Low-Frequency Non-Separable Transform Signaling Reduces Coding Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques face inefficiencies in signaling overhead due to explicit signaling of last transform coefficient positions and transform indices, particularly in advanced codecs like HEVC and the next generation of video coding standards such as VVC/H.266.
Innovation Solution
Implementing location-based restrictions for signaling the last transform coefficient position and transform indices for Low-Frequency Non-separable Transforms (LFNSTs), allowing the video encoder and decoder to infer zero-out patterns and reduce the need for explicit signaling of these positions, thereby reducing bitstream size and improving coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If explicit signaling of last transform coefficient positions and transform indices is used, then transform coefficient coding can be performed, but signaling overhead increases
Solution Approach 1:
The patent extracts only the necessary transform index information from the full transform coefficient data, signaling only the difference between the actual transform index and a predicted index rather than the complete index value. This reduces the number of bits required for signaling while maintaining the ability to reconstruct the original transform coefficients.
Solution Approach 2:
Instead of signaling the absolute transform index directly, the patent inverts the approach by signaling the difference or offset from a predicted index. The decoder reconstructs the actual index by adding this signaled difference to the predicted index, thereby reducing signaling overhead.
2Measurement precision
If more transform indices are signaled explicitly, then transform accuracy improves, but bitstream size increases
Solution Approach 1:
The patent changes the parameter being signaled from the absolute transform index to the difference between the actual index and a predicted index. This parameter transformation reduces the magnitude and variability of the signaled values, allowing for more efficient entropy coding and reducing the average bitstream size while preserving transform precision.
Solution Approach 2:
The patent signals only the necessary portion of transform index information (the difference from predicted index) rather than the complete index value. This partial signaling approach provides sufficient information for accurate reconstruction while minimizing bitstream size.
Data Source
AI summary
A video decoder determines, based on a block size of a current block and a low-frequency non-separable transform (LFNST) syntax element, a zero-out pattern of normatively defined zero-coefficients. The LFNST syntax element is signaled at a transform unit (TU) level. Additionally, the video decoder determines transform coefficients of the current block. The transform coefficients of the current block include transform coefficients in an LFNST region of the current block and transform coefficients outside the LFNST region of the current block. As part of determining the transform coefficients of the current block, the video decoder applies an inverse LFNST to determine values of one or more transform coefficients in the LFNST region of the current block. The video decoder also determines that transform coefficients of the current block in a region of the current block defined by the zero-out pattern are equal to 0.