Non-Separable Transform Kernel Selection for Adaptive Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding designs do not provide sufficient adaptation to content being coded, as input statistics vary significantly for each transform kernel, leading to suboptimal coding efficiency.
Innovation Solution
Select a non-separable transform kernel set based on prediction type and transform block dimensions, using techniques such as directional intra prediction, template-based intra mode derivation, and spatial geometric partitioning mode to improve video compression performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed kernel set is used for non-separable transforms, then the implementation is simple, but the adaptation to content characteristics is insufficient
Solution Approach 1:
The patent implements dynamic kernel selection by making the transform kernel set dependent on the transform block size. Different kernel sets are selected based on the size of the transform block (e.g., 4x4, 8x8, 16x16, 32x32), allowing the system to adapt to different content characteristics and block dimensions automatically during video coding processing.
Solution Approach 2:
The patent changes the parameter of kernel set selection based on transform block size. By using transform block size as the selecting parameter, the system can switch between different kernel sets (e.g., first kernel set for smaller blocks, second kernel set for larger blocks) to optimize compression performance for different content types and block dimensions.
2Productivity
If transform kernel is selected based on prediction type, then coding efficiency is improved, but selection complexity increases
Solution Approach 1:
The patent uses feedback from the prediction type to guide kernel selection. The prediction type information (intra or inter prediction, directional, planar, etc.) is fed back to determine which kernel set should be applied, creating a closed-loop system that optimizes compression based on the actual prediction characteristics of each block.
Solution Approach 2:
The patent segments the kernel selection process into different categories based on prediction types. By dividing prediction modes into segments (intra prediction, inter prediction, directional, planar, etc.), the system can select appropriate kernel sets for each segment, improving compression efficiency while maintaining manageable selection complexity through structured categorization.
Data Source
AI summary
A device for decoding video data, the device comprising: a memory configured to store the video data; and one or more processors implemented in circuitry, the one or more processors configured to: determine, based on a prediction type associated with a transform block, a non-separable transform kernel set, wherein the non-separable transform kernel set is a non-separable primary transform (NSPT) kernel set or a low-frequency non-separable transform (LFNST) kernel set; select a kernel from the non-separable transform kernel set; apply one or more transforms to the transform block to reconstruct a residual block, wherein applying the one or more transforms comprises applying a non-separable transform to the transform block using the kernel; and reconstruct a block of the video data based on the residual block and a prediction block.