JPS6227866A - Compressing and recording method for medical picture - Google Patents

Abstract

PURPOSE:To reduce the quantity of information to be recorded while maintaining the picture quality of an original picture as high as possible, by enlarging and interpolating the compressed picture having picture elements forming the original picture sampled at the proper intervals up to the size of the original picture to obtain the deviation between the picture elements of both pictures and recording the deviation picture after reversible compression. CONSTITUTION:An original picture is read out of a magnetic disk by a CPU 14 and a compressed picture for storage is obtained in accordance with a program. This compressed picture includes a basic compressed picture obtained through 1/4 compression, etc. and a deviation compressed picture. In addition, a part of a memory 16 and a memory 17 store the various types of information which are under processing. The CPU 14 records for storage the compressed picture for storage consisting of said two types of compressed pictures into an optical disk 18 according to another program. Then the CPU 14 reads the compressed pictures out of the disk and restores them by the programs.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to an image compression/recording method for compressing and recording digitized medical images.

[Background of the invention]

Medical images generally involve large amounts of data. For example, for X-ray fluoroscopic images, instead of the conventional film recording method, an accumulation recording method is being adopted in which X-ray fluoroscopic images are stored as data. The number of data is determined by the resolution. Generally, about 4 Mbytes are required per image. A chest X-ray fluoroscopic image may require about three screens, whereas an X-ray fluoroscopic image may require 10 or more screens. Therefore, in the case of three chest X-ray fluoroscopic images, the amount of data is 12M-ite. 10 sheets can reach 40MB.

Furthermore, the number of subjects is extremely large. Therefore, for each subject,
Storing and recording all the images results in an increase in the number of memories and storage space for the custodian. On the other hand, data recording devices include magnetic disks and optical disks. In particular, optical discs play a role as large-capacity recording devices.

Even when such optical discs are used, as described above, the number of optical discs increases and the storage space increases.

On the other hand, there is a problem with the accuracy of recorded information when restoring it after recording. It is desirable that the restoration after recording be a faithful reproduction of the original image. However, the question is how strictly to require fidelity. Although some of the recorded information is extremely important information that requires faithful reproduction of the information before recording in a completely strict sense, there is also information that does not require such fidelity. The former requires completely faithful records, while the latter requires less rigorous records.

Therefore, it is desirable to record information according to its importance.

[Purpose of the invention]

An object of the present invention is to provide a medical image compression/recording method that reduces the amount of information to be recorded while maintaining the image quality of the original image as much as possible when recording image data.

[Summary of the invention]

The present invention is based on the premise that pixels constituting the original image are extracted at appropriate intervals to obtain a type of compressed image, the compressed image is enlarged and interpolated to the size of the original image, and the enlarged and interpolated image and the original image are A deviation image is obtained by finding the deviation for each pixel. Above 1
We decided to record the compressed image of the seeds and the deviation image on a large-capacity recording device.

Incidentally, the deviation image can often be reversibly compressed by run-length coding, even if it is only recorded. In that case, instead of recording the deviation image as is,
It is possible to record the data after losslessly compressing it.

A type of image compression is determined depending on the precision of the reproduced image, and can be variously performed, such as 1/4 compression, 9/64 compression, etc.

[Embodiments of the invention]

FIG. 1 shows an embodiment of an information processing apparatus according to the present invention. The information processing device has a common path 20, and has a CPU 14, a magnetic disk 15, and a memory 16, 17 connected to this computer.
, has an optical disc 18.

The information processing device performs two processes: compression processing to obtain a compressed image for storage from the original image, and storage processing to record and store the compressed image for storage on the optical disc 18.

The optical disc 18 becomes a large capacity recording device. Further, the information processing device reads and restores the compressed image stored on the optical disc 18.

The above three processes are achieved by software.

This software is stored in memory 16. C.P.
By accessing the memory 16, U14 reads the nologogram and executes the program. At this time, it is assumed that the original image to be processed is stored on the magnetic disk 15. The original image on the magnetic disk 15 is transferred to the CPU 1
4 reads and obtains a compressed image for storage according to the program. There are two compressed images for storage: a basic compressed image obtained by compression such as 1/4, and a deviation compressed image.

Furthermore, a part of the memory 16, the memory 17, stores various information that is being processed.

The CPU 14 records and stores the compressed image for storage, which is composed of two compressed images obtained according to the nologram, on the optical disk 18 according to another nologram. This compressed image for storage has a smaller amount of information than the original image, and its accuracy satisfies predetermined content. The accuracy is determined by the degree of compression of the basic compressed image and the degree of deviation of the deviation compressed image.

The CPU 14 also reads and restores the compressed image stored on the optical disc 18 according to the nologogram.

Next, the processing contents will be specifically explained.

(1) Compression processing First, an overview of the compression process will be described.

Now, assume that the original image is H+ (L y). Each point corresponds to a pixel location. Each point is multi-valued information, and consists of information obtained by, for example, gradation of brightness. If there are 64 gradations, each point becomes 6-bit information.

