JPH09168149A - Digital image data compression method and digital image processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To calculate an optimum coefficient of quantization for every image based on the statistic featured value of the DCT(discrete cosine transformation) value that directly affects the compressibility when the digital image data are compressed by a compression system of a static image coding system JPEG. SOLUTION: This system includes the processing 1 which defines a relation table 5 between the statistic featured value of the DCT value and the compressibility and the processing 2 which calculates a coefficient 8 of quantization. In an operation mode, an optional image 4 is inputted and the table 5 is defined through the processing 1 between the statistic featured value of the DCT value and the compressibility. Then the table 5 is outputted. The compressibility 6 that is used for compression is inputted together with an original image 7 and the table 5. Then the processing 2 is carried out and the coefficient 8 of quantization is outputted. The coefficient 8 is inputted and the compression processing 3 is carried out. Then the compression data 9 are outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital image data compression method, and more particularly to a discrete cosine transform processing (D
CT processing) and digital image data compression processing including processing for quantizing DCT values, and a method for obtaining a quantization coefficient that can obtain a specified compression rate.

As an example thereof, a still image coding system JP
The present invention relates to a method for obtaining a quantized coefficient in a digital image data compression process based on a compression method of EG (Joint Photographic Experts Group).

[0003]

2. Description of the Related Art In digital image data compression processing according to the compression method of the still image encoding method JPEG,
It is said that it is not possible to control the image compression rate (reference: Nikkei Communication October 1992).
5th issue). Then, the same document describes that a parameter called a quantization coefficient is used to instruct the degree of compression.

On the other hand, Japanese Patent Laid-Open No. 06-326872
In the issue (hereinafter, known example (1)), the relationship between the compression rate and the quantized coefficient is linear when the image is compressed by using the expected maximum and minimum values of the quantized coefficient that can control the compression rate. The quantized coefficient that is defined by the equation and obtains the specified compression rate is obtained by this linear equation. However, since the relationship between the compression rate and the quantization coefficient cannot be defined by the linear equation in practice,
It is not possible to obtain the image with the desired compression ratio by performing the calculation twice. Therefore, the linear equation is redefined by using the obtained relationship between the quantized coefficient and the compression rate, and recalculation is performed (feedback processing is performed). The calculation is repeated until the quantized coefficient that creates the image of the desired compression ratio is obtained.

IEICE Transactions BI Vol. J75-B
-I pp.353-361, published in May 1992, entitled "Adaptive DCT Coding Method Based on Statistical Properties of Image" (hereinafter, known example (2)) describes CCITT (International Telegraph and Telephone Consultation). Based on the reference image defined by the committee), the quantized coefficient for each compression rate obtained in advance using a total of nine images, and the image feature amount calculated by the following formula using the image data (pixel value), The quantized coefficient for obtaining the specified compression rate of the specified image is calculated.

ACT (H) = ΣΣ | X (i, j + 2)-
X (i, j) | (i = 1,2, ..., 8; j = 1,2, ..., 6) ACT (V) = ΣΣ | X (i + 2, j) -X (i, j)
| (I = 1, 2, ..., 6; j = 1, 2, ..., 8) ACT (Block) = ACT (H) + ACT (V) where X (i, j) is a position in the block. (I, j)
Pixel value at ACT (H), ACT (V),
ACT (Block) indicates horizontal, vertical direction, and block activity, respectively.

That is, the relationship between the image feature amount calculated using the pixel value and the quantization coefficient is defined by a linear equation. Since this relational expression defines one linear expression for each compression rate,
When obtaining the quantized coefficients for all the required compression rates, the quantized coefficients for all the compression rates are obtained for a plurality of reference images.

Japanese Unexamined Patent Publication No. 03-85682 (hereinafter, known example (3)) discloses an image in which an optimum quantization coefficient is provided in advance for each information such as modality (imaging device or imaging method) and region regarding medical image data. The present invention relates to a data compression device.

[0009]

In the above-mentioned known example (1), the compression calculation process is repeated many times for each image to be compressed in order to obtain the quantization coefficient for obtaining the image of the desired compression ratio. It is necessary to carry out (feedback processing), and when a large number of images are compressed at one time, the processing time greatly increases.

