A multi -reference prediction method, device, storage medium and terminal in the frame

Abstract

The invention discloses an intra-frame multi-reference line prediction method, device, storage medium and terminal. The method includes: a coding end calculates a rate-distortion cost corresponding to a basic reference line; the coding end uses a plurality of different relative reference lines to predict Basic reference line, calculate the sum of squares of pixel differences between the prediction block corresponding to each relative reference line and the basic reference line; the encoding end calculates the rate-distortion cost of the relative reference line with the smallest square sum; the encoding end selects the rate-distortion cost The smallest reference line is used as the optimal reference line for intra-frame prediction; the encoding end encodes the identifier corresponding to the optimal reference line and the intra-frame prediction mode into the code stream; the decoding end parses the code stream to obtain the reference line identifier and intra-frame prediction mode; the decoding end determines the optimal reference line based on the reference line identifier; the decoding end performs prediction according to the optimal reference line and the intra-frame prediction mode. Therefore, by adopting the embodiments of the present application, noise caused by prediction of a single reference line can be removed and bits consumed by the identification of multiple reference lines can be reduced, thereby improving the performance of intra-frame prediction.

Description

technical field [0001] The present invention relates to the field of digital signal processing, in particular to an intra-frame multi-reference line prediction method, device, storage medium and terminal. Background technique [0002] In the current image and video coding technology, compression is mainly achieved by removing spatial redundancy, temporal redundancy (video field) and coding redundancy. The encoder effectively combines a variety of algorithms to achieve high compression performance. Related techniques include predictive coding, transform coding, quantization, and entropy coding. The prediction includes intra-frame prediction and inter-frame prediction. The principle of intra-frame prediction is mainly to use the correlation of the video signal in the spatial domain. The intra prediction technology mainly includes five steps: obtaining reference pixels, filtering the reference pixels, obtaining the predicted value according to the selected prediction mode, fi...

AI Technical Summary

Problems solved by technology

If there are noisy points in the reference line, it will greatly affect the prediction performance

Although VVC proposes multi-reference line prediction and expands more reference lines, it needs to identify each reference line number, which increases the bit cost.

The multi-angle intra prediction method can predict more directional textures. In the current multi-angle intra prediction method, the AVS3 intra prediction only has the nearest single reference line, and the predicted value is easily affected by the noise points in the reference line. Influence, while the multi-reference line in VVC requires more bit identification, which affects the prediction performance, resulting in reduced intra-frame prediction performance

Images