JPH08201464A - Method for detecting S / N value of television video signal - Google Patents

Abstract

(57) [Summary] [Purpose] To improve the detection accuracy of the S / N value in the video signal. [Structure] When detecting the S / N value of a television video signal, the input television video signal is divided into a plurality of small blocks (1) on the screen, and an image is obtained for each pixel of each divided small block. A difference (4) between the signal and the signal (2) averaged in the time axis direction in each of the small blocks, and /
Alternatively, the difference (5) between the signal (3) two-dimensionally averaged in each of the small blocks is obtained, and the distribution of these differences is compared with the noise distribution of the statistical distribution rule prepared in advance for S / N value detection. The significance (6) is obtained by comparison, and the S / N value based on the difference between the small blocks determined to have significance is adopted, and the comparison and examination are sequentially continued (7) in all screens. , Occurrence frequency (8) of those S / N values adopted in all screens is calculated, and effective S is calculated from the distribution of the occurrence frequency.
/ N value is detected (9).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of detecting noise in a television video signal, and more particularly to the measurement of its S / N value.

[0002]

2. Description of the Related Art Conventionally, in detecting the S / N value in a television image signal of this type, a signal having a known level is defined as a part of the image, and the difference between the signal and the known level is determined. The noise amount was calculated and the S / N value was calculated. In other words, it is not a true evaluation of the amount of noise contained in the video,
It was detecting the amount of noise that the system experienced in the video.
Alternatively, if the image is limited to a still image, the S / N value is calculated by regarding noise that changes with time. However, when limiting to this still image, it is necessary to detect the noise amount only in the still image portion obtained by a method such as motion detection or motion correction. However, it is generally difficult to detect an accurate moving part and a vector from an image, and it cannot be applied to all motions, so that the effect is limited for the complexity of the algorithm and the increase of the hardware scale. Came.

[0003]

That is, conventionally, in order to detect the S / N value in a television video signal of this type, it is necessary to set a signal of a known level in a part of the video,
In other words, there is a problem in that the amount of noise incurred in the system through which the video has passed is not a true evaluation of the amount of noise included in the video, and that it is only possible to detect the S / N value limited to the still image. there were. Therefore, an object of the present invention is to solve the above-mentioned problems and to use the complicated motion detection and motion correction by focusing on the difference between the video amplitude and the stochastic distribution of noise to separate and test the noise from the video. The present invention aims to provide a S / N value detection method for a television video signal, which is capable of estimating the S / N value in an actual video.

[0004]

To achieve this object, a television image signal S / N value detecting method of the present invention detects an input television image signal when detecting an S / N value of the television image signal. It is divided into a plurality of small blocks on the screen, and for each pixel of each divided small block, the difference between the video signal and the signal averaged in the time axis direction in each of the small blocks and / or each of the small blocks. The difference between the two-dimensional averaged signal and the noise distribution of the statistical distribution rule prepared for the S / N value detection is compared to obtain the difference, and the difference is calculated. The S / N value based on the difference of the small blocks determined to be adopted is adopted, and the comparative examination is sequentially continued in the entire screen, and the occurrence frequency of those S / N values adopted in the entire screen is determined. Seeking, then It is characterized in that detecting the effective S / N value from the frequency distribution.

[0005]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The principle of the present invention is that S of a television video signal is S.
In detecting the / N value, the entire screen is divided into a plurality of small blocks, the noise distribution of each small block is compared with a statistical distribution rule prepared in advance, and the significance is calculated. Discard the measured value, apply this to all screens, then find the occurrence frequency of S / N values adopted in all screens, and detect the effective S / N values from this occurrence frequency distribution. This is because it reduces false detections caused by video.

In general, when the image is a still image, noise is easily detected from the difference between the image signal of each pixel and the averaged signal in the time axis direction. However, in the case of a moving image, an error due to motion occurs in this difference. On the other hand, noise can be detected even by using the difference from the signal averaged two-dimensionally in the small block so as not to be affected by the movement of the image. However, an error due to a video signal component is mixed in this difference. As described above, each noise detection method has an error depending on the image content, and thus it has been difficult to detect noise in a conventional general image.

The method of the present invention will be described in detail below with reference to the block diagram shown in FIG. In the method of the present invention, first, the input video signal is divided into a plurality of small blocks (for example, a size of 8 pixels × 8 pixels) on the screen by using the small block formation processing 1, and each of the subdivided small blocks is divided. , A difference 4 from the signal averaged in the time-axis direction by the time-axis direction averaging process 2 and / or a signal averaged in the x, y two-dimensional space by the two-dimensional averaging process 3. Find the difference 5.