Further, in the above-mentioned publicly known example (2), in order to create a table that defines a common quantization coefficient for any image without classifying the image, an original image data (pixel value) is used. The quantized coefficient is calculated using the feature quantity of. Further, the compression rate based on the JPEG compression method originally depends on the DCT value, but since the feature amount of the DCT value differs for each image, the compression rate of the same degree is obtained for all images. You can't really get an image.

Further, the above-mentioned publicly known example (3) relates to a device which is previously provided with an optimum quantization table for each information such as modality (imaging device or imaging method) and region regarding medical image data, and the compression ratio is controlled. Not a table to do. Also, classifying images is a general idea, not a special method.

An object of the present invention is to obtain a quantized coefficient for creating an image of a desired compression ratio, one image to be compressed 1
It is to provide a method that does not need to perform compression calculation processing many times (feedback processing) for a sheet. Another object of the present invention is to provide a method for obtaining a quantization coefficient for creating an image having a desired compression rate for each image that has been categorized to some extent, by using the statistical feature amount of the DCT value that directly affects the compression rate. .

[0013]

In order to solve the above problems, the following means are provided.

Means is provided for predefining a reference table showing the relationship between the statistical feature amount and the compression rate of the frequency domain data of the DCT processed data (DCT value),
Further, when the image is compressed, the statistical feature amount corresponding to the designated compression ratio is referred to from the table, and the statistical feature amount of the DCT value of the image to be compressed is converted into the referred statistical feature amount. A means for controlling the compression rate of the image is provided by calculating the quantized coefficient and controlling the statistical feature amount of the DCT value using the quantized coefficient.

Further, as a statistical feature amount, a variance value or a standard deviation value of the DCT value, a squared error from 0, a DCT
Means are provided for utilizing any of the zero frequency values.

Further, the process of defining a table showing the relationship between the statistical feature amount of the DCT value and the compression rate is performed at least for each group classified by the medical image capturing apparatus, or for each group classified by the capturing method, or A means for performing based on any one of the groups classified by the imaged site is provided.

Further, the processing for defining the table showing the relationship between the statistical feature amount of the DCT value and the compression rate is a means for utilizing only the DCT value of one point within a specific frequency region for each classified group, or A means for utilizing the DCT values at a plurality of points is provided.

Further, in the processing using only the DCT value of one point in the frequency domain, means for automatically determining the point in the frequency domain is provided.

The processing procedure in the present invention will be described with reference to FIG.

In step 1, a process of defining a relational table between the statistical feature amount of the DCT value and the compression rate is performed. Here, the relationship that the statistical feature amount of the DCT value and the compression rate are almost constant depending on the image classification method is used. Specifically, the arbitrary image 4 classified according to the information of the medical image capturing apparatus and the imaged region is input, and DC is input.
The relationship table 5 between the statistical feature amount of the T value and the compression rate is defined and output. Next, in step 2, quantized coefficient calculation processing is performed. Specifically, input the original image 7 to be compressed,
The statistical feature amount of the DCT value of the original image 7 is calculated. DCT
By using the relationship table 5 between the statistical feature amount of the value and the compression rate, the statistical feature amount corresponding to the input compression rate 6 is referred to. Quantization coefficient calculation processing for converting the statistical feature amount of the DCT value of the original image 7 into the referenced statistical feature amount is performed, and the quantized coefficient 8 is output. This quantized coefficient realizes a compression rate of 6. Next, in step 3, compression processing is performed. The quantized coefficient 8 and the original image 7 are input, compression processing is performed, and compressed data 9 is output. The compressed data 9 is data that realizes a compression rate of 6 with respect to the original image 7.

Steps 1 and 2 (thick line frames) are processing unique to the present invention, and step 3 is digital image data compression processing based on the compression method of the still image coding method JPEG.

After the relational table 5 of the statistical feature amount of the DCT value and the compression rate is defined by the process 1 for defining the relational table of the statistical feature amount of the DCT value and the compression rate, the quantum for each image to be compressed is defined. The necessary compression processing can be performed only by performing the conversion factor calculation processing 2. Furthermore, the point that the compression rate can be controlled by defining the relationship table between the statistical feature amount of the DCT value and the compression rate will be described with reference to FIG.

It was confirmed by experiments that the relationship between the statistical feature amount of the DCT value and the compression rate is almost constant. Therefore, the compression rate can be controlled by defining the relationship table between the statistical feature amount of the DCT value and the compression rate of the arbitrary image and controlling the statistical feature amount of the DCT value.

