Affine Chroma Prediction Using Mean Luma Motion Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for encoding and decoding high-resolution or high-definition image content struggle with efficient prediction of chroma blocks based on luma blocks, particularly when using affine modes, leading to suboptimal compression and decoding performance.
Innovation Solution
The proposed method involves determining whether a prediction mode of a current block is affine, splitting the luma block into sub luma blocks of a predefined size, calculating a mean luma motion vector from neighboring blocks, and using this vector for chroma block prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If chroma prediction is performed using conventional intra prediction mode of luma block, then prediction process is simple, but prediction efficiency for chroma blocks is insufficient
Solution Approach 1:
The luma block is divided into multiple sub luma blocks (e.g., 4 sub luma blocks) to enable more precise motion vector calculation. Each sub luma block can have its motion vector calculated independently, improving the accuracy of chroma prediction while maintaining manageable process complexity through systematic division.
Solution Approach 2:
Different motion vectors are calculated for different sub luma blocks within the same luma block, allowing each region to have optimized prediction parameters. This local quality approach enables better prediction efficiency for chroma blocks by using region-specific motion characteristics rather than a single global motion vector.
2Device complexity
If motion vector is calculated using only upper-left and lower-right sub luma blocks, then calculation complexity is reduced, but prediction accuracy may be compromised
Solution Approach 1:
Motion vectors from multiple sub luma blocks (upper-left and lower-right) are merged or averaged to produce a unified motion vector for chroma prediction. This combining approach leverages information from multiple regions to achieve both reduced calculation complexity compared to using all possible vectors and maintained or improved prediction accuracy through aggregation of motion information.
Data Source
AI summary
A video decoding method and device for determining whether a prediction mode of a current block is an affine mode; splitting, when the prediction mode of the current block is the affine mode, a luma block of the current block into a plurality of sub luma blocks having a square shape based on a predefined sub block size; determining a mean luma motion vector for four neighboring sub luma blocks among the plurality of sub luma blocks, by using a motion vector of an upper-left sub luma block of the four sub luma blocks and a motion vector of a lower-right sub luma block of the four sub luma blocks; determining the mean luma motion vector to be a motion vector of a current sub chroma block corresponding to the four sub luma blocks; and performing prediction on the current sub chroma block by using the determined motion vector, in a video encoding and decoding process are suggested.