Intra Prediction Reference Block Selection for Bandwidth-Limited Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intra (picture) prediction coding modes in multimedia data processing suffer from poor coding performance due to limited reference data blocks, particularly in hardware implementations, which restrict the use of adjacent blocks as references and impose challenges on memory bandwidth and decoding efficiency.
Innovation Solution
A method that extends the selection scope of reference data blocks by adaptively selecting from both global and local reference data block sets, based on the attributes of the multimedia data frame, allowing for predictive coding that utilizes both adjacent and non-adjacent coded data blocks, thereby optimizing the coding performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only local reference data blocks (adjacent blocks) are used for intra prediction coding, then hardware implementation complexity is reduced, but coding performance deteriorates due to limited reference data availability
Solution Approach 1:
The reference data block selection is segmented into two distinct sets: local reference data blocks (adjacent to current block) and global reference data blocks (non-adjacent to current block). This segmentation allows the system to selectively use different reference sources based on content characteristics, thereby improving coding performance while maintaining manageable hardware complexity through structured organization.
Solution Approach 2:
The reference data block selection mechanism is designed to be universal by supporting both local and global reference blocks through a unified selection process. The system can adaptively choose from multiple reference sources (adjacent and non-adjacent blocks) depending on the prediction mode and content requirements, making the hardware implementation versatile without proportionally increasing complexity.
2Manufacturing precision
If global reference data blocks (non-adjacent blocks) are included in addition to local reference blocks, then coding performance is improved through expanded reference selection, but memory bandwidth requirements increase
Solution Approach 1:
The reference data block selection is made dynamic and adaptive rather than static. The system determines whether to use local or global reference blocks based on content characteristics, prediction mode, and other contextual factors. This dynamic selection optimizes memory bandwidth usage by accessing global reference blocks only when necessary, rather than continuously accessing all possible reference sources.
Solution Approach 2:
The selection of reference data blocks is controlled by changing parameters such as prediction mode indicators and block position information. By using parameter-based selection (e.g., block vector differences, mode flags), the system can efficiently control which global reference blocks to access without requiring continuous high-bandwidth memory access, thus reducing overall memory bandwidth requirements while maintaining coding performance.
3Manufacturing precision
If adaptive selection between global and local reference blocks is implemented, then coding performance is enhanced, but processing complexity increases
Solution Approach 1:
The adaptive selection process relies on changing key parameters such as block position coordinates, prediction mode indicators, and block vector differences. By base the selection on parameter comparisons rather than complex content analysis, the system achieves adaptive reference block selection with manageable processing complexity. The parameters provide a compact representation that simplifies the decision-making process.
4Quantity of substance
If adjacent coded data blocks are used as reference blocks, then memory bandwidth requirements are reduced, but the selection scope of reference blocks is limited
Solution Approach 1:
The reference block pool is segmented into local (adjacent) and global (non-adjacent) categories. This segmentation enables the system to primarily use memory-efficient local blocks while selectively incorporating global blocks when broader reference scope is needed. The segmented structure maintains memory efficiency by organizing reference blocks by their spatial relationship to the current block.
Data Source
AI summary
Example embodiments provide a multimedia data processing method and apparatus, a computer device, a computer-readable storage medium, and a computer program product. The method includes: obtaining a global reference data block set and a local reference data block set, the global reference data block set and the local reference data block set being associated with a first data block in a multimedia data frame, the global reference data block set including a coded data block having no adjacent relationship with the first data block in the multimedia data frame, and the local reference data block set including a coded data block having an adjacent relationship with the first data block in the multimedia data frame; and determining a reference data block for coding or decoding of the first data block from the global reference data block set and the local reference data block set.