FIG. 2 shows the relationship between the statistical feature amount of the DCT value and the compression rate. FIG. 2A is an example of a frequency distribution of DCT values before compression processing (quantization processing) is performed for each of certain images A and B classified by the medical image capturing apparatus, the imaging method, and the information of the imaging region. Is represented. The image A has a high mountain distribution, and the image B has a gentle distribution. Thus, depending on the image, DC
The characteristics of the T value are usually different.

FIG. 2B shows D when compressed to the same compression ratio.
An example of the frequency distribution of CT values is shown, but the shape of the distribution is almost the same. Further, an example of the relationship of the statistical feature amount of the DCT value for each compression rate is shown in FIG. 2 (c) and FIG. 2 (d). FIG. 2C is an example of a variance value (or standard deviation value) or a squared error value from 0 as the statistical feature amount, and FIG. 2D is an example of a frequency value of 0. . Both show almost the same numerical changes. in this way,
It can be confirmed that the relationship between the statistical feature amount of the DCT value and the compression rate is almost constant depending on the image classification method.

The graph shown in FIG. 2 (c) or FIG. 2 (d) is a table which is a relational table of the statistical feature amount of the DCT value and the compression rate of the present invention. The numerical values to be actually tabulated can be obtained by using only the data of either image A or image B. Alternatively, a method of using the average value of the images A and B or the average value of a plurality of images may be used.

By using the relationship table between the statistical feature amount of the DCT value and the compression rate, the quantization coefficient when the image x is compressed at the compression rate m is calculated by the following equation.

Quantization coefficient (K) = statistical feature of DCT value of image x / compression ratio m defined in the relation table
Feature value of DCT value in / / value of quantization table in limited frequency band

[0029]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below. Example of compressed image management in medical image network management system 1. Example of image management on the same terminal Three examples of storing a required number of images in an image storage medium having a limited capacity will be sequentially described.

1. Example of Compressed Image Management in Medical Image Network Management System (1) Example System Outline In this example, medical images stored in a database are managed and operated using a network, and images are referred to by a plurality of departments. The present invention relates to a medical image network management system configured for the purpose. The system configuration is shown in FIG.

First, an imaging department 32, an image database (abbreviated as DB hereinafter) department 35, a hospital DB department 310, and an examination department 313 are connected to a network 31 in the hospital. The imaging department 32 includes a medical image capturing device 34 such as an X-ray, X-ray CT, MRI, and DR, and a data processing personal computer (hereinafter referred to as P
(Abbreviated as C) 33. In the photographing department 32, the image data photographed by the image photographing device 34 is stored in the image DB department.
Send to 35. In this embodiment, no compression is performed at this point. The image DB section 35 includes an image server 36, a data processing PC 37, and image DBs 38 and 39. Image D
The B department 35 stores and manages all image data generated in the photographing department 32.

In this embodiment, on the data processing PC 37 of the image DB section, an image having the set compression rate is created based on the compression rate set for each purpose of use of the image in the image DB of each section. , Is an example of transmission. First, the image DB 38 stores all images for the purpose of diagnosis or long-term storage. Therefore, image deterioration does not occur 1/3
Set the compression rate. The image DB 39 stores images with a high compression rate that are used as indexes for easy management and retrieval of the images stored in the image DB 38. Here, the 1/50 compression ratio is set.

The in-hospital DB department 310 is the in-hospital image server 311.
And an image DB 312. The image DB 312 stores and manages, for example, all image data of patients who are in the hospital, and other departments in the hospital, for example, the examination department 313.
The image for the purpose of image reference from is accumulated. Here, the 1/8 compression rate is set.

The examination department 313 is composed of a department image server 314, an image DB 316, and a management PC 315. Image D
B316 stores all the image data of the patient for one day to be examined, and is used by the doctor to confirm the image or to use it for informed consent to the patient. Here, the 1/20 compression rate is set.

The compression ratio control processing according to the present invention is performed in the image DB
It runs on the management PC 37 in department 35. That is, using an arbitrary image on the management PC 37 in the image DB department 35,
According to the present invention, a relation table between the statistical feature amount of the DCT value and the compression rate is defined. This process is not performed after it is performed once when the medical image network management system is introduced. Further, the quantization coefficient for controlling the compression rate is calculated based on the defined relationship table between the statistical feature amount of the DCT value and the compression rate and the compression rate set for each DB purpose, Perform compression. Then, the image is automatically transmitted to each DB.