Pixel information is extracted from the original image H+ (+y) at one-point intervals and the image is compressed to one-fourth. In other words, "(1/2 the vertical size) x (1/2 the horizontal size) = 4
According to 1 minute. This 1/4 compressed image becomes the aforementioned basic compressed image.

This basic compressed image becomes a compressed image for storage, and also serves as an image from which to obtain the above-mentioned deviation compressed image.

The procedure for obtaining a deviation compressed image is as follows.

First, a basic compressed image compressed to one-fourth is enlarged to the size of the original image. H,! This enlarged image! Let it be (x, y). Enlargement is performed by performing enlargement complementation on the basic compressed image. The interpolation is performed by, for example, an interpolation method. This interpolation method includes
There are linear interpolation, quadratic interpolation, curved interpolation, etc.

At this time, in the compression process, the frequency spectrum of the original image is
Only a fraction of the frequency of the original image is passed through, and when enlarging, the interpolation function restores the frequency to a range close to that of the original image. but,
In general, information in high-frequency vector regions is lost. In real space, this means that rapidly changing parts and fine details of the image are lost. The degree of compression and the contents of the enlargement process are determined.The degree of precision to be allowed depends on the nature of the original image and the degree of reproduction image required.

Next, the deviation ΔH(
x*y).

ΔH (x, y) = H1 (x, y) - Hl (x, y)...
(1) The deviation ΔH generally takes a large value at locations where there are sudden changes or small changes in the original image.

In areas where changes are gradual, the deviation ΔH is a small value of 0. In medical image diagnosis, rapid changes and small changes are often monitored. Areas that change slowly are not important areas.

Therefore, threshold processing is performed on portions where changes are gradual.

Let the threshold be a, (I)1ΔH(L y ) If l≦a, ΔH(x, y
)=0(II) lΔH(Ly)l)a Then ΔH
Let (x, y)=ΔH(x, y). That is, it is assumed that the deviation ΔH does not exist for the deviation ΔH below the threshold value,
For a deviation ΔH greater than the threshold value, the deviation ΔH itself is given as deviation information. A specific example of threshold processing on one scanning line is shown in FIG.

The threshold value a varies depending on the image and accuracy. For example, it is determined by a binary number such as a=1.2, 4, 8.16.

``If the threshold value a is increased, the compression ratio η increases, but the image quality of the restored image deteriorates.

Here, the compression ratio η is It is.

Thresholding also performs image compression. That is, since ΔH below the threshold value a becomes a rate of Kr0J, the ΔH at that time is lost. This reduces the amount of information and effectively achieves the same effect as image compression.

The image composed of the deviations thus obtained is called a first deviation compressed image. This first deviation compressed image can also be positioned as the aforementioned storage deviation compressed image.

However, the first deviation compressed image may be further compressed by a reversible compression method to obtain a second deviation compressed image, and this second deviation compressed image may be positioned as the above-mentioned storage deviation compressed image.

Here, lossless compression means that information is not lost due to compression.
Compression means that even after playback, the original pre-compression information can be restored. For example, there is compression using the run-length method.

Through the above process, a basic compressed image and a deviation compressed image are obtained.

(2) Storage Processing The CPU 14 writes the basic compressed image and the deviation compressed image onto the optical disc 18 according to the storage program. However, C.P.
The data may be transferred and stored using DMA without involving U14.

(3) Restoration Processing The CPU 14 reads out the basic compressed image and the deviation compressed image according to the program, and performs restoration processing. This restoration process is
This is a process in which the compression process shown in FIG. 3 is performed in reverse order. Details will be described later with reference to FIG.

Next, explanation will be given using a specific example. In Figure 3, the original image 1
A simple example is shown in which the pixel is composed of 8×8 pixels.

Information is extracted from this original image l in the vertical and horizontal directions pixel by pixel. The black circles in the original image l are information that should be extracted, and the white circles are information that is not extracted, that is, information that is not related to so-called compressed information.

As a result, the basic compressed image 2 consisting of the extracted information
get. This compressed image 2 is an image obtained by compressing the original image 1 to one fourth. That is, 8×8 pixels became 4×4 pixels.

Next, the basic compressed image 2 is quadrupled by enlargement interpolation to obtain an enlarged interpolated image 3. Expansion interpolation is performed by interpolation. In the enlarged interpolated image 3 in the figure, the black circle mark is 1/4
This is the pixel information itself of the basic compressed image during compression. rx
The J mark is interpolated information.

Next, the deviation is determined for each pixel between the original image 1 and the enlarged interpolated image 3. As a result, a deviation image 4 is obtained. In the deviation image 4 in the figure, the "+" mark indicates the deviation at the extracted pixel position =
An example of 0 is shown. That is, at the sampled pixel position, the information of the original image and the information of the compressed image 1 (information at the same position of the enlarged interpolated image 2) are the same, and the deviation thereof is naturally zero. Therefore, this 4×4 position information, which is naturally zero, is effectively compressed. On the other hand, in deviation image 4 in the figure,
The "mouth" mark indicates the deviation ΔH between the original information of the original image 1 and the information obtained by interpolation. This deviation ΔH is subjected to threshold processing.

