Inter Residual Non-Separable Transforms for Video Coding Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in improving the quality and efficiency of video encoding and decoding, particularly in applying transforms on inter mode block residuals, which can lead to suboptimal computational costs and video quality.
Innovation Solution
Implementing non-separable transform kernels on inter mode residual blocks, with a flag indicating whether to apply these transforms, allowing for improved video quality and reduced computational costs through efficient encoding and decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-separable transforms are applied to inter prediction residuals, then video quality is improved, but computational complexity increases
Solution Approach 1:
The patent implements dynamic selection of transform types (separable vs. non-separable) based on prediction mode and block characteristics. The system adapts the transform processing to match the specific coding scenario, applying non-separable transforms only when they provide quality benefits, thereby balancing video quality improvement with computational complexity management.
Solution Approach 2:
The patent applies different transform types to different blocks based on local characteristics such as prediction mode and block content. Non-separable transforms are selectively applied to specific blocks where they provide the most benefit, rather than uniformly to all blocks, optimizing the trade-off between quality and complexity at the block level.
2Manufacturing precision
If non-separable transform kernels are applied to all inter mode residual blocks, then reconstruction accuracy is improved, but encoding and decoding time increases
Solution Approach 1:
The patent applies non-separable transforms selectively to only those blocks where they provide significant quality improvement, rather than to all blocks. This partial application approach maintains reconstruction accuracy for critical blocks while avoiding the performance penalty of applying complex transforms universally.
Solution Approach 2:
The patent divides the video stream into different block types and applies appropriate transform types to each segment. By segmenting the processing based on block characteristics, the system achieves high reconstruction accuracy where needed while maintaining overall encoding and decoding efficiency.
3Productivity
If a flag is signaled in the bitstream to indicate non-separable transform application, then coding efficiency is improved, but bitstream overhead increases
Solution Approach 1:
The patent uses existing syntax elements and flag structures in the bitstream to convey transform selection information, making these existing elements serve multiple functions. This approach communicates the necessary transform application information without requiring entirely new dedicated syntax, thereby improving coding efficiency while minimizing bitstream overhead.
Solution Approach 2:
The patent combines the transform selection indication with other existing signaling mechanisms in the bitstream. By merging the non-separable transform flag with existing syntax structures, the system conveys additional information without proportionally increasing bitstream size, thus improving coding efficiency while controlling overhead.
Data Source
AI summary
The various implementations described herein include methods and systems for coding video. In one aspect, a method includes receiving a video bitstream that includes a set of inter mode encoded blocks and a corresponding set of transform coefficients. The method includes deriving a set of inter mode residual blocks from the set of transform coefficients. The method includes determining, according to a value of a first indicator in the video bitstream, whether one or more non-separable transform kernels are to be applied to the set of inter mode residual blocks. The method includes applying a first non-separable transform kernel when the indicator has a first value, and forgoing applying the first non-separable transform kernels when the indicator has a second value. The method also includes reconstructing a set of video blocks using the set of inter mode residual blocks and a corresponding set of prediction blocks.