JP2887168B2 - Method for reducing block distortion of reproduced image - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] The present invention relates to a method for reducing a block distortion of a reproduced image in an image encoding transmission system in which an image is encoded and transmitted in block units by a required prediction method including intra-frame prediction. In order to reduce block distortion that occurs at the edge of a predicted block, when playing back an image coded in block units using multiple types of prediction methods including intra-frame prediction, the block that has undergone intra-frame prediction For a target pixel near the boundary with another block adjacent to the block and causing block distortion, the reproduction value of a pixel adjacent to the target pixel and the adjacent block And a value for interpolating between the reproduction values of the pixels adjacent to the target pixel based on the reproduction value of the pixel adjacent to the target pixel. It is configured to perform filter processing for reducing block distortion by changing to a value to be interpolated.

[Industrial applications]

The present invention relates to a method for reducing block distortion of a reproduced image for removing block distortion of a reproduced image in an image encoding transmission system in which an image is encoded and transmitted in units of blocks by a prediction method including at least intra-frame prediction.

[Conventional technology]

As a moving image band compression encoding method for realizing a video conference system, etc., a method of predictively encoding an image in block units is known.
A coding method using any one of prediction methods such as motion compensation prediction or intra-frame prediction, or adaptively selecting these inter-frame prediction, motion compensation prediction, and intra-frame prediction for each block according to screen motion. There is a coding method. Further, the present invention is referred to as “variable block size motion compensation system” of the present invention.
There is a method proposed in the Japanese Patent Application dated to perform coding by adaptively selecting a prediction method and a block size according to the properties of each part such as a still part or a moving part of an image.

[Problems to be solved by the invention]

In the method of encoding an image in these block units,
For example, when intra-frame prediction is used to encode a block, in a screen reproduced on the receiving side, a block in which the intra-frame prediction is performed becomes noticeable at an edge of the block so that block distortion that becomes unfamiliar with other adjacent blocks is noticeable. Tend to be. Such block distortion tends to be conspicuous even though the prediction error is not large, and deteriorates the quality of an actual reproduction screen more than expected from the actual prediction error.

Therefore, an object of the present invention is to provide an image encoding / transmission system in which an image is encoded and transmitted in units of blocks using a plurality of types of prediction methods including intra-frame prediction, and a block generated at an edge of an intra-frame predicted block in a reproduced image. It is to reduce distortion.

[Means for solving the problem]

FIG. 1 is a principle explanatory diagram for explaining a method for reducing block distortion of a reproduced image in image encoding transmission according to the present invention.

According to the method for reducing block distortion of a reproduced image according to the present invention, when reproducing an image encoded in a block unit by a plurality of types of prediction methods including intra-frame prediction, a block subjected to intra-frame prediction is adjacent to the block. For the target pixel near the boundary with another block and causing block distortion, the reproduction value of the pixel adjacent to the target pixel in the same block and the reproduction value of the pixel in the adjacent other block Filter processing for reducing block distortion by obtaining a value for interpolating between reproduction values of adjacent pixels based on the reproduction value of adjacent pixels and changing the value of the target pixel to the value to be interpolated. Configured to do.

The value to be interpolated is, for the target pixel, a reproduction value of a pixel adjacent to the target pixel in the adjacent other block, and a reproduction value of a pixel adjacent to the target pixel in the other block within the same block. At the opposite position, the average value of the target pixel and the reproduction value of the adjacent pixel is available.

Further, the filter processing may be configured to perform interpolation only when a difference between a reproduction value of a pixel of a block on which intra-frame prediction has been performed and a reproduction value of a pixel of a block adjacent to the block exceeds a certain threshold. It is possible.

[Action]

Prediction was performed using different prediction methods within one screen, for example, intra-frame prediction method and inter-frame prediction method, such as performing coding by adaptively switching a plurality of prediction methods including intra-frame prediction within a screen. In this case, an error (distortion mode) having a different property occurs between the intra-frame prediction and the inter-frame prediction. Specifically, for example, the mode of the blur of the reproduced image is different from the error having the different property. For this reason, the appearance of errors differs between intra-frame prediction and inter-frame prediction even if the coding efficiency is improved compared to the case where coding is performed using the same prediction method in one screen. For these reasons, the boundaries of the blocks are conspicuous and visually obstructive. For this reason, either intra-frame prediction or inter-frame prediction,
Or, at both boundaries, the visual disturbance is less noticeable when a filter for reducing distortion is applied. However,
For example, when a non-linear average filter is applied as a filter for reducing distortion, for example, when such a filter is applied to all boundaries between intra-frame prediction and inter-frame prediction, "blur" at the boundary increases and the image quality deteriorates rather. . Therefore, it is desirable to apply a filter to an area as small as possible. Therefore, commonly used image block coding methods (eg, H.261 and MPEG)
Then, since “the number of blocks for intra-frame prediction 数 the number of blocks for inter-frame prediction”, the method for reducing block distortion of a reproduced image according to the present invention pays attention to this, and applies Only on the receiving side, filter processing is performed on the receiving side.

