JPH11112978A - Signal processing method and signal reproducing device - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To reduce mosquito noise. SOLUTION: At the time of decoding band compression-encoded data, a plurality of adjacent pixels of a video signal are treated as one block,
A difference between pixel levels adjacent in the block in the horizontal or vertical direction is detected, a pixel whose difference in pixel level is larger than a predetermined value T1 is set as an edge pixel, a pixel whose difference in pixel level is smaller than a predetermined value T2 is set as a flat pixel, When edge pixels and flat pixels coexist in the same block, the level of the flat pixels is corrected so that the level difference between the flat pixels is reduced, and then output is performed. The level correction is performed only when the flat pixels are continuous in at least three pixels in the horizontal or vertical direction. This makes it possible to effectively reduce mosquito noise generated at an edge portion of an image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of processing a video signal encoded by band compression involving block processing such as (discrete cosine transform), and to a signal processing method for decoding and a signal reproducing apparatus.

[0002]

2. Description of the Related Art In recent years, transform coding such as DCT has become the mainstream in band compression. These use the information of the original image as 1 /
This is a method of highly efficient compression to about 10 to 1/40.

However, as the compression ratio increases, so-called mosquito noise, in which noise in a block located at an edge portion of an image becomes conspicuous, has occurred. Conventionally, a technique for reducing the mosquito noise is disclosed in
There is one disclosed in Japanese Patent No. 9126. this is,
An edge portion is detected based on a difference between pixel levels, and a filter process is performed on a fixed number of pixels on the left and right sides.

[0004]

However, in this conventional method, all the pixels around the edge are subjected to the filtering, so that even the pixel indicating the original signal around the edge is subject to the filtering. There is a problem that the original signal that is originally required is attenuated.

[0005]

SUMMARY OF THE INVENTION The present invention relates to a method of processing a video signal which is band-compressed and encoded using a plurality of pixels adjacent to the video signal as one block. A difference between pixel levels adjacent in the direction is detected, a pixel having a pixel level difference larger than a predetermined value T1 is defined as an edge pixel, a pixel having a pixel level difference smaller than a predetermined value T2 is defined as a flat pixel, and a pixel level difference is determined as a predetermined pixel. When the pixel larger than the value T2 and smaller than T1 is set as an intermediate pixel, and an edge pixel and a flat pixel are mixed in the same block, the level difference between the flat pixels is reduced without changing the edge pixel and the intermediate pixel. It is configured to output after correcting the level of the flat pixel. Further, the correction of the flat pixel level is such that the flat pixel is at least horizontally or vertically. The correction value of the flat pixel level is obtained only by a filter weighting the average value level or the center of a plurality of continuous flat pixels, and a part of the flat pixels. Alternatively, the levels of all pixels are corrected.

Accordingly, even when mosquito noise occurs at an edge portion of an image, it is possible to selectively filter only flat pixels without subjecting a pixel indicating an original signal originally around the edge to filtering. As a result, only the mosquito noise can be effectively reduced without attenuating the originally required original signal.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention is a method for processing a video signal which is band-compressed and encoded using a plurality of adjacent pixels as one block.
A difference between pixel levels adjacent in the block in the horizontal or vertical direction is detected, a pixel whose difference in pixel level is larger than a predetermined value T1 is set as an edge pixel, a pixel whose difference in pixel level is smaller than a predetermined value T2 is set as a flat pixel, A pixel whose difference in pixel level is larger than the predetermined value T2 and smaller than T1 is set as an intermediate pixel, and when edge pixels and flat pixels are mixed in the same block, some or all of the pixels are reduced so that the level difference between the flat pixels is reduced. After the level of the flat pixel is corrected, the signal is output, and has an effect that only the mosquito noise can be effectively reduced without attenuating the originally required original signal.

According to a second aspect of the present invention, the correction of the flat pixel level is performed only when the flat pixels are continuous in at least three pixels in the horizontal or vertical direction, and only the mosquito noise is corrected. It has the effect of enabling effective reduction.

According to a third aspect of the present invention, in the correction of the flat pixel level, an average value level of a plurality of continuous flat pixels is set to a level of a part or all of the flat pixels. Has the effect of effectively reducing only mosquito noise.

According to a fourth aspect of the present invention, the correction of the flat pixel level is such that the average value of three consecutive flat pixels is set to the middle pixel level, and only the mosquito noise is effectively reduced. It has the effect of doing.