Next, a first deviation image 5 is obtained from the deviation image 4 by threshold processing. The deviation ΔH of the "mouth" position in deviation image 4 is compared in size with the threshold a, and if |ΔH|≦a, the deviation ΔH at that position is set to zero, and if 1ΔHl)a, the deviation ΔH at that position is set to zero. The deviation ΔH is left as is as deviation information. In this way, a first deviation image 5 is obtained.

This first-order deviation image 5 is subjected to reversible compression using the run-length method to obtain a second-order deviation image 5.

The basic compressed image 2 and the second deviation compressed image 5 become compressed images for storage, and are stored on the optical disk 18. FIG. 4 shows an example of obtaining a 9/64 basic compressed image.

This is an example in which information is extracted at two intervals in the horizontal and vertical directions. The basic compressed image 7 is a 3×3 image. The compression ratio is approximately 1/7. The subsequent steps are the same as in FIG.

FIG. 5 shows a procedure for restoring a compressed image stored on the optical disc 18. This restoration procedure is also performed by the CPU 14! Execute by program. However, the timing of restoration does not matter. It may also be restored by other equipment.

In FIG. 5, the compressed image 2 and the reversibly compressed second deviation compressed image 11 are read from the optical disk 18. The read compressed image 2 is subjected to quadruple enlargement interpolation to obtain an enlarged interpolated image 3A. On the other hand, the read second deviation compressed image 11 is enlarged to obtain a first deviation compressed image 12 before the reversible compression image. The enlarged interpolated image 3A has the same content as the enlarged interpolated image 3 in FIG. 3, and the first deviation compressed image 12 has the same content as the first deviation compressed image 5 in FIG.

Next, the enlarged interpolated image 3A and the first deviation compressed image 5 are
Add pixel by pixel. The result of this addition becomes the restored image 13.

This restored image is not the original image itself, but is an image in which parts with large changes or parts with small changes are specially preserved. As a result, necessary information can be stored and recorded as a small amount of information, and when restoring, the amount of information can be correctly restored as is.

There are various methods of deformation.

(1) The compression ratio when obtaining a basic compressed image is originally
Optional.

(II) Pixel extraction to obtain a basic compressed image is as follows:
They do not necessarily have to be spaced at constant intervals vertically and horizontally. For example, 1
The rows are 1, 3, 5. Extract the odd numbers of ..., and the third line is 2, 4, 6, etc. You can also take the other option, such as extracting even numbers from ....

(III) In the restoration process, instead of adding the images 12, the images 12 may be subjected to low-pass filter processing and the results may be added.

In addition to medical images, the present invention can also be applied to other fields where images need to be preserved.

〔Effect of the invention〕

According to the present invention, it has become possible to record a necessary amount of information with a small amount of information. In comparison with the conventional example of simple reversible compression, a large amount of data can be compressed with only a slight deterioration in image quality, and the effective storage capacity of the data recording device increases accordingly.

In particular, in terms of medical images, single-plane image information can be easily stored and remains as diagnostic information, providing an economically superior means for hospitals and doctors.

[Brief explanation of the drawing]

Fig. 1 is an illustration of an embodiment of the present invention, Fig. 2 is an illustration of threshold processing, Fig. 3 is an illustration of compression processing, Fig. 4 is an illustration of other compression processing, and Fig. 5 is an illustration of image restoration. It is an explanatory diagram. 14...Norosessa (CPU), 20...Common lotus. Patent Applicant: Hitachi Medical Co., Ltd. Attorney: Tadashi Akimoto Figure 1 Figure 2 Figure 3 8x8 4x4
Takumi Yot no Shi-fILT Spear 2 Sunsun H-i Evasion 3 Otori 2i Figure 4

Claims (1)

[Claims] 1. Regularly extract pixel information from the original medical image H_1 (x, y) stored in the memory to obtain a basic compressed image, and expand and interpolate the basic compressed image to obtain the original medical image. An enlarged interpolated image H_2 (x, y) of the size is obtained, and the original medical image H_
1(x, y) and the enlarged interpolated image H_2(x, y) deviation Δ
H = H_1 (x, y) - H_2 (x, y) is calculated for each point (pixel), and the deviation ΔH at each point is compared with the threshold value a. If |ΔH|≦a, then The deviation information of a point is set to zero, and if |ΔH|>a, ΔH itself is given as the deviation information of that point to obtain a first deviation compressed image, and the basic compressed image and the first deviation compressed image, or A medical image compression method characterized in that, in place of the first-order deviation-compressed image, a second-order deviation-compressed image obtained by reversibly compressing the first-order deviation-compressed image is recorded on a recording device. Recording method. 2. The medical image compression/recording method according to claim 1, wherein the regular pixel information is extracted every other point in the length and width of the original medical image. 3. The medical image compression/recording method according to claim 1, wherein the regular extraction of pixel information is performed at every second point in the vertical and horizontal directions of the original medical image.