In FIG. 1 (1) and (2), block 3 is a block for which intra-frame prediction has been performed.
And 4 are blocks that have been subjected to intra prediction or other prediction.

As shown in FIG. 1 (1), when the value of the pixel at the edge of the block 3 is noticeably different from the value of the pixel at the edge of the other adjacent blocks 1, 2, and 4, Block distortion will occur at the edges. Therefore, in the present invention, for example, the following filter processing is performed on the block 3 on which intra-frame prediction has been performed, thereby suppressing the occurrence of block distortion.

That is, when the difference between the pixels in the block 3 and the pixels in the block 4 exceeds a certain threshold value, interpolation processing is performed so that the value of the pixel is an average value of the pixels in the block 3 and the pixels in the block 4. When the difference between the pixel and the pixel in the block 1 exceeds a certain threshold, the average value of the pixel and the pixel in the block 3 is set as the pixel value. When the difference between the pixel and the pixel in the block 2 exceeds a certain threshold value, the average value of the pixel and the pixel in the block 1 is set as the pixel value.

Further, a description will be given, with reference to FIG. 1 (2), of the processing in the case where the distortion becomes conspicuous even in the block 3 as a result of the processing as described above. That is, if the difference between the pixels in the block 3 exceeds the threshold value, the average value of the pixels is set as the pixel value.

As described above, for only the block region for which intra-frame prediction has been performed, interpolation is performed using the pixel values of the block and the pixel values of the neighboring blocks adjacent thereto.

〔Example〕

 Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 2 is a flowchart showing a method for reducing block distortion of a reproduced image according to one embodiment of the present invention. In this embodiment, a required prediction method including intra prediction in a block unit is adaptively selected, encoded and transmitted.
Screen its magnitude, as shown in FIG. 3, the horizontal direction B _x, longitudinally is of B _y, the pixel in the screen F (i,
j).

First, both the variables i and j indicating the pixel position are set to 1 (steps S2 and S3), and it is determined whether or not the focused pixel F (i, j) belongs to the area where the intra-frame prediction has been performed ( Steps
S4). If it is outside the region, the process proceeds to step S11, performs a calculation of j = j + 1, if j does not exceed the size of B _y of the screen (step S12), the repeated processing in step S4.

If the pixel F (i, j) is in the intra prediction region, the pixels F (i−1, j) and F (i +, j) to the left or right of the target pixel F (i, j) are further added. ) Protrudes from the screen to the left or right in the horizontal axis direction. This can be determined by satisfying the following conditional expression (step S5).

i + 1 ≦ _Bx and ij ≧ 1 In this conditional expression, the pixel position (ij) on the left side of the target pixel F (i, j) is 1 or more, and the pixel position (i + j) on the right side is the screen. by determining is within the size B _x, it is to determine that the target pixel F (i, j) is in the horizontal axis direction on the inside of the screen.

When the target pixel F (i, j) is inside the screen, the absolute value of the difference between the target pixel F (i, j) and the right pixel F (i + 1, j) or the target pixel F (i, j) ) And its left pixel F
Whether the absolute value of the difference from (i−1, j) is equal to or greater than a predetermined threshold value TH is determined by the following equation (step S6).

| F (i, j) −F (i + 1, j) | ≧ TH F (i, j) −F (i−1, j) ≧ TH If these differences exceed the threshold value TH, the target pixel Interpolation processing in the horizontal axis direction is performed on F (i, j) (step S7).
Proceed to.

The interpolation process is performed by calculating the average value of the left and right pixels of the target pixel F (i, j) by the following equation and using the average value as the value of the target pixel F (i, j).

