Adaptive Spatially Varying Transform for Lower Video Coding Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing spatially varying transform (SVT) schemes for video coding suffer from heavy signaling overhead and increased encoder complexity due to the large number of candidate positions for transform block placement, and the transform blocks may not adequately cover the major residuals in larger residual blocks.
Innovation Solution
Adaptive use of simplified SVT types (SVT-V and SVT-H) with reduced candidate positions and discrete sine transforms (DST) for horizontal and vertical transforms, based on the SVT type and position, to improve decoding quality and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple SVT types with many candidate positions are used to capture residual distribution, then decoding quality is improved, but signaling overhead and encoder complexity increase significantly
Solution Approach 1:
The patent changes the parameters of SVT by limiting the transform block size ratios to specific values (1:1, 1:2, or 2:1) and restricting candidate positions to only four locations (top-left, top-right, bottom-left, bottom-right corners). This parameter simplification reduces the number of possible transform configurations from many combinations to just four, significantly reducing signaling overhead while maintaining effective residual coverage.
Solution Approach 2:
Instead of exhaustively testing all possible transform block positions and sizes (which would be excessive action), the patent applies partial action by selecting only the four corner positions and specific size ratios. This partial approach is sufficient to capture the major residual energy while avoiding the complexity of evaluating all possibilities.
2Productivity
If transform block size is reduced to capture major residual, then coding efficiency is improved, but the transform block may not adequately cover residual content in larger residual blocks
Solution Approach 1:
The patent introduces dynamic adaptability by allowing the transform block size to be dynamically selected from three possible ratios (1:1, 1:2, or 2:1) based on the residual block characteristics. This dynamic sizing enables the transform block to adapt to different residual distributions, ensuring adequate coverage while maintaining coding efficiency.
Solution Approach 2:
The patent applies local quality by focusing transform blocks at the four corner positions of the residual block, where residual energy is most likely to be concentrated. By strategically placing transform blocks at these critical locations rather than uniformly distributing them, the method achieves effective residual coverage with fewer, larger transform blocks.
Data Source
AI summary
A video decoding method includes determining a usage of an SVT-vertical (V) or an SVT-horizontal (H) for a residual block; determining a transform block position of a transform block of the residual block; determining a transform type of the transform block, wherein the transform type indicates a horizontal transform and a vertical transform for the transform block, wherein at least one of the horizontal transform or the vertical transform is a discrete sine transform (DST)-7; and reconstructing the residual block based on the transform type, the transform block position, and transform coefficients of the transform block.