The above is an outline of the system configuration and processing of this embodiment.

(2) Outline of DCT process and feature amount of image data in frequency domain Before describing details of the compression ratio control method in this embodiment, digital image data based on the compression method of the still image encoding method JPEG The DCT processing method and the quantization processing in the compression processing will be described with reference to FIG.

With respect to the arbitrary image 41 (size m pixels × n pixels), the image is first divided into blocks. Specifically, 8 ×
It is divided into blocks of 8 pixels (42 in the figure), and a two-dimensional discrete cosine transform process (DCT process) is performed on the divided plurality of pixel data blocks 43 by the following equation (Equation 1).

[0039]

[Equation 1]

However, x, y: position of pixel in block u, v: position of DCT coefficient C _u , C _v = 1 / √2; u, v = 0 = 1; other L _x = 128; P _xy Bit precision = 8 bits = 2048; _Pxy bit precision = 12 bits Thus, in one 8 × 8 block 44, frequency data corresponding to the pixel data block 43 is calculated. From left to right of the block, horizontal low frequency components to high frequency components are described, and from the top to the bottom of the block, numerical values including vertical low frequency components to high frequency components are described. The above is the DCT processing method.

Further, the quantization processing will be described.

To quantize the DCT value, a quantization table is used. This quantization table is composed of numerical values set for each frequency band divided into 8 × 8 blocks. An example of the quantization table is shown in 45.

The quantization processing is a numerical value obtained by multiplying the quantization table value by the quantization coefficient (quantization parameter: see the following formula).
Is to divide the numerical value at the same position in the frequency data block.

Quantization parameter (Q _xy ) = quantization table value (T _xy ) × quantization coefficient (K) where x, y: position in the table (= 0; 7) Multiply each numerical value in the quantization table 45 by the quantization coefficient value, and add 5
It becomes possible to change the compression rate of the image by controlling the numerical value of the quantization parameter by performing the process of dividing by 0.

Next, the statistical feature amount of the DCT value in the present invention will be described.

The statistical feature amount of the DCT value is the DCT value.
Frequency data block 44 in FIG. 4 in the explanation of the processing outline
It means the feature quantity of the numerical value inside. Therefore, the process of calculating the statistical feature amount of the DCT value is for calculating the statistical feature amount regarding the data for each same frequency, and is represented by the following calculation formula.

DCT value feature: P = f ((S _uv ) _ij ) where u, v: position of DCT coefficient i, j: position of block f: function for calculating feature Here, the function f is It means the variance value (or standard deviation), the squared error from 0, or the frequency of 0.

Further, the process of calculating the characteristic amount of a specific frequency band will be described. This means a process of limiting the data in the frequency data block 46 in FIG. For example, the feature amount of the DCT value within the band is calculated by limiting to the band indicated by the shaded portion 47.

The method for limiting the area can be set arbitrarily and may be, for example, the shaded portion 48 and the shaded portion 49. fundamentally,
The shaded area 47 is a band including many low frequency components, and the shaded area 48 is
Indicates a band including a large amount of medium frequency components, and a shaded portion 49 indicates a band including a large amount of high frequency components.

As an example of further limiting the area, for example,
It is also possible to use the DCT value of one point 440. Alternatively, it is possible to use the average value of a plurality of neighboring points. At this time, as described above with reference to FIG. 2, the present invention is used by utilizing the characteristic that the statistical feature amount of the DCT value is almost the same for each compression rate depending on the image classification method. The compression rate control process of is realized.

In this embodiment, as a method of classifying images, a chest X-ray image (classification according to photographing device and photographing region), a head X-CT image (classification according to photographing device and photographing region), and a stomach X-ray image are collected. 4 (points 410, 411, and points in FIG. 4) for each of the images (classification according to the imaging device, the imaging method, and the imaging site) and the double contrast image of the stomach X-ray image (classification according to the imaging device, the imaging method, and the imaging site) 412, the method of calculating the feature amount in the frequency domain of the image data using the point 413 is adopted. That is, the characteristic amount in the frequency domain of the image data is calculated by the following formula.

DCT value feature quantity: P = f ((S _uv ) _ij ) In case of point 410: u = 0, v = 2 In case of point 411: u = 3, v = 3 In case of point 412: u = 0 , V = 3 In the case of point 413: u = 0, v = 1 As described above, there is a method of designating points in the frequency domain by the medical image classification method, and in addition, points in the frequency domain are automatically determined. It has the function of determining. A method of automatically determining points in the frequency domain will be described later.

