Adaptive Secondary Transform Cores for Variable Video Block Sizes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in intra prediction and motion compensation, particularly in handling various block sizes and transform coefficients, leading to suboptimal compression ratios and increased bandwidth requirements.
Innovation Solution
The implementation of a secondary transform process that includes determining a secondary transform core based on intra prediction modes and indices, allowing for the generation and application of primary transform coefficient blocks with reduced sizes, and utilizing context-dependent entropy coding for improved efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional transform processes are used for various block sizes, then device complexity is reduced, but video coding efficiency and compression ratio deteriorate
Solution Approach 1:
The patent implements dynamic selection of transform cores based on block size and intra prediction mode. Different transform cores (e.g., 4×4, 8×8, 16×16) are selected adaptively according to the current block characteristics, allowing the system to optimize coding efficiency for each specific case while managing complexity through conditional logic rather than fixed complex structures
Solution Approach 2:
The patent changes transform parameters (transform core selection, transform type) based on block size and prediction mode. By varying these parameters dynamically, the system achieves higher coding efficiency without requiring a fundamentally complex transform architecture, as the base transform mechanism remains relatively simple while its application varies
2Productivity
If secondary transform is applied to all block sizes, then compression ratio improves, but device complexity and processing overhead increase
Solution Approach 1:
The patent applies secondary transform selectively to specific regions or blocks based on their characteristics (block size, prediction mode, coefficient distribution). Rather than uniformly applying secondary transform to all blocks, the system identifies which blocks would benefit most from it and applies the transform only there, improving compression ratio while limiting the increase in processing overhead
Solution Approach 2:
The patent implements partial application of secondary transform - not to all blocks but to a subset that meets certain criteria. This partial action approach achieves compression improvement where needed without incurring the full processing overhead of universal application, balancing compression ratio gains against complexity costs
3Quantity of substance
If context-dependent entropy coding is used, then bit requirements decrease, but processing complexity increases
Solution Approach 1:
The patent performs preliminary classification of transform coefficients based on their characteristics (position, magnitude, pattern) before entropy coding. By pre-grouping coefficients into contexts with similar statistical properties, the system enables more efficient entropy coding with fewer bits, while the classification overhead is managed through simple rules rather than complex processing
Data Source
AI summary
Aspects of the disclosure provide methods, apparatuses, and non-transitory computer-readable storage mediums for video encoding/decoding. In a method, prediction information for a current block is decoded. The prediction information indicates a first intra prediction mode and a secondary transform index, based on which a secondary transform core is determined. A first transform coefficient block is de-quantized from the prediction information. A size of the first transform coefficient block is less than a size of the secondary transform core. A part of a second transform coefficient block is generated based on the first transform coefficient block and the secondary transform core. A size of the second transform coefficient block equals the size of the secondary transform core. A size of the part of the second transform coefficient equals the size of the first transform coefficient block. The current block is reconstructed based on the part of the second transform coefficient block.