Next, the distribution of these differences is calculated in advance by S / N.
The significance is calculated by comparing with the noise distribution of the statistical distribution rule prepared for value detection, and the S / N value based on the difference between the small blocks determined to have significance is adopted. This process is shown in Figure 1.
Is executed in the difference significance degree determination process 6. When the noise distribution is, for example, normally distributed noise, a normal distribution test, that is, a normal chi-square test is used. Thus, it is possible to eliminate erroneous detection in a video portion in which a difference is caused by a motion and a video portion in which a difference is caused due to some two-dimensional strong correlation video component in each of the small blocks of the subdivided screen.

The verification of these small blocks is sequentially continued within the entire screen, and the S / N values of only the small blocks whose distribution has passed the verification as noise are accumulated by the S / N value accumulation processing 7. A frequency distribution (histogram) is further created by the frequency distribution creating process 8 for the accumulated S / N values over the entire screen. When the best S / N value is immediately determined from the created histogram, it may be output via the S / N value determination processing 9. Generally, however, the peak value, average value, and variance are calculated from the histogram. A threshold value is given to the histogram, and the best S / N value out of the best S / N value and the peak value and the average value above the threshold value is determined by the S / N value determination processing 9. Output through. As a result, the S / N value can be detected by removing the influence of the image as much as possible, and even if the detected S / N value is erroneously detected, a better S / N value can be output. Is fail-safe (of course, the opposite can be reverse-safe).

Further, the influence of the image can be further reduced by dispersing the threshold value of the occurrence frequency and referring to the past value. This is shown in FIG.
It is performed in the 9th route. Furthermore, by detecting the time change of the detection value, it is possible to further reduce the false detection due to the image and to estimate the noise amount in the actual image. This processing system path is shown by feedback processing from the output of the processing block 9 through the delay circuit 12 in FIG.

The above description is about the embodiment relating to claims 1 and 3 to 5, but in claim 2, the pixel of interest moves as the pixels on the screen are scanned, and the small block moves to this pixel of interest. It can be said that more detailed processing methods can be achieved by providing them one by one. Further, it is obvious that the present invention is not limited to the embodiments described herein, and various modifications and changes can be made within the scope of the invention.

[0012]

According to the S / N value detecting method of the present invention, the S / N value in a normal television moving image can be detected with high accuracy. As a result, the noise reduce amount can be optimized according to the change of the S / N value of the image, the noise reducer can be effectively operated, and fine image quality management can be performed.

[Brief description of drawings]

FIG. 1 is a processing block diagram showing a signal processing flow system for achieving a method of the present invention.

[Explanation of symbols]

 1 Small block processing 2 Time-axis direction averaging processing 3 Two-dimensional averaging processing 4,5 Subtractor 6 Difference significance judgment processing 7 S / N value accumulation processing 8 Histogram creation processing 9 S / N value judgment processing 10 Peaks Detection of value, average value, variance, etc. 11, 12 Delay processing

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Masahide Naemura 1-10-11 Kinuta, Setagaya-ku, Tokyo Inside Broadcasting Technology Laboratory, Japan Broadcasting Corporation (72) Inventor Jun Fukuda 1-1-10 Kinuta, Setagaya-ku, Tokyo No. Japan Broadcasting Corporation Broadcasting Technology Research Institute

Claims (5)

[Claims] 1. When detecting an S / N value of a television video signal, an input television video signal is divided into a plurality of small blocks on a screen, and a video signal is obtained for each pixel of each divided small block. And / or a difference with a signal averaged in the time axis direction in each of the small blocks and / or a difference with a signal averaged two-dimensionally in each of the small blocks, and the distribution of these differences is calculated in advance by S / N. The significance is calculated by comparing with the noise distribution of the statistical distribution rule prepared for value detection, and the S / N value based on the difference of the relevant small blocks determined to have significance is adopted, and all such comparative examinations are carried out. A television image characterized by continuing to sequentially within the screen, determining the frequency of occurrence of those S / N values adopted in all screens, and detecting an effective S / N value from the distribution of the frequency of occurrence. Signal S / N value Detection method. 2. The S of a television video signal according to claim 1, wherein said small block is moved from left to right and from top to bottom on the screen for each pixel and for each scanning line. / N value detection method. 3. The method according to claim 1 or 2, wherein
When detecting an effective S / N value from the distribution of the occurrence frequency of the obtained S / N value in the entire screen, it is necessary to set a determination threshold value that takes into account the variance of the occurrence frequency distribution. A characteristic S / N value detection method for a television video signal. 4. The method for detecting an S / N value of a television video signal according to claim 3, wherein a temporal change is taken into consideration in the determination threshold value. 5. The S / N of a television video signal according to any one of claims 1 to 4, wherein a temporal change is taken into consideration in detecting the effective S / N value.
Value detection method.