(3) Details of compression rate control processing In this embodiment, the compression rate control processing is performed on the management PC 37 in the image DB section 35 in FIG. Details will be described below with reference to the drawings.

First, the processing procedure of the relational expression definition processing of information regarding compression will be described with reference to FIG. In the "DCT process" of step 51, the arbitrary image 501 classified by the information of the image capturing device and the imaged region of the medical image is input, and the discrete cosine transform process (DCT process) is performed. Next, in the "DCT value statistical feature amount calculation process" in step 52, D
As the statistical feature amount of CT, the variance value (or standard deviation) of the DCT value, the square error from 0, or
Calculate the frequency of 0. In the "quantization coefficient variation process" of step 53. A process of changing the quantization coefficient is performed. Here, for example, starting the setting of the quantization coefficient from 1,
Processing for increasing by 5, such as 5, 10, 15 ... Alternatively, the quantized coefficient that realizes the desired compression rate is obtained for each compression rate. “Quantization process” in step 54
Then, the quantization table 502 is input, and the DCT value is quantized using the quantization coefficient set in step 53.
In the "statistical feature amount calculation process of DCT value" in step 55, the feature amount regarding the DCT value quantized in step 53 is calculated. Similar to step 52, the variance value (or standard deviation) of the DCT value or the square error from 0,
Alternatively, the frequency of 0 is calculated. In the "Huffman coding process" of step 56, the coding table 503 is input,
Encode the quantized DCT value.

Note that steps 52, 53 and 55 (bold line frames) are the processes according to the present invention, and steps 51 and 5
Reference numerals 4 and 56 denote digital image data compression processing based on the compression method of the still image encoding method JPEG.

Steps 53, 54, 55 and 56 are repeatedly processed, and steps 53, 55 and 5 are repeated each time.
The compression ratio and the DCT value statistical feature amount obtained in 6 are output, and the relation table 50 between the DCT value statistical feature amount and the compression ratio is output.
Define 4

Next, FIG. 6 shows a specific example of the relationship table 504 between the statistical feature amount of the DCT value and the compression rate. Compression rate, DC
The T-value feature amount is shown. First, the compression rate 1/1, that is, the information when not compressed is described. At this time, the variance value 50 is described as the DCT value feature amount. Similarly, at the compression rate of 1/2, the variance value 40 is described as the DCT value feature amount. The same applies to the compression rates below. The table 61
Is set for at least one point in the frequency domain for at least the imaging device, the imaging method, or the imaging site.

Next, an example of quantized coefficient calculation processing will be described with reference to FIG.

In step 71, "DCT processing", the original image 701 to be compressed is input and DCT processing is performed. In the “statistical feature amount calculation process of DCT value” in step 72, the DCT value is calculated as the statistical feature amount of the DCT value of the original image 701.
The variance value (or standard deviation) of the values, the squared error from 0, or the frequency of 0 is calculated. Next, in the “quantization coefficient calculation process” of step 73, the statistical feature amount of the DCT value calculated in step 72, the compression rate 702,
The quantization coefficient calculation process is performed using the relationship table 703 between the statistical feature amount of the DCT value and the compression rate, and the quantization coefficient 704
Is output.

The steps 72 and 73 (thick line frame) are
The process according to the present invention is step 71, which is a digital image data compression process based on the compression method of the still image encoding method JPEG.

Further, the details of the quantization coefficient calculation processing in step 73 will be described. The data to be used are the statistical feature amount of the DCT value calculated in step 72 of FIG. 7, the compression ratio 702, the relational table 703 of the statistical feature amount of the DCT value and the compression ratio, and the quantization table. As an example, a case will be described in which a chest X-ray image is compressed and accumulated in the image DB 38 in FIG. 3 in which the compression ratio ⅓ is designated. The calculation formula for obtaining the quantization coefficient in that case is described as follows.

Quantization coefficient coefficient = statistical feature amount of DCT value of original image / statistical feature amount of DCT value at specified compression rate defined in relation table / value of quantization table in specified frequency band × In the quantization process in 50 JPEG, when the quantization parameter is obtained, since division is performed by 50, 50 is finally multiplied. Then, the statistical feature amount of the DCT value of the original image = 50 The statistical feature amount of the DCT value at the designated compression ratio defined in the relation table = 35 The value of the quantization table in the designated frequency band = 14 , Quantization coefficient = 50/35/14 x 50
≈5.1.

