Sub-Group Secondary Transform Signaling for Video Coefficients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as ITU-T H.265 and JEM, face inefficiencies in signaling secondary transforms for transform coefficients, particularly in determining whether and which transform matrix to apply for intra prediction modes, leading to suboptimal coding efficiency.
Innovation Solution
The proposed techniques enhance transform coefficient coding by efficiently indicating whether a secondary transform is applied to sub-groups of core transform coefficients and selecting an appropriate transform matrix based on intra prediction modes, using improved signaling mechanisms like a CU-level flag and TU-level index, thereby optimizing coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If secondary transforms are applied to transform coefficients, then coding efficiency is improved, but signaling complexity increases
Solution Approach 1:
The transform coefficients are divided into sub-groups (e.g., 4x4 sub-groups within 8x8 or 16x16 blocks), and secondary transforms are applied selectively to individual sub-groups based on local characteristics. This segmentation allows the encoder to apply secondary transforms only where beneficial, reducing the need for extensive signaling while maintaining coding efficiency gains in regions that benefit from secondary transforms.
Solution Approach 2:
Different secondary transform matrices are applied to different sub-groups of transform coefficients based on local characteristics such as prediction mode and coefficient energy distribution. The encoder determines on a per-sub-group basis whether to apply secondary transforms and which matrix to use, allowing local optimization without requiring complex global signaling. This local decision-making reduces signaling overhead while preserving coding efficiency improvements.
2Measurement precision
If multiple transform matrices are provided for different intra prediction modes, then transform precision is improved, but device complexity increases
Solution Approach 1:
A set of candidate secondary transform matrices is pre-defined and standardized in the video coding framework. These matrices are designed in advance to correspond to different intra prediction modes and coefficient patterns. During encoding, the encoder selects from this pre-prepared set based on local characteristics, avoiding the need to compute or signal custom matrices. This preliminary preparation maintains transform precision while reducing runtime complexity and signaling requirements.
Solution Approach 2:
The patent uses compact parameters (such as small index values or flags) to indicate which pre-defined transform matrix should be applied to each sub-group. Instead of signaling full matrix data, the encoder transmits minimal parameter information that references the appropriate matrix from the pre-defined set. This parameter-based selection maintains transform precision across different intra prediction modes while significantly reducing signaling complexity and device complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A video coding device may be configured to perform transform data coding according to one or more of the techniques described herein.