JP2002232876A - Method for efficiently reducing picture data obtained in remote picture monitoring system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for performing wavelet transformation on picture data obtained in a remote picture monitoring system arranged in a nuclear facility and efficiently reducing data. SOLUTION: A differential processing is performed by subtracting picture data being the standard of comparison from picture data obtained at a prescribed interval, and wavelet transformation is performed on the result. Respective coefficients obtained by a wavelet transformation are quantized and are encoded. Thus, reduced picture data is obtained. Reduced picture data is decoded and is inversely quantized, and the wavelet transformation coefficients are restored. Inverse wavelet transformation is performed on the restored respective transformation coefficients and differential picture data is restored. Differential picture data is added to picture data being the reference of comparison and therefore restored picture data is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a remote image monitoring system for performing a wavelet transform on image data obtained by a remote image monitoring system installed in a nuclear facility or the like and performing intensive monitoring at a remote position. The present invention relates to a method for efficiently reducing image data.

[0002] The wavelet transform was originally devised by a French oil exploration engineer, and has been developed and used in various fields. That is, the wavelet transform is a time scale transform for a signal, and is a transform for mapping a signal to a time scale region by a wavelet function including two parameters of time and scale.

There are various types of wavelet transform, and a wavelet proposed by a person named Daubechies is called "Daubechies wavelet".
In this case, N is used as a parameter, and there are five types of N, 2, 4, 6, 8, and 10. In the present invention, Daubechies wavelet when N = 4 is used. The outer shape of the wavelet changes depending on the value of N. As N increases, the waveform becomes smoother and the existing section becomes longer.

[0004]

2. Description of the Related Art A remote image monitoring system, which is one of the conventional monitoring technologies, is for remotely and intensively monitoring image data obtained by a monitoring system installed in a nuclear facility or the like. The compressed image data is large in size, and if it is used as it is, the transmission cost and the recording cost are too high. Therefore, data compression is an indispensable technique.

[0005]

In this remote image monitoring system, JPEG (Joint Photographic Experts Group) is used.
Is standardly adopted as an image data compression method. JPEG, which is a conventional data compression method for still images, performs processing in units of blocks (8 × 8 pixels). Therefore, if the compression ratio is increased, block distortion due to inconsistency of surrounding block boundaries will appear on the image. There is a disadvantage that the image quality deteriorates. Therefore, when image quality is taken into consideration as a monitoring system, it is not realistic to compress monitoring image data at a very high compression ratio using the JPEG method.

[0006]

In order to solve this problem, the present invention has devised a new image data reduction and restoration method using a wavelet transform. The features of the method of the present invention are as follows.

1) A black and white still image (256 pixels per pixel)
Gradation). 2) Since the processing is not performed in units of blocks, block distortion that causes image quality deterioration does not occur in principle.

[0008] 3) The surveillance image data is acquired from a fixed camera, and it is assumed that the frame of the image is also fixed (that is, the viewing direction and angle of the camera are always constant, and enlargement / reduction and rotation of the image are also performed) None). It is also assumed that the brightness of the monitoring area during the monitoring period is substantially constant (in a place where the influence of sunlight or the like can be ignored indoors).

4) The smaller the difference between the reference image and the image to be processed, the smaller the size of the reduced image data. That is, the method of the present invention is performed as follows.

1) In order to reduce the image data and restore the image according to the present invention, two image data are always required.
One is image data serving as a reference for comparison, and the other is image data to be processed.

2) Image data is reduced as follows. First, subtraction processing is performed by subtracting image data serving as a standard for comparison from image data acquired at certain intervals, and a wavelet transform is performed on the result. Quantization is performed on each coefficient obtained by the wavelet transform, and the resulting image is coded to obtain reduced image data.

3) Next, image data is restored as follows. First, the reduced image data is decoded (decoded), and then dequantized to restore the wavelet transform coefficients. Wavelet inverse transform is performed on each of the restored transform coefficients to obtain restored difference image data.
The restored image data is obtained by adding the restored difference image data to reference image data serving as a reference for comparison.

In the present invention, a monitoring image is obtained at a certain interval, the obtained image is compared with a reference image to obtain a difference image, and the difference image data is subjected to processing such as wavelet transform and quantization to obtain data. Since the restoration processing is performed after the reduction, the monitoring image data in the facility (the fuel storage facility or the like) where the scene change is small is greatly reduced according to the present invention, and the inspection work is made more efficient.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS When a certain image is subjected to wavelet transform, wavelet transform coefficients corresponding to the number of pixels constituting the image can be obtained. All the obtained transform coefficients are not reduced as they are, and therefore have a large data amount.

Therefore, first, the transform coefficients are approximated by some representative values by quantization. There are various methods for this quantization. As a simple method, a representative value is set at equal intervals, and quantization is performed using the representative value.
For example, when the distribution of the transform coefficients is distributed between 0 and 100 and quantized uniformly at a quantization interval of 10, the distribution of 0 to 100 is changed to 0 to 10 as shown in Table 1 below.
It can be expressed only by the value between. Since this is a kind of data decimation, in order to reproduce the original data (approximate value) from the representative value after quantization, from Table 1, if the representative value is 1, 1-
5 in the middle of 10, 8 if the representative value is 9 in the middle of 81-90
An inverse quantization process is performed by setting a rule such as 5 or the like.

[0016]

[Table 1] From the distribution of the quantized representative values, encoding is performed by a method such as assigning short codes in order from the representative value appearing with the highest frequency, and the final reduction is performed.

FIGS. 1 and 2 are block diagrams of the image data reduction processing and the image restoration processing of the present invention, respectively. As shown in FIG.
Perform as follows. First, a difference image is obtained by subtracting a reference image serving as a standard for comparison from a processing reference image acquired at a certain interval, and a wavelet transform is performed on the difference image data. Quantization is performed on each coefficient obtained by the wavelet transform, and the resulting data is encoded to obtain reduced data.

