Template matching based method for encoding self-adaption intra-frame prediction filtering

Abstract

The invention relates to a template matching based method for encoding self-adaption intra-frame prediction filtering, belonging to the technical field of image processing. The method comprises the following steps of: acquiring a template of a current encoding block, respectively searching n optimal template matchings in a matching way according to the acquired template; obtaining the prediction of the current encoding block according to the template matchings; obtaining three pixel values of pixel predictions generated in the prediction of the current encoding block according to a standard supply method; computing the mean values of the encoding block predictions obtained by all the template matchings, and using the mean values as targets of the predictions; computing the mean values of n groups of the three pixel values of pixels, using the mean values as the support for generating target values, and computing the optimal self-adaption filter coefficient of a mode according to a least square principle; and generating a predicting value of the intra-frame prediction of the current encoding block. The invention can effectively reduce the residual error after the intra-frame prediction and enhance the objective image quality of decoding by more than 0.2 dB under the condition of same code rate.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a template matching-based adaptive intra-frame predictive filter coding method. Background technique [0002] The current typical video coding algorithm adopts the hybrid coding technology framework based on prediction / transformation that has been used until now. Following this framework, two series of international video coding standards have been formulated: MPEG-X series and H.26X series, forming the H.264 A number of video coding standards represented by / AVC, AVS, VC-1, etc. have been widely recognized by the industry. [0003] Hybrid coding technology is based on Shannong information theory, uses classical set theory as a tool, and uses probability and statistics models to describe information sources. Its compression idea is based on data statistics, which can only remove data redundancy and completely ignore visual redundancy. Remain. The development of this te...

AI Technical Summary

Problems solved by technology

Due to the non-stationarity of the video signal, it is difficult for the fixed-tap filter to meet the change of the image content, resulting in (1, 2, 1) filtering, the residual error of the intra-frame prediction is still relatively large, and it is difficult to compress, which becomes the intra-frame prediction coding The bottleneck that needs to be broken through

Images