According to a fifth aspect of the present invention, the correction of the flat pixel level is obtained by filtering a plurality of continuous flat pixels by weighting each pixel, and a part or all of the flat pixels are obtained. The level is corrected every time, and has an effect of effectively reducing only mosquito noise.

According to a sixth aspect of the present invention, there is provided a signal reproducing apparatus for reproducing a video signal.
The signal processing method according to any one of 2, 3, 4, and 5, wherein the signal is output by the signal processing method, and a signal reproducing apparatus that outputs a signal in which only mosquito noise is effectively reduced can be obtained. Having.

Hereinafter, an embodiment of the present invention will be described with reference to FIG.
This will be described with reference to FIG. (Embodiment 1) Assume that 8 × 8 pixels are one encoding processing block in order to simplify the description of the embodiment of the present invention.

FIG. 1 is a schematic diagram of a pixel level, which schematically shows data in a horizontal direction. Here, pixels 1 to
8, pixels 9 to 16 are each one block. Further, the pixel levels of the pixels 1 to 8 and the pixels 9 to 16 are represented by a to h and i to p, respectively, and are shown in the upper part of FIG. 1, and the differences between the pixels are represented by A to O, and are shown in the lower part of FIG. That is,
A = | ab |, B = | bc |, C = | cd |, and so on.

First, a method of selecting a pixel to be corrected will be described.
The present invention detects differences A to O between pixels, and discriminates an edge pixel when it is larger than a certain threshold level T1, a flat pixel when it is smaller than another threshold level T2, and a middle pixel between them. I do. T1 for 8-bit system
About 120 and T2 about 20.

In the case of the embodiment, the edge pixel is a pixel which generates a pixel difference (C, H, J in the example of FIG. 1) larger than the threshold level T1. , 4, 8, 9, 10, 11 are selected for the time being. However, since only pixels in a block are targeted, pixels between blocks are excluded. Therefore, pixels 8 and 9 are excluded, and edge pixels such as pixels 3, 4 and
10 and 11. This is because the presence of an edge between blocks does not cause mosquito noise.

A flat pixel has a pixel difference (A, B, E, F, G, I, K,
L, M, O), the flat pixel portion includes pixels 1, 2, 3 and pixels 5, 6, 7, 8
, The pixels 9, 10, the pixels 11, 12, 13,
The portion 14 and the pixels 15 and 16 are selected for the time being.

The intermediate pixel has a threshold level T1 ·
The pixels 4, 5, 14, and 15 are selected as the intermediate pixels because the pixels generate the pixel difference (D, N) between T2.

Here, as a result of the image evaluation, the inventor has found a method for correcting all flat pixels and a method for correcting flat pixels excluding edges and intermediate pixels.

In the embodiment of FIG. 1 described in detail here, when all the flat pixels are corrected, the target pixels are pixels 1, 2, 3, 5, 6, 7, 8, 9, 10,. 11, 12,
13, 14, 15, and 16.

When edge pixels and intermediate pixels are excluded, pixels 1, 2, 6, 7, 8, 9, 12, 13, and 16 are obtained.

Next, a method of correcting a pixel to be corrected will be described.
The first correction method completely replaces the level of a pixel in a flat portion with the average value of the pixel in the flat portion. FIG. 2 is a schematic pixel level diagram showing a case where all the flat pixels are corrected. That is, in the case of the first correction method, pixel 1 = pixel 2 = pixel 3 = (a + b + c) / 3 pixel 5 = pixel 6 = pixel 7 = pixel 8 = (e + f + g + h)
/ 4 Pixel 9 = Pixel 10 = (i + j) / 2 Pixel 11 = Pixel 12 = Pixel 13 = Pixel 14 = (k + 1 +
m + n) / 4 pixel 15 = pixel 16 = (o + p) / 2.

FIG. 3 is a schematic pixel level diagram showing a case where flat pixels excluding edge pixels and intermediate pixels are corrected by the first correction method. Pixels 3, 5, 10, 11, 14, and 15 are not corrected as compared with FIG. 2 in which all the flat pixels are corrected.

In the case of the embodiment using the first correction method,
Although there is a feature that the noise reduction effect is high, the influence on the original signal is likely to occur.

Next, a second correction method will be described with reference to a schematic pixel level diagram shown in FIG. This is because 3 flat pixels
The correction is performed only when the number of consecutive pixels is equal to or more than one, and the correction of the flat pixel level is performed so that the average value level of three continuous flat pixels is set to the level of the middle pixel of the flat pixels. . That is, in this case, among the flat pixels, the pixels to be corrected are pixels 2, 6, 7, 12, and 13.
5 pixels. Therefore, pixel 2 = (a + b + c) / 3 pixel 6 = (e + f + g) / 3 pixel 7 = (f + g + h) / 3 pixel 12 = (k + 1 + m) / 3 pixel 13 = (l + m + n) / 3.

