Adaptive Transform Kernel Selection for Intra Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in selecting and deriving transform kernels for intra prediction, leading to suboptimal compression and reconstruction of video data.
Innovation Solution
A method for video decoding that determines a transform kernel set based on the histogram of occurrence of intra prediction modes in neighboring samples of a current block, and partitions the block adaptively for intra and inter predictions, using decoder-side intra mode derivation to enhance the reconstruction process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If transform kernel selection is performed using conventional methods, then the decoding process is simple, but compression efficiency and reconstruction quality are suboptimal
Solution Approach 1:
The patent applies dynamics by making the transform kernel selection adaptive rather than fixed. The system dynamically selects transform kernels based on the actual intra prediction modes used in each block, allowing the transformation to adapt to different image characteristics and content types, thereby improving compression efficiency without using overly complex selection mechanisms
Solution Approach 2:
The patent changes the parameter of transform kernel selection by deriving it from the intra prediction mode parameters. Instead of using fixed or randomly selected kernels, the system modifies the transform kernel parameters based on the specific prediction mode (e.g., angular, planar, DC) used for each block, optimizing the transformation for each local region's characteristics
2Manufacturing precision
If transform kernel derivation is simplified, then the processing time is reduced, but reconstruction quality deteriorates
Solution Approach 1:
The system applies self-service by having the transform kernel derivation process utilize information already available during the prediction stage. The intra prediction mode, which is determined during the prediction process anyway, is used to derive the transform kernel, eliminating the need for separate complex analysis steps and avoiding additional processing time while improving reconstruction quality
Solution Approach 2:
The patent applies preliminary action by determining the transform kernel derivation mode and selecting appropriate kernels during the prediction stage rather than after reconstruction. This preliminary selection allows the transform to be optimized for the specific prediction characteristics before the actual reconstruction occurs, improving quality without adding post-processing time
3Productivity
If intra prediction modes are derived using decoder-side techniques, then coding efficiency is improved, but the complexity of the decoding algorithm increases
Solution Approach 1:
The patent applies feedback by using the actual intra prediction modes that are determined during the prediction process to feed back into the transform kernel selection. This feedback mechanism ensures that the transform kernels are optimized for the specific prediction characteristics of each block, improving coding efficiency through adaptive transformation while using a relatively simple feedback loop rather than complex algorithms
Data Source
AI summary
An aspect of the disclosure provides a method of video decoding. For example, a coded video bitstream is received. The coded video bitstream includes coded information of a plurality of pictures. Based on coded information of a current block in a current picture, it is determined that the current block is coded using an intra prediction that generates prediction samples of the current block based on reference samples in the current picture. At least a first transform kernel set is determined based on a histogram of occurrence (HoC) of intra prediction modes in neighboring samples of the current block. Based on the coded information of the current block, a residual block of the current block is calculated according to at least the first transform kernel set. The current block is reconstructed based on the residual block and the intra prediction of the current block.