JPS60176180A - Picture data compression system - Google Patents

Abstract

PURPOSE:To obtain a data compression ratio efficient for various pictures and to improve processing efficiency by deciding automatically the most suitable number of differential processings at every partial area in accordance with the partial data characteristic of a picture. CONSTITUTION:After a picture written from a film reader unit 61 is A/D-converted, it is stored in a magnetic disk 63. An image processor 62 chops a partial picture from the picture in the magnetic disk 63, executes differential processing at every partial area and compresses data. Then the processor 63 evaluates the result after the differential processing, and decides whether or not additional differential processing is necessary. When the additional differential processing is necessary, the differential processing is executed, its result is evaluated again. When the additional differential processing is unnecessary, data are encoded and their partial pictures are stored in, for instant, an optical disk 64, etc. By repeating such processing by one picture, compressed pictures are stored in the optical disk 64.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to an image data compression method, and particularly to an image data compression method suitable for efficiently compressing large amounts of data such as digital X-ray images. Regarding the method.

[Background of the invention]

In band compression processing of image data, typified by conventional differential processing, uniform processing is performed on one image. Therefore, even if data bands or characteristics differ within one image, the same processing is performed, resulting in poor data compression efficiency.

Therefore, not only does the compression ratio differ for each image, but in order to increase the compression ratio, it is necessary to repeat experiments by changing the number of differences and the like through trial and error.

[Purpose of the invention]

The object of the present invention is to, depending on the partial data characteristics of an image,
The purpose of this invention is to provide a method for efficiently compressing large amounts of image data by automatically determining the optimal number of differential processing steps.

[Summary of the invention]

Normally, data compression processing is divided into three processes as shown in FIG. The first is scanning 12 of two-dimensional image data.
, and two-star scanning in the horizontal direction is common. Second
is achieved by compressing the data bandwidth (dynamic range).
Differential predictive coding (abbreviated as DPCM) and the like are widely used, taking advantage of the high correlation between pixels. The third is data encoding, and natural codes, Huffman codes, etc. are typical examples.

The present invention relates to the second process described above.

The basic idea of the present invention will be explained below with reference to FIG. Conventionally, the data band was compressed by applying DPCM once to the entire image as shown in 21, but in the present invention, the image is divided into several partial areas as shown in 22, For each region, for example, the original image itself was used. 9 (216 DPCM') difference is executed twice (215 DPCM''), and the number of repetitions is changed for each partial area to achieve a higher compression ratio.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to FIGS. 3 to 5.

FIG. 3 shows the processing flow of the present invention. First, in step 31, the image is divided into partial areas. In the following explanation, an example in which the area is divided into rectangular areas will be explained, but it is also possible to divide the area into meaningful areas (for example, parts of the human body) using other advanced area division methods.

In step 32, differential processing is performed for each partial region to compress the data. The differential processing is the simplest processing of a so-called predictive coding method, and corresponds to previous value prediction. That is, when the current value is Xi, the previous value X1-1 is multiplied by the prediction coefficient a to calculate the prediction error ε. Tsumashi, ε
l""XI-aXI-1. Here, the case where a=1 is described as an example.

In step 33, the result after the difference processing is evaluated, and it is determined whether the difference processing should be performed again. Specifically, the following evaluation criteria are used. Regarding this process,
The following three examples will be described for each encoding processing method following differential processing.

(1) When saving the difference data itself In this case, the maximum value of the image data after the difference processing is searched, and if that value is smaller than the previous value, the difference processing is repeated again.

(2) Natural code In this case, the compression ratio depends on the type of data. Therefore, the data types are counted from the difference data, and if the previous value is smaller than the previous value, further difference processing is performed.

(3) Huffman code In this case, the compression rate depends on the number of data types and their frequency distribution. In this case, the r' error is evaluated based on the frequency distribution shown in FIGS. 4a and 4b.

First, count the data types, and if there are fewer data types than last time, perform differential processing and return. Also, when the data type is almost the same as the previous time, the variance value for that frequency, that is, the frequency, is Dl
When (N is the number of data), a judgment is made based on the size of fcDa''.In the example in Figure 4, comparing the distributions of a and b, assuming Huffman coding, in the case of fc , a has a higher compression ratio. Looking at Do'' in this case, a is about 36 and b is about 4
．． There is a huge difference between 8 and 8. Therefore, when Dσ2 is larger than the previous time, the differential processing is returned.

In addition to the method described above, it is also possible to make a determination by comparing the compression ratios after actually converting into 7-man code.

In step 34, it is determined whether to repeat the differential processing. In step 35, it is determined whether all partial areas have been completed. In step 36, the partial area to be processed next is set.

Next, the effect of changing the band compression method for each partial area will be explained using a specific numerical example with reference to FIG. 5
1.52 is the original partial image, 511° and 512 are the values that have been subjected to differential processing once, and 522 are the values that have been subjected to differential processing twice. First, when comparing 51 and 511, it can be seen that the band is sufficiently compressed by one difference. The absolute bandwidth has been compressed by more than 60 bits, from a maximum of 20 bits (5 bits) to 4 bits (3 bits), and the number of data types has also been reduced from 13 bits to 4 bits, to about 30 bits.

52 indicates a case where the width of density value fluctuation is larger than 51. In 521, which is the result of one-time difference processing, the absolute band ranges from a maximum of 2'00 (8 pits) to 40 (6 pits). bit) to about 75 *, the data type goes from 13 to 4, about 30
It has decreased to chi. In addition, the results of the second difference processing show that the absolute bandwidth has changed from maximum 20G (8 bits) to -10 (5 pits), about 62 inches, and the data type has changed from 13 to 3, about 11
FIG. 6 shows an example of a system configuration for realizing the processing of the present invention, and FIG. 7 shows a correspondence relationship between the configuration diagram of FIG. #I6 and the processing process. The image read from the film reader unit 61 is A/D converted and then transferred to the magnetic disk 6.
3. The image processor 62 cuts out a partial image from the image 63, performs an image compression process as shown in FIG. 3, encodes it, and stores the partial image on, for example, an optical disk 64. A compressed image is stored on the optical disk 64 by repeating the above-described processing for one screen.

[Effects of the Invention] According to the present invention, it is possible to automatically determine the most suitable number of times of difference processing for each partial region according to the partial data characteristics of an image, and to perform efficient data processing for a straight image. Since it is possible to achieve a high compression ratio, it is effective in improving processing efficiency and facilitating the storage of large image data.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing an overview of data compression processing, Fig. 2 is a detailed explanatory diagram of the present invention, Fig. 3 is an overall processing flow of the present invention,
Figures 4a and b are histograms of differential data, Figure 5 is an explanatory diagram showing one embodiment of the present invention, Figure 6 is an example of a system configuration suitable for implementing the present invention, and Figure 7 is the system configuration and processing. It is a schematic diagram showing the relationship between processes. 22... Image divided into partial areas, 52... Original partial image data, 521... One-time difference data, 52
2... Twice difference data, 62... Image processor. y 1 Figure 2 Ro Y 3 Country% 4 Port (A) Figure 5 Figure 2 Item 72 Continuation of page 1 ■Int, CI,' Identification code Internal reference number 0 Inventor Sato-Hiroshi Aiichi New 10-Year-Old 22-Ta

Claims (1)

[Claims] 1. It has means for storing an image, means for compressing the band of image data by differential processing, and means for encoding the compressed image data, and divides the image into a plurality of regions,
An image data compression method characterized by repeating differential processing multiple times for each area and providing means for stopping the repeating processing based on a certain standard.