Video Transform Block Encoding With Selective Secondary Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video signal processing technologies face challenges in efficiently handling high spatial resolution, high frame rate, and high dimensionality of scene representation, leading to increased computational complexity and memory requirements.
Innovation Solution
A method and apparatus for encoding and decoding video signals that selectively apply a non-separable secondary transform based on the size of the transform block, omitting it for blocks smaller than or equal to 4×4 and applying it to larger blocks, using transforms like reduced secondary transform (RST), sparse orthonormal transform (SOT), Givens rotation-based transforms, and combinations of DCT-2, DST-7, and DCT-8.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a non-separable secondary transform is applied to all transform blocks to improve transform efficiency, then video signal processing quality is improved, but computational complexity increases
Solution Approach 1:
The patent applies different transform strategies to different regions of the transform block based on local characteristics. Specifically, the non-separable secondary transform is applied only to certain regions (e.g., 4x4 regions) of the transform block rather than uniformly to the entire block, optimizing processing quality where needed while reducing complexity in other areas
Solution Approach 2:
The transform block is divided into multiple 4x4 regions, and the non-separable secondary transform is selectively applied to specific segments rather than the entire block. This segmentation allows the system to process only the necessary portions with high computational cost operations
2Reliability
If transform processing is performed on all block sizes to maintain consistent quality, then video quality is maintained, but processing time increases
Solution Approach 1:
The patent applies the non-separable secondary transform partially rather than completely to all transform blocks. By applying the transform only to specific 4x4 regions within larger blocks or to blocks meeting certain criteria, the system maintains adequate video quality while significantly reducing processing time compared to applying the transform universally
3Measurement precision
If memory is allocated for storing transform coefficients from non-separable secondary transform to improve accuracy, then transform accuracy is improved, but memory requirements increase
Solution Approach 1:
The patent stores transform coefficients locally only for the specific 4x4 regions where the non-separable secondary transform is applied, rather than allocating memory for entire large transform blocks. This localized storage approach maintains transform accuracy for processed regions while minimizing memory allocation
Data Source
AI summary
A video signal encoding method according to an embodiment of the present invention comprises checking a transform block including residual samples except a prediction sample from a picture of the video signal, generating transform coefficients through a transform for the residual samples of the transform block based on a size of the transform block, and performing quantization and entropy coding the transform coefficients, wherein generating the transform coefficients includes, applying a forward primary transform to each of a horizontal direction and vertical direction of the transform block including the residual samples, and not applying a forward non-separable secondary transform to the transform block to which the primary transform has been applied when the size of the transform block is smaller than or equal to 4×4, and applying the forward non-separable secondary transform to the transform block when the size of the transform block is greater than 4×4.