Inter Prediction Luminance Compensation for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of inter prediction in video coding is compromised by luminance differences between current and reference blocks, leading to suboptimal video quality and increased bit rates.
Innovation Solution
An inter prediction method that involves determining luminance compensation blocks based on motion vectors of subblocks within the current affine block, using reconstructed samples from the block edges to update the predicted sample values through a linear model, thereby improving prediction accuracy and reducing residual values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inter prediction is performed based on affine motion model without luminance compensation, then device complexity is reduced, but prediction accuracy deteriorates due to luminance differences between current and reference blocks
Solution Approach 1:
The patent applies preliminary action by performing luminance compensation before the main prediction process. It calculates a linear model coefficient based on luminance differences between current and reference blocks, and uses this coefficient to compensate predicted sample values in advance, thereby improving prediction accuracy without adding significant complexity to the main coding flow
Solution Approach 2:
The patent changes the parameter of predicted sample values by applying a linear model coefficient that adjusts for luminance differences. This parameter transformation compensates for illumination variations between current and reference blocks, improving prediction accuracy while maintaining computational efficiency
2Measurement precision
If luminance compensation is performed on all subblocks, then prediction accuracy is improved, but computational complexity and bit rate increase
Solution Approach 1:
The patent applies local quality by performing luminance compensation selectively rather than uniformly across all blocks. It identifies subblocks with significant luminance differences and applies compensation only to those regions, thereby improving prediction accuracy where needed while reducing overall computational cost
Solution Approach 2:
The patent applies partial action by performing luminance compensation on only certain subblocks rather than all subblocks. It uses a threshold or significance test to determine which subblocks require compensation, applying the linear model coefficient only where luminance differences exceed acceptable levels, thus balancing accuracy improvement with computational efficiency
Data Source
AI summary
The inter prediction method is provided, which includes: obtaining a predicted sample value of a current affine block; determining one or more luminance compensation blocks based on motion vectors of one or more subblocks of the current affine block; determining a linear model coefficient based on a first sample set and a second sample set, where the first sample set includes at least one reconstructed sample neighboring to an edge of the current affine block, the second sample set includes at least one reconstructed sample neighboring to an edge of the one or more luminance compensation blocks, and the sample in the first sample set is in a one-to-one correspondence with the sample in the second sample set; and updating the predicted sample value of the current affine block based on the linear model coefficient, to obtain an updated predicted sample value of the current affine block.