Next, a third correction method will be described. this is,
This is basically the same as the second correction method, and the correction target pixels are five pixels 2, 6, 7, 12, and 13, but the coefficients of the correction values are changed. That is, the coefficient is obtained by slightly increasing the weight of the middle pixel level. As a result, the division value becomes 4, which is a coefficient suitable for digital IC. In addition, the adverse effect on the original signal is reduced.
That is, pixel 2 = (a + 2b + c) / 4 pixel 6 = (e + 2f + g) / 4 pixel 7 = (f + 2g + h) / 4 pixel 12 = (k + 21 + m) / 4 pixel 13 = (l + 2m + n) / 4.

In the embodiment according to the above-described second and third correction methods, the pixel 16 is excluded from the correction target. However, when the flat pixel portion including the adjacent pixel 17 is formed, the pixel 16 is not corrected.
May be the object of correction. In this case, the correction value is pixel 16 = (o + p + q) / 3: the case of the second correction method or pixel 16 = (o + 2p + q) / 4: the case of the third correction method.

In the above-described embodiment, a description has been given of a pixel in the horizontal direction. However, the present invention is not limited to this. The correction of the present invention is performed in all or any of the horizontal and vertical directions of the screen. It goes without saying that it is a thing.

The thresholds T1 and T2 are not limited to the values described in the embodiment, and T1 and T2 may have the same value.

Further, the values of the thresholds T1 and T2 may be adaptively changed according to the screen, for example, according to the noise amount, the recording / reproducing mode, and the like.

The block size is not limited to 8 × 8, but may be anything such as 4 × 4.

Further, it goes without saying that the coefficients of the filter are not limited to the coefficients described in the embodiment, but can be arbitrarily taken.

Also, the correction itself according to the present invention is turned ON,
It may be turned off and controlled based on information such as a compression ratio and motion detection.

The description of the present embodiment is exactly the same for a signal reproducing apparatus which outputs after performing processing using the present invention.

[0035]

As described in detail above, according to the present invention,
Even when mosquito noise occurs at the edge of the image, it is possible to selectively filter only flat pixels without filtering the original signal surrounding the edge, which is the original signal, as a result. It is possible to effectively reduce only mosquito noise without attenuating a necessary original signal.

[Brief description of the drawings]

FIG. 1 is a schematic pixel level diagram illustrating an embodiment of the present invention.

FIG. 2 is a schematic pixel level diagram illustrating a first correction method according to the present invention.

FIG. 3 is a schematic pixel level diagram illustrating a first correction method according to the present invention.

FIG. 4 is a schematic pixel level diagram illustrating a second correction method according to the present invention.

Claims (6)

[Claims] 1. A method for processing a video signal which is band-compressed and encoded using a plurality of adjacent pixels as one block, wherein a difference between levels of horizontally or vertically adjacent pixels in a block is detected during decoding. , The difference between pixel levels is a predetermined value T
A pixel whose pixel level difference is smaller than a predetermined value T2 is a flat pixel, a pixel whose pixel level difference is larger than a predetermined value T2 and smaller than T1 is an intermediate pixel, and an edge pixel is in the same block. A signal processing method comprising: correcting the level of some or all of the flat pixels so as to reduce the level difference between the flat pixels, and outputting the corrected data when the flat pixels and the flat pixels coexist. 2. The correction of the flat pixel level is as follows.
2. The signal processing method according to claim 1, wherein the method is performed only when at least three or more pixels continue in the horizontal or vertical direction. 3. The flat pixel level correction, wherein an average value level of a plurality of continuous flat pixels is set as a level of some or all of the flat pixels.
The signal processing method as described. 4. The signal processing method according to claim 1, wherein the correction of the flat pixel level is performed by setting an average value of three consecutive flat pixels to a middle pixel level. 5. A method for correcting a level of a flat pixel, wherein the level of a plurality of flat pixels is obtained by filtering a plurality of continuous flat pixels by weighting each pixel, and the level is corrected for a part or all of the flat pixels. The signal processing method according to claim 1, wherein 6. A signal reproducing apparatus for reproducing a video signal, wherein the video signal is processed by a signal processing method according to any one of claims 1, 2, 3, 4, and 5, and is output. Signal playback device.