METHOD FOR IMAGE DECODING BASED ON MUTUAL-COMPONENT LINEAR MODEL PREDICTION IN IMAGE CODING SYSTEM AND DEVICE FOR IT Russian patent published in 2023 - IPC H04N19/186 H04N19/11 

FIELD: video decoding.

SUBSTANCE: coding and decoding, in particular, arrangement and method for decoding video based on the prediction of a co-component linear model (CCLM) in a video coding system. A video decoding method is performed by a decoding apparatus and consists of the following steps: extracting one of the plurality of CCLM prediction modes as the CCLM prediction mode of the current chrominance unit; extracting the number of adjacent chrominance samples of the current chrominance unit based on the CCLM prediction mode of the current chrominance unit, the size of the current chrominance unit, and a specific value; extracting adjacent chrominance samples in the number of samples; calculating CCLM parameters based on adjacent chrominance samples and adjacent luminance samples after downsampling; extracting the prediction samples for the current chrominance unit based on the CCLM parameters and the downsampled luminance samples, and generating reconstructed samples for the current chrominance unit based on the prediction samples, wherein the specific value is extracted as 2.

EFFECT: highly efficient image compression for efficient transmission, storage and reproduction of image information with high resolution and high quality.

11 cl, 42 dwg, 41 tbl