Chroma Transform Selection Using Reference-Block Direction Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current video encoding and decoding technologies, particularly in the context of the Versatile Video Coding (VVC) standard, face challenges in selecting an appropriate transform set for chroma components due to the presence of directional patterns, leading to suboptimal decorrelation of residual signals when cross-component linear model (CCLM) is used.
Innovation Solution
An improved process for selecting a transform set for prediction blocks, specifically using information from a reference block to determine a directional pattern, allowing for the selection of an appropriate Non-Separable Secondary Transform (NSST) set based on this pattern, which is applied in both encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a fixed transform set is used for chroma components, then the encoding process is simple, but the decorrelation of residual signals is suboptimal when directional patterns are present
Solution Approach 1:
The patent implements dynamic transform set selection by determining directional patterns in prediction blocks and selecting transform sets based on these patterns. Instead of using a fixed transform set, the system adapts the transform set choice to the local directional characteristics of the chroma components, thereby improving decorrelation of residual signals while managing complexity through pattern-based selection criteria
Solution Approach 2:
The patent changes the parameter of transform set selection from a fixed value to a dynamically determined value based on directional patterns. By analyzing the directional pattern parameters in prediction blocks and mapping them to appropriate transform sets, the system optimizes the transform parameters to match the local signal characteristics, improving residual signal decorrelation
2Reliability
If transform set selection is based on directional pattern analysis, then decorrelation of residual signals is improved, but the encoding complexity increases
Solution Approach 1:
The patent performs preliminary directional pattern determination on prediction blocks before selecting transform sets. By pre-analyzing the directional characteristics and preparing the transform set selection in advance, the system improves decorrelation while managing encoding complexity through structured preliminary analysis rather than complex real-time optimization
Solution Approach 2:
The patent applies different transform sets to different regions based on local directional patterns in prediction blocks. Instead of using a uniform transform approach, the system adapts the transform characteristics to match local signal properties, improving decorrelation where directional patterns exist while maintaining efficiency through localized rather than global optimization
3Productivity
If the same transform set is used for all prediction blocks, then the decoding process is straightforward, but the video encoding and decoding efficiency is suboptimal
Solution Approach 1:
The patent implements a feedback mechanism where the encoder determines directional patterns and selects transform sets, then signals this selection to the decoder. The decoder uses this feedback information to select the corresponding inverse transform sets, ensuring that both encoder and decoder use matching transform pairs while maintaining straightforward processing through explicit signaling
Solution Approach 2:
The patent creates a universal transform selection mechanism that works for both encoding and decoding processes. The same directional pattern analysis and transform set selection logic is applied in both encoder and decoder, with the added function of signaling the selection to ensure synchronization, thereby improving overall video encoding and decoding efficiency
Data Source
AI summary
A process for selecting a transform set for a prediction block. The process can be used in both an encoder and a decoder. For example, the process can be used in both an encoder and a decoder for a prediction block that has been predicted from a reference block. In some embodiments, both the prediction block and the reference block are intra blocks.