F (i, j) = 1/2 [F (i + 1, j) + F (i−1, j)] After performing such interpolation processing in the horizontal axis direction, the pixel is further added to the target pixel F (i, j) or The lower pixels F (i, j + 1), F (i, j−
It is determined whether 1) protrudes upward or downward from the screen in the vertical axis direction. This can be determined by satisfying the following conditional expression (step S8).

j + 1 ≦ B _y and j-1 ≧ 1 pixel F (i, j) The conditional expression as an object under the pixel position (j-1) is not less than 1, and the upper pixel position (j + 1) is by determining is within the screen size B _y, it is to determine that the target pixel F (i, j) is in the vertical axis direction inside the screen.

When the target pixel F (i, j) is inside the screen, the absolute value of the difference between the target pixel F (i, j) and its upper pixel F (i, j + 1) or the target pixel F (i, j) ) And its lower pixel F
Whether the absolute value of the difference from (i, j-1) is equal to or greater than a predetermined threshold value TH is determined by the following equation (step S9).

| F (i, j) −F (i, j + 1) | ≧ TH | F (i, j) −F (i, j−1) | ≧ TH If these differences exceed the threshold TH, It is assumed that interpolation processing in the vertical axis direction is performed on the target pixel F (i, j) (step S10).
Go to 11.

The interpolation processing in the vertical axis direction is based on calculating the average value of the pixels above and below the target pixel F (i, j) by the following equation to obtain the value of the target pixel F (i, j).

F (i, j) = 1/2 [F (i, j + 1) + F (i, j-1)] After performing the interpolation processing in the vertical axis direction, the value of j is increased by one, and Repeat the operation (Steps S4 to S1
2), by repeating this process until the value of j exceeds the screen size B _y (step S12), the all pixels of the screen in the vertical axis direction in the horizontal direction i position performs interpolation processing.

Horizontal axis i When the longitudinal axial position of the pixel exceeds the screen size B _y at position (step S12), the next one to increase the value of i (step S13), and reset the value of j to 1 (Step S3), the above operation is repeated again (Step S3).
S3-S14). The value of this process i (i.e. the horizontal axis direction position of the pixel) is repeated until more than the screen size B _x (step S14).

As described above, the interpolation processing in the vertical and horizontal directions can be performed on the blocks for which intra-frame prediction has been performed among all the pixels in the area of the screen size B _x × B _y , and the interpolation processing can be performed between adjacent blocks or the own block. It is possible to reduce block distortion generated in the inside.

Various modifications are possible in practicing the present invention. For example, in the above-described embodiment, the case where the interpolation is performed in the vertical axis direction and the horizontal axis direction has been described. However, the interpolation is not limited to this, and it is of course possible to perform the interpolation in the oblique direction. Further, in the above-described embodiment, the interpolation process is performed when the difference between the target pixel and the pixels located on the left, right, upper and lower sides thereof exceeds a predetermined threshold. However, the present invention is not limited to this. The interpolation processing may be uniformly performed on all the pixels.

[Effects of the Invention] According to the present invention, since a filter process is performed on a portion where intra-frame prediction has been performed, block distortion is not conspicuous even when a reproduction screen is viewed on the receiving side, and a natural image can be reproduced. .

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the principle of a method for reducing the block distortion of a reproduced image according to the present invention. FIG. 2 is a flowchart showing the processing procedure of the method for reducing the block distortion of a reproduced image according to an embodiment of the present invention. FIG. 3 is a view for explaining the size of the screen. In the figure, 1-4 ... blocks ... pixels

Claims (2)

(57) [Claims] When reproducing an image coded in a block unit by a plurality of types of prediction methods including intra-frame prediction, a block subjected to intra-frame prediction is located near a boundary with another block adjacent to the block. For a target pixel causing block distortion, the reproduction value of a pixel adjacent to the target pixel in the same block and the reproduction value of a pixel adjacent to the target pixel in the other adjacent block are calculated. A block distortion reduction method for a reproduced image in which a value for interpolating between reproduction values of those adjacent pixels is obtained based on the calculated value and the value of the target pixel is changed to the value to be interpolated, thereby performing a filtering process for reducing block distortion. . 2. The value to be interpolated is a reproduction value of a pixel adjacent to the target pixel in the adjacent other block for the target pixel and a reproduction value of the target pixel in the other block within the same block. 2. The method according to claim 1, wherein the average value is the average value of the reproduction value of the pixel adjacent to the target pixel at a position opposite to the pixel adjacent to the target pixel.