the difference corresponding to the block representative value is undefined, and since this value can be reproduced from the representative value file, it is excluded from the compressed data file. and,
To facilitate playback, the difference values are written to the compressed image data file in the same order as when creating the mode selection file.

The following table shows the state of thinning that performs such gradation reduction.

Table 5 The compression ratio in this example was 14% in Example 1 and 20% in Example 2.

Effects of the Invention As described above, according to the present invention, the gradation information reduction process and the pixel density reduction (thinning) process of each pixel data are performed from the mode data determined according to the size of the difference between the block-formed pixel data. Since compressed data is obtained by performing the following steps, it is possible to efficiently compress an arbitrary image while minimizing deterioration in reproduced image quality, and the encoding efficiency can also be improved.

[Brief explanation of drawings]

The drawing is a block diagram showing one embodiment of the present invention. 1... Image memory, 2... l block memory, 3...
...Block representative value memory, 4...Difference calculation circuit, 5
...Mode determining circuit, 6...Gradation information reduction circuit, 7
... thinning circuit, 8 ... encoder, 9 ... predictive coding circuit, 10 ... compressed image data.

Claims (1)

[Claims] 1) A plurality of pieces of each pixel data of image data having halftones are grouped together into a block, the block is classified into a plurality of modes according to the size of the difference between the pixel data within the block, and parameters specified by the mode are determined. A gradation image compression method characterized in that gradation information reduction processing and pixel density reduction processing are performed on pixel data within each block according to the following.