Non-Separable Transform Selection for High-Resolution Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently processing next-generation video content with high spatial resolution, high frame rate, and high dimensionality, requiring improved transform techniques for spatial domain video signals to enhance memory storage and processing power.
Innovation Solution
The method involves determining a non-separable transform set index based on the intra-prediction mode of a current block, applying a non-separable transform matrix to the left top region of the block, and using horizontal and vertical direction transforms, with specific index values assigned to different intra-prediction modes to optimize transform efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional transform techniques are used for next-generation video content, then processing complexity increases due to high spatial resolution and high frame rate, but transform efficiency remains insufficient
Solution Approach 1:
The transform process is segmented into multiple stages: primary transform, secondary transform, and tertiary transform. Each stage processes specific frequency components with appropriate transform types, allowing complex high-resolution video to be handled in manageable segments rather than as a single complex operation
Solution Approach 2:
The patent implements dynamic transform selection where the transform type and parameters are adaptively chosen based on the characteristics of the current video block, including prediction mode, block size, and frequency content. This allows the system to optimize transform efficiency for each specific region rather than applying a uniform complex transform throughout
2Manufacturing precision
If transform techniques are applied to high spatial resolution video, then memory storage and processing power requirements increase significantly
Solution Approach 1:
The patent extracts and processes different frequency components separately through multiple transform stages. By separating low-frequency and high-frequency components and applying appropriate transforms to each, the system reduces the memory required to store and process the complete high-resolution video data at once
Solution Approach 2:
The patent transforms spatial domain video data into frequency domain representation through multiple transform stages. This dimensional transformation from spatial to frequency domain allows for more efficient compression and reduced memory requirements while maintaining high spatial resolution quality
3Measurement precision
If multiple transform types are used for different prediction modes, then transform accuracy improves, but system complexity increases
Solution Approach 1:
The patent applies different transform types and parameters to different regions of the video based on local characteristics such as prediction mode, block size, and frequency content. This local adaptation ensures high transform accuracy for each specific region while avoiding the need to complexity throughout the entire system
Solution Approach 2:
The system dynamically selects transform types based on the prediction mode and block characteristics. For example, different transform kernels are selected for intra-prediction versus inter-prediction modes, and for different block sizes, optimizing accuracy without requiring all transform types to be active simultaneously throughout the system
Data Source
AI summary
Embodiments of the present disclosure provide a method and device for processing a video signal. A method for decoding an image signal according to an embodiment of the present disclosure comprises the steps of: determining a non-separable transform set index indicating a non-separable transform set used for a non-separable transform of a current block from among non-separable transform sets predefined on the basis of an intra-prediction mode of the current block; determining. as a non-separable transform matrix, a transform kernel indicated by a non-separable transform index for the current block from among transform kernels included in the non-separable transform set indicated by the non-separable transform set index; and applying the non-separable transform matrix to an upper left region which of the current block, which is determined according to the width and height of the current block, wherein each of the predefined non-separable transform sets includes two transform kernels.