LFNST Index Context Coding for Video Signaling Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding technologies face challenges in reducing signaling overhead for transform-related information, particularly for low-frequency non-separable transforms (LFNST), which can lead to increased data transmission and storage requirements without significant improvements in prediction accuracy or complexity.
Innovation Solution
The proposed solution involves context coding both bins for the LFNST index instead of context coding one bin and bypass coding the other, allowing the video coder to determine the LFNST index with reduced signaling overhead, thereby decreasing the amount of video data transmitted with minimal loss in prediction accuracy and complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If context coding is applied to both bins for LFNST index instead of mixed context and bypass coding, then signaling overhead is reduced and data transmission efficiency is improved, but encoding and decoding complexity increases
Solution Approach 1:
The patent changes the coding parameter from mixed context-bypass coding to uniform context coding for both bins of LFNST index. This parameter change in the coding strategy reduces signaling overhead by ensuring both bins are transmitted with full contextual information, improving data transmission efficiency despite the increased processing complexity.
2Measurement precision
If LFNST is applied to video blocks, then prediction accuracy is improved, but the amount of data to be transmitted increases
Solution Approach 1:
The patent extracts only the essential LFNST index information (two bins) from the full transform parameters and transmits them using context coding. This selective extraction approach maintains prediction accuracy by preserving the critical LFNST identification data while minimizing the transmitted data volume through efficient entropy coding.
Data Source
AI summary
A method of decoding video data includes receiving encoded data for a current block and decoding N bins for a low-frequency non-separable transform (LFNST) index from the encoded data. The N bins comprises a first bin and a second bin. Decoding the N bins comprises context decoding each bin of the N bins. The method further includes determining the LFNST index using the N bins and decoding the encoded data to generate transform coefficients. The method further includes applying an inverse LFNST to the transform coefficients using the LFNST index to produce a residual block for the current block and reconstructing the current block of the video data using the residual block and a prediction block for the current block.