Similarly, a case will be described in which the same image is compressed and transmitted to the image DB 312 in FIG. 3 in which the compression rate ⅛ is designated.

Statistical feature amount of DCT value of original image = 50 Statistical feature amount of DCT value at designated compression rate defined in relation table = 10 Quantization table value in designated frequency band = 14 Therefore, the quantization coefficient = 50/10/14 × 50
≈17. Becomes

Using these quantized coefficients, digital image data compression processing according to the compression method of the still image coding method JPEG is performed, and the compressed image data is transmitted to each image DB.

Here, the function of automatically determining one point in the frequency domain will be described.

The determination procedure is as follows.

A table 61 shown in FIG. 6 is created for each of a plurality of points in the frequency domain.

With respect to a plurality of medical images having the same classification information, the quantized coefficient calculation processing is performed for each of the plurality of points used in the above by the method described above.

For each point, based on the statistics of the target compression ratio and the actually controlled compression ratio,
Automatically determine one point.

As a statistic, there is a method of using a squared error amount between a target compression rate and an actually controlled compression rate, or a method of utilizing a difference.

When using the respective numerical values, the following calculation method can be arbitrarily selected.

(A) The point that the average value of the statistics for the set compression rate is the minimum.

(B) The point that the maximum value is the minimum among the statistic amounts corresponding to the set compression rates.

The number of designated compression ratios means the number of compression ratios to be controlled, and the compression ratio of a certain image is 1/8, 1
If it is controlled to / 20 and 1/50, the designated compression rate number is 3.

The above is the embodiment of the compressed image management in the medical image network management system.

2. Example of image management on the same terminal FIG. 8 shows a device configuration to which the present invention is applied. here,
The image capturing device 81, the processing device 82, the image display device 83, and the image DB 84 are included.

The photographed image is sent from the image photographing device 81 to the processing device 82. Here, an image is displayed on the image display device 83, and confirmation work is performed. After that, an arbitrary compression rate is designated and stored in the image DB 84. An example of compressing an image at an arbitrary compression rate is the same as in the first embodiment.

Here, when the image is stored in the image DB 81, the quantization coefficient is also stored together. Therefore, when the compressed image is displayed again, this quantization coefficient is used.
Image display device that performs decompression processing based on PEG processing
Redisplay the image on 83.

3. Example of storing required number of images in image storage medium having limited capacity FIG. 9 shows an apparatus configuration to which the present invention is applied. Digital camera 91, magnetic card reader 92, data recording magnetic card
It comprises 93, a processing device 94, and an image display device 95.

The digital image taken by the digital camera 91 is sent to the processing device 94. Here, the captured image is 5
If the size is 12 pixels × 512 pixels and each pixel has a capacity of 8 bits, the size of one image is about 2 Mbytes. When the number of captured images is four and the capacity of the data recording magnetic card 93 is 1 Mbytes, it is necessary to compress the images at a 1/8 compression rate in order to store these four images. There is. Here, on the processing device 94, the quantized coefficient that achieves the ⅛ compression rate is calculated and the compression processing is performed by the compression rate control method according to the present invention.

Similar to the second embodiment, when the image is stored in the data recording magnetic card 93, the quantization coefficient is also stored together. Therefore, when the compressed image is displayed again, the quantization is performed. Decompression processing is performed based on the JPEG processing using the coefficient, and the image is displayed again on the image display device 95.

[0083]

According to the present invention, the following effects can be obtained in obtaining a quantization factor for creating an image having a desired compression ratio.

(1) When a quantized coefficient that realizes a desired compression rate is obtained by providing a reference table that directly uses the statistical feature amount of the DCT value for controlling the compression rate,
There is no need to perform compression processing (feedback processing) many times for each image.

(2) Since the relationship between the statistical feature amount of the DCT value and the compression rate is almost constant depending on the image classification method, the statistical feature amount of the DCT value and the compression rate are used. It suffices to define only one table showing the relationship of.

(3) Still image coding method Only data obtained by the basic method of JPEG is used (no special calculation is required).

