Boundary-Based Template Refinement for Adaptive Intra Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding and decoding technologies face challenges in efficiently determining intra prediction modes due to fixed template sizes that require larger line buffers than necessary, leading to increased signaling overhead and reduced performance.
Innovation Solution
Adjusting the length of templates for intra mode derivation based on available reference lines, block position, and second template length to enhance intra mode derivation, allowing for reduced signaling overhead and improved prediction performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a fixed template size is used for intra mode derivation, then the algorithm is simple to implement, but the line buffer size must be larger than necessary, increasing memory requirements and signaling overhead
Solution Approach 1:
The template length is made dynamic rather than fixed. The encoder and decoder adjust the template length based on the number of available reference lines and the block position relative to CTU boundaries. This allows the template to adapt to different coding scenarios, using only the necessary number of reference lines, thereby reducing line buffer size requirements while maintaining implementation feasibility through conditional logic
Solution Approach 2:
The template length parameter is changed from a fixed value to a variable that depends on available reference lines and block position. By changing this parameter dynamically, the system reduces the maximum line buffer size needed while maintaining the simplicity of the algorithm through standardized adjustment rules that both encoder and decoder follow
2Reliability
If a larger line buffer is allocated to accommodate fixed template sizes, then proper operations of intra mode derivation algorithms are ensured, but signaling overhead increases
Solution Approach 1:
The template length is dynamically adjusted to match the actual number of available reference lines, ensuring that intra mode derivation algorithms operate reliably with the exact data they need. This dynamic adaptation eliminates the need for oversized line buffers and reduces signaling overhead by only transmitting information about the actual template length used, rather than accommodating a fixed maximum size
Solution Approach 2:
By changing the template length parameter based on available reference lines and block position, the system ensures reliable algorithm operation with minimal buffer size. The parameter adjustment rules are standardized between encoder and decoder, ensuring consistent behavior without requiring additional signaling for buffer size management
3Productivity
If template length is adjusted based on available reference lines and block position, then signaling overhead is reduced and prediction performance is improved, but the algorithm complexity increases
Solution Approach 1:
The template length parameter is adjusted based on simple conditions (number of available reference lines and block position relative to CTU boundaries). These parameter changes improve prediction performance by using only relevant reference data, while the conditional logic remains straightforward enough to avoid significant increases in algorithm complexity
Solution Approach 2:
Different template lengths are applied to different blocks based on their specific characteristics (position relative to CTU boundaries and available reference lines). This local adaptation improves prediction performance for each block without requiring a completely complex algorithm, as the adjustment rules are simple and localized to each block's context
Data Source
AI summary
Encoding and/or decoding a block of a video frame may be based on a template associated with the block. The length of the template may be determined based on at least one of: a position of the block relative to a boundary, available reference samples above the boundary, and/or the length of a second template of one or more other template derivation algorithms. The length may be adjusted to improve prediction efficiencies and accuracies of template-based predictions.