Video coding reference block filtering method

Abstract

The invention relates to a video coding reference block filtering method. The method includes: A, calculating the texture complexity C of a current reference pixel block Bref; B, obtaining a transformation coefficient matrix Iref through DCT of the reference pixel block Bref, and calculating the transverse and vertical frequency distribution intensities of the Bref and the corresponding transverse and vertical filtering factors; C, establishing a filtering template U having the same size with the reference pixel block Bref; D, multiplying the obtained transformation coefficient matrix Iref and a matrix of the filtering template U to obtain a filtered coefficient matrix, and performing IDCT to obtain the filtered reference pixel block; and E, determining whether the filtered reference pixel block is written back to the reference pixel block Bref according to the fact that whether the mean absolute difference and the complexity of the filtered reference pixel block are reduced. According to the method, smooth processing of image textures can be effectively performed, the performance of image coding is obviously enhanced, the filtering of the pixel blocks with good quality and reference is avoided, and the goal of adaptive improvement of the quality of reference images is achieved.

Description

technical field [0001] The invention relates to a processing method of video coding, in particular to a video coding reference block filtering method. Background technique [0002] The compression of digital images and videos needs to remove redundant information as much as possible during the compression process. In video compression, the time domain redundancy is very considerable, so eliminating the time domain redundancy as much as possible has long been a hot research topic. Most of the current mainstream encoders use inter-frame prediction technology, and use methods such as motion estimation and motion compensation to reduce time-domain redundancy. Inter-frame predictive coding technology is the best guarantee for obtaining good coding results. [0003] At present, mainstream encoders use discrete cosine transform (DCT) as a key technology to transform and compress the residual between images and images. Cosine transform (DCT), quantization, entropy coding. Since ...

AI Technical Summary

Problems solved by technology

After prediction and quantization coding, the problem of boundary mismatch between blocks may be introduced, resulting in incoherent boundaries

This not only affects the user's subjective experience and reduces the coding quality, but also weakens its own reference, objectively increasing the difficulty of coding subsequent images

Images