Video Signal Transform Kernel Selection by Block Dimensions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video signal processing methods struggle to achieve efficient coding of video signals, particularly in transforming blocks where the choice of transform kernel is not optimized for the block size.
Innovation Solution
A method and apparatus for processing video signals that dynamically determine and apply horizontal and vertical transform types based on the width and height of transform blocks, respectively, using predefined transform types such as DCT-2 and DST-7.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed transform kernel is used for all block sizes, then device complexity is reduced, but coding efficiency deteriorates
Solution Approach 1:
The patent implements dynamic transform kernel selection where the transform type (DST-7 or DCT-2) is adaptively chosen based on the transform block size. Specifically, DST-7 is applied when the block width or height is less than 32 pixels, while DCT-2 is used for larger blocks. This dynamic adaptation resolves the contradiction by allowing the system to optimize coding efficiency for different block sizes without requiring complex manual configuration.
Solution Approach 2:
The patent changes the transform kernel parameter based on the block size parameter. By establishing a relationship between block dimension (width or height) and transform type selection, the system automatically adjusts the transform parameters to match the characteristics of each block, thereby improving coding efficiency while maintaining manageable device complexity through rule-based parameter adaptation.
2Productivity
If transform kernel is selected based on block size, then coding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent applies different transform kernels to different local regions (block sizes) based on their specific characteristics. Small blocks (width or height < 32) use DST-7 which is optimized for their size, while larger blocks use DCT-2. This local quality approach improves coding efficiency by matching the transform to the block characteristics without requiring global system complexity.
Solution Approach 2:
The system automatically changes the transform kernel parameter based on the block size parameter, using simple threshold-based rules (32 pixels). This parameter change mechanism achieves improved coding efficiency through adaptive selection while keeping device complexity low by using straightforward conditional logic rather than complex optimization algorithms.
Data Source
AI summary
A method for processing a video signal comprises the steps of: obtaining, from the video signal, an intra sub-partitions (ISP) mode flag indicating whether the ISP mode is applied to a current block; when the ISP mode is applied to the current block, obtaining, from the video signal, an ISP partitioning flag indicating a partitioning direction of the current block; partitioning the current block into a plurality of transform blocks on the basis of the partitioning direction; generating residual blocks of the transform blocks by performing an inverse transformation on each of the transform blocks; and restoring the current block on the basis of the residual blocks.