The image restoration is performed as follows, as shown in FIG. First, the obtained reduced data is decoded and then dequantized to restore the wavelet transform coefficients. By performing a wavelet inverse transform on each of the restored transform coefficients, restoring a difference image, and adding the difference image to a reference image serving as a reference for comparison,
Obtain the restored image.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In the present invention, a wavelet transform is first performed on a difference image as shown in FIG. 1 for data reduction processing. In order to obtain this difference image (an image indicating a change between two images), two image data are always required, one of which is a reference image to be compared, and the other is an image to be processed. (Processing target image). In the present invention, a test is performed using FIG. 3 as a reference image and FIG. 4 as a processing target image. The difference between the reference image and the processing target image is that the processing target image shows two persons working on both sides of the desk at the innermost part of the image. In order to evaluate the performance of the present invention, the same processing target image (FIG. 4) is compressed at several compression ratios by the conventional JPEG compression method, and the result and the execution result of the present invention are summarized. Table 2. In both methods, data compression is performed in both quantization and encoding steps.

[0020]

Table 2 Comparison between JPEG compression method and the method of the present invention after compression processing File size after compression Compression ratio PSNR File name Byte (dB) JPG5 33,624 9.17 40 0.02 JPG3 25,697 12.00 37.74 JPG1 16,594 18.58 33.71 JPG0 15,068 20.46 32.91 WV4 12,418 24.83 39.73 In Table 2, "after compression" The “file size” and the “compression ratio” are data indicating performance regarding data compression. The PSNR refers to a peak signal-to-noise ratio, and indicates the degree to which an image reproduced from data after compression has deteriorated due to compression processing with respect to an image before compression. The smaller the value is, the more the image is degraded.

In Table 2, those whose file names are JPG are compressed by the JPEG compression method, and the data is reduced by WV according to the present invention. In the JPEG compression method,
Although 0, 1, 3, and 5 are used as the compression parameters, the compression ratio increases as the numerical value of the parameter decreases. The wavelet used in the embodiment is Daub
echies wavelet (N = 4).

From this table, PSN which is one of the indexes of the image quality is shown.
R (peak signal-to-noise ratio) is 39.73 for WV4, which is considered to be almost comparable to JP5 or JP5.
The compression ratio is 24.83 or more times that of G3.

In order to see the difference in image quality between the image reproduced from the reduced data obtained by the method of the present invention and the image reproduced from the compressed data obtained by the JPEG compression method, an approximate data compression is shown in Table 2. JPEG compression method (JP
G0: compression ratio of 20.46) and reproduced images of the method of the present invention (WV4: compression ratio of 24.83) are shown in FIGS. 5 and 6, respectively.

FIG. 5 is an image compressed by the JPEG compression method with respect to FIG. 4, and FIG. 6 is a diagram restored from the result of data reduction by the method of the present invention with respect to FIG. Naturally, both figures are the same as those in FIG. 4, but the processing for reducing data (such as quantization) is performed in both the JPEG compression method and the method of the present invention. Has deteriorated image quality. However, PS in Table 2
The NR (peak signal to noise ratio) is determined by the method of the present invention (WV4: P
SNR39.73) is the JPEG compression method (JPG0: PS)
NR 32.91), indicating that the method of the present invention has less degree of image quality deterioration accompanying the data compression processing.

FIGS. 5 and 6 show the degree of the deterioration in detail, since the difference is not clearly seen, and FIG. 7 (JPG0) which is a partial enlarged view of FIG. 5 and a partial enlarged view of FIG. 8 (WV4), the block-like pattern shown in FIG. 7 appears, and the degree of image quality deterioration is J
It is clear that the method of the present invention is less than the PEG compression method.

[0026]

According to the present invention, it is expected that surveillance image data in a facility (a fuel storage facility or the like) with a small scene change will be greatly reduced, and the monitoring work will be more efficient.
Also, by using the monitoring method of the present invention, it was discovered early that nuclear material for peaceful use was used for nuclear weapons and the like,
Measures can be taken to prevent the production of nuclear weapons. Still further, the present invention facilitates on-site activities performed by inspectors to verify that safeguards-targeted nuclear material, equipment, or facilities are used in accordance with safeguards agreements.

[Brief description of the drawings]

FIG. 1 is a diagram showing a block diagram of an image data reduction process according to the present invention.

FIG. 2 is a diagram showing a block diagram of a reduced data restoration process according to the present invention.

FIG. 3 is a diagram showing a reference image.

FIG. 4 is a diagram showing a processing target image.

FIG. 5 is a diagram showing a reproduced image by the JPEG compression method.

FIG. 6 is a diagram showing a reproduced image according to the method of the present invention.

FIG. 7 is a diagram showing a partially enlarged view of FIG. 5;

FIG. 8 is a diagram showing a partially enlarged view of FIG. 6;

Claims (1)

[Claims] 1. A method for efficiently reducing image data obtained by a remote image monitoring system, comprising the following steps: 1) When performing image data reduction processing, a monitoring area is photographed at a certain interval by a camera and a monitoring image 2) Compare the acquired processing target image with a reference image serving as a reference for comparison, take out an image having a small difference as a difference image, perform wavelet transform on the difference image data, and obtain a transform coefficient. 3) Quantize the obtained wavelet transform coefficients; 4) Create reduced data by encoding the quantized data; 5) When performing the next image restoration process, Decoding, inverse quantization, inverse wavelet transform to obtain a restored difference image, and 6) adding the obtained restored difference image to a reference image to obtain a restored image. Law.