JPH0346482A - Method and apparatus for eliminating block distortion for moving picture coding - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

Detailed Description of the Invention (Industrial Application Field) The present invention divides a video signal such as a video conference or a video phone call into pixel blocks, and performs calculations such as conversion for each pixel block.
The present invention relates to a method and apparatus for performing quantization for data compression, reducing the bit rate, and digitally transmitting the data.

(Prior Art) Many attempts have been made in the past to remove superimposed noise during image signal transmission without impairing the clarity of the original image. Among these, many image filters have been proposed that remove random noise and the like while preserving edges in images.

In such a filter, for each pixel, the signal level of surrounding pixels is referred to, calculations are performed based on the signal level, and a process of overwriting the pixel with a new signal level is repeatedly performed over the entire image.

Examples of the above calculations include (a) one that refers to the signal levels of the relevant pixel and its surrounding pixels, multiplies the values by a coefficient for smoothing, and calculates a weighted average; (b) calculates the weighted average of the signals of the surrounding reference pixels. (c) A method that determines the edge direction based on the arrangement of signal levels and adaptively selects a combination of weighted average coefficients to preserve this. There is a method that sets the new signal level of the pixel.

(Problem to be Solved by the Invention) In an encoding method that divides a moving image signal into pixel blocks and performs calculations such as conversion and quantization for data compression for each pixel block, when the transmission bit rate is low, Since coarse quantization is performed, an unnatural discontinuous state (block distortion) occurs at the pixel block boundary, which adversely affects image quality.

In order to perform image processing that removes such distortion and also prevents the original edges of the image from becoming as blunt as possible, it is necessary to perform edge preservation smoothing processing. However, since conventional edge-preserving filters perform calculations uniformly over the entire image, there is a problem in that block distortions are treated as edges and effective smoothing cannot be achieved.

The present invention focuses on the boundaries of pixel blocks and aims to effectively remove block distortion while preserving the original edges of the image.

(Means for Solving the Problems) A feature of the present invention is that it can be independently connected to a low bit rate moving image decoder, and the original edges of the image can be determined based on the output of the decoder and the size of the pixel block. The present invention provides an edge-preserving block distortion removal device that can effectively remove block distortion without causing any dullness.

FIG. 1 shows a block diagram of this device. An input unit 11 that selects pixel signal levels around block boundaries, an input signal delay unit 12 that arranges input signals as a reference pixel string, and performs smoothing processing on the block boundaries based on the signal levels of reference pixels. A smoothing determination unit 14 receives a smoothing necessity (ONloFF) control signal from the smoothing determination unit, which determines whether or not the block distortion is detected, that is, performs block distortion detection.
), it is an edge-preserving block distortion removal device that is configured with a smoothing calculation unit 13 that performs a calculation for smoothing the pixel, that is, performs block distortion removal, and outputs the calculation processing result.

(Embodiment) In the input unit 11 that selects pixel signal levels around block boundaries, signal levels of four pixels arranged symmetrically across the block boundary and arranged in a direction perpendicular to the block boundary are input. An example of selective input is shown in FIG. As shown in FIG. 2, signals of four pixels in the vertical direction are inputted to the block boundaries in the horizontal direction, and signals of four pixels in the horizontal direction are inputted to the block boundaries in the vertical direction. The output signal of the input section 11 is arranged in the input signal delay section 12 as a reference pixel array.

The smoothing determination unit 14 determines and detects whether or not smoothing processing is required at the block boundary, that is, block distortion detection processing, based on the signal level of the reference pixel string using the following determination algorithm.

FIG. 3 shows an example of the relationship between pixel arrangement and signal level in calculations and determinations in the algorithm. Determination parameters do, d, d2 are calculated from the signal level of the reference pixel.

d, = lso -S21 dl = ls+ -5ol di: ls'o -s? 1 From the comparison of these values and the thresholds ThA and Tha, d0
≦ThA: Smoothing non-smoothing (OFF) d, >TlIAand (d, ≦ThB and di
≦Th8): Smoothing required (ON) do>ThA and (dt>Thl Or d
a>Tha): Determine whether smoothing is required (OFF) and whether smoothing is necessary (ONloFF).

In the smoothing calculation unit 13, only when receiving a smoothing required (ON) signal from the smoothing determination unit 14, for example, simultaneously performs the following smoothing process, that is, block distortion removal process, on two pixels that directly contact the block boundary. conduct. This is a weighted averaging process. Taking FIG. 3 as an example, the smoothing target pixel X. , x2 new signal level S. -new1s Shizuku.

, W is So-new = (s+ + 2XSo +
So) / 4so-r,. ” (So + 2xSo
+S:)/4.

The test video CLAIRE, which is actually used in the C:CITT standardization work, was encoded at 64KBIT/S3
As a result of applying the present invention using the above-mentioned algorithm to two frames, it was confirmed that the S/N ratio of the decoded image to the original image to which the present invention was not applied was improved. Furthermore, a subjective examination with the naked eye revealed that the visual impression was improved more than the S/N ratio. In addition, the smoothing required (ON) rate in the smoothing judgment section at this time was 17.30%.
It is.

As a guide to show the effect of the present invention in preserving edges, the present invention was applied directly to the original image before encoding, and as a result, the smoothing required (ON) rate was 2.42%, and the S/N ratio was 67.18 dB.
It was found that almost no unnecessary smoothing processing was performed.

(Effects of the Invention) The present invention eliminates unnatural discontinuity (block distortion) at pixel block boundaries that occurs when the encoding bit rate is low in video encoding, by preserving the original edges of the image. can be removed effectively. This means that visual discomfort caused by block distortion can be removed.

Furthermore, since the present invention does not use variable parameters used during encoding, such as motion compensation amount and quantization step size, it can be independently connected to a decoder, making it easy to add functions using the device of the present invention. .

Furthermore, since a relatively simple smoothing determination algorithm and smoothing calculation algorithm are used, the apparatus configuration is simple, and edge preserving block distortion removal processing can be performed effectively and easily.

[Brief explanation of drawings]

FIG. 1 shows a device block diagram of the present invention. FIG. 2 shows the positional relationship of pixel blocks, block boundaries, and reference pixels. FIG. 3 shows the relationship between the signal level of the reference pixel and the 0N10FF determination parameter. 11... Input section, 12... Input signal delay section, 13... Smoothing calculation section, 14... Smoothing determination section.

Claims (2)

[Claims] (1) In a method in which an image is divided into multiple pixel blocks and quantization is performed for each block, calculations on multiple pixels at the boundaries of pixel blocks are performed to determine whether the image is the original edge part or the block due to quantization. A pixel characterized in that it includes a first step of detecting whether the block distortion is present, and a second step of performing smoothing processing to remove the block distortion on pixels that are in contact with a block boundary when block distortion is detected. Block distortion removal method. (2) In a device that divides an image into multiple blocks and performs quantization for each block, it is possible to determine whether the area in question is the original edge of the image by calculating multiple pixels at the boundary of the pixel block. a determining means for determining whether the block distortion is due to block distortion; and a smoothing processing means for removing the block distortion from pixels adjacent to a block boundary when the determining means determines that the block distortion is caused by block distortion. Pixel block distortion removal device.