(4) Statistical feature amount and compression ratio of DCT value for each group classified by the medical image capturing apparatus, each group classified by the imaging method, or each group classified by the imaging region By creating a table showing the relationship of, the compression rate control accuracy is further improved.

(5) When calculating various statistical feature values of the DCT value, the DCT value at one point in the frequency domain is used, or the DCT values at a plurality of points are used to further improve the accuracy of the compression rate control. Is improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing an operation of the present invention.

FIG. 2 is a diagram showing a relationship between a compression rate and an image feature amount.

FIG. 3 is a diagram showing a configuration of a medical image network management system to which the present invention is applied.

FIG. 4 is a diagram showing DCT processing and frequency component region limiting operation.

FIG. 5 is a diagram illustrating a process of defining a relationship table between a statistical feature amount of a DCT value and a compression rate.

FIG. 6 is a diagram showing a relationship table between statistical feature amounts of DCT values and compression rates.

FIG. 7 is a diagram illustrating a quantized coefficient calculation process.

FIG. 8 is a configuration diagram of one terminal to which the present invention is applied.

FIG. 9 is a block diagram of a digital image capturing apparatus to which the present invention is applied.

[Explanation of symbols]

31 ... Network, 33 ... PC for data processing, 34 ... Medical image capturing device, 36 image server, 3
8, 39 ... Image database.

Claims (10)

[Claims] 1. A digital image data compression method including a discrete cosine transform process (DCT process) and a process for quantizing a DCT value. Statistical characteristics of frequency domain data of data after DCT process (DCT value). A table showing the relationship between the amount and the compression rate is defined in advance, and when the image is compressed, the statistical feature amount corresponding to the specified compression rate is referred to from the table, and the D of the image to be compressed is referred to.
A process for calculating a quantized coefficient that converts a statistical feature amount of a CT value into a referenced statistical feature amount is provided, and the quantized coefficient is used during the quantization process to control the statistical feature amount of the DCT value. A digital image data compression method, characterized in that the compression rate of the image is controlled by doing so. 2. The statistical feature quantity according to claim 1 is DC.
A method for compressing digital image data, characterized by using a dispersion value or a quasi-deviation value of data (DCT value) after T processing. 3. The statistical feature quantity according to claim 1 is DC.
A method of compressing digital image data, characterized by using a squared error value from 0 of data (DCT value) after T processing. 4. The statistical feature quantity according to claim 1 is DC.
A digital image data compression control method, wherein a frequency value of 0 of data (DCT value) after T processing is used. 5. The process of defining a table showing the relationship between various statistical feature amounts of DCT values and compression ratios according to claim 1, is classified at least for each group classified by a medical image capturing apparatus or by a capturing method. For each group
Alternatively, the method for compressing digital image data is characterized in that it is performed based on any one of the groups classified by the imaging region. 6. The process of defining a table showing the relationship between various statistical feature amounts of DCT values and compression rates according to claim 5,
D at one point in a specific frequency region for each classified group
Function to define using only CT value, or DCT of multiple points in specific frequency domain for each classified group
A digital image data compression method having a function of defining using a value. 7. A DC at a point in the frequency domain according to claim 6.
The process of defining the T value has a function of automatically determining one point in the frequency domain based on the statistical amount of the target compression ratio and the controlled compression ratio. Compression method. 8. A method of compressing digital image data according to claim 7, wherein a squared error amount with respect to a target compression rate is used as the statistic amount. 9. A digital image data compression method, wherein a difference amount from a target compression ratio is used as the statistic amount according to claim 7. 10. An image processing system comprising at least two image processing devices each comprising a data processing device, an image display device and an image database, each of which is connected via a transmission line, at least 1. On one or more image processing devices, the relationship between the statistical feature amount and the compression rate of the data (DCT value) after DCT processing for each same frequency region is substantially constant depending on the image classification method. Using the relationship, DC
A table indicating the relationship between the statistical feature amount of the T value and the compression rate is defined in advance, and when the image is compressed, the statistical feature amount corresponding to the designated compression rate is referred to from the table to perform the compression. Image D
A process for calculating a quantized coefficient that converts a statistical feature amount of a CT value into a referred statistical feature amount is provided, and the quantized coefficient is used during the quantization process to control the statistical feature amount of the DCT value. By doing so, a process for controlling the image compression ratio is provided, and further, a process for expanding the compressed image is provided for all the image processing devices, and the digital image processing system is characterized.