Dynamic Co-located Luma Pixel Selection for Video Edge Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in achieving high class accuracy for edge offset corresponding to color component pixels, which affects overall encoding performance.
Innovation Solution
A data processing method and apparatus that determine classification mode information for a to-be-encoded block, including an extended co-located luma reconstructed pixel and a target classification mode, to improve edge class accuracy. This involves determining an edge class based on the extended co-located luma reconstructed pixel and the target classification mode, and offsetting the reconstructed pixel accordingly to enhance encoding performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed position of co-located luma component is used in classification, then the device complexity is reduced, but the class accuracy of edge offset for color component pixels deteriorates
Solution Approach 1:
The patent makes the co-located luma reconstructed pixel position dynamic by selecting it from a target region centered on the true co-located luma reconstructed pixel based on classification mode information. This allows the system to adapt the pixel position to different classification modes (horizontal, vertical, diagonal, anti-diagonal), improving class accuracy while maintaining reasonable device complexity through mode-based adaptation rather than exhaustive search.
Solution Approach 2:
The patent changes the parameter of co-located luma reconstructed pixel position from fixed to variable, where the position is selected from a target region based on classification mode information. This parameter change enables the system to achieve higher class accuracy by adapting the pixel position to the specific edge orientation being classified, without significantly increasing device complexity.
2Ease of manufacture
If the co-located luma reconstructed pixel position is fixed, then the encoding process is simplified, but the overall encoding performance deteriorates
Solution Approach 1:
The patent introduces dynamic selection of co-located luma reconstructed pixels from a target region based on classification mode information. This dynamic approach improves overall encoding performance by adapting to different edge orientations, while the encoding process remains relatively simple through mode-based selection rather than complex optimization algorithms.
Solution Approach 2:
The patent changes the position parameter of the co-located luma reconstructed pixel from fixed to variable, selected from a target region based on classification mode. This parameter change improves encoding performance by better matching the pixel position to the edge orientation, while maintaining encoding process simplicity through predefined selection rules.
3Speed
If a fixed co-located luma reconstructed pixel is used, then the classification process is faster, but the edge offset accuracy for color component pixels deteriorates
Solution Approach 1:
The patent implements dynamic selection of the co-located luma reconstructed pixel from a target region based on classification mode information. This dynamic selection improves edge offset accuracy by adapting to different edge orientations, while maintaining relatively fast processing speed through mode-based selection rather than exhaustive search or complex optimization.
Solution Approach 2:
The patent changes the pixel position parameter from fixed to variable, where the position is selected from a target region based on classification mode. This parameter change improves edge offset accuracy by better aligning the pixel position with the edge orientation, while keeping the classification process relatively fast through predefined selection rules.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Embodiments of this application provide a data processing method and apparatus, and a device. The method includes: determining classification mode information corresponding to a to-be-encoded block in video data, the classification mode information including an extended co-located luma reconstructed pixel and a target classification mode corresponding to a color component pixel in the to-be-encoded block, and the extended co-located luma reconstructed pixel belonging to a target region centered on a true co-located luma reconstructed pixel of the color component pixel; determining an edge class corresponding to the color component pixel based on the extended co-located luma reconstructed pixel and the target classification mode; and offsetting a reconstructed pixel of the color component pixel based on the edge class to obtain an offset reconstructed pixel, and performing encoding processing on the to-be-encoded block based on the offset reconstructed pixel. According to the embodiments of this application, class accuracy of edge offset corresponding to the color component pixel can be improved, thereby improving overall encoding performance for the video data.