Directional Intra-Prediction Mapping for Rectangular Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing QTBT partitioning framework in video coding standards does not efficiently handle rectangular video coding blocks, as it does not account for the asymmetry in aspect ratios and availability of reference samples, leading to suboptimal directional intra-prediction modes.
Innovation Solution
The solution involves extending the set of directional intra-prediction modes based on the aspect ratio of the block and the availability of reference samples, using a signaling indicator to map and enable or disable modes, and employing a mirroring procedure to complement existing modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the QTBT partitioning framework uses a fixed set of directional intra-prediction modes, then the implementation complexity is low, but the coding efficiency for rectangular blocks is suboptimal
Solution Approach 1:
The patent implements dynamic selection of directional intra-prediction modes based on the aspect ratio of rectangular blocks. The encoder adapts the set of available prediction modes according to block geometry, transitioning from a static fixed set to a dynamic configuration that optimizes coding efficiency for different block shapes while maintaining manageable complexity through conditional logic.
Solution Approach 2:
The patent applies different directional intra-prediction modes to different regions or orientations of rectangular blocks based on their aspect ratios. By tailoring the prediction mode selection to local block characteristics (aspect ratio categories), the system achieves optimized coding efficiency for each specific block type without requiring complete redesign of the prediction framework.
2Measurement precision
If the set of directional intra-prediction modes is extended to account for aspect ratio asymmetry, then the prediction accuracy improves, but the signaling overhead increases
Solution Approach 1:
The patent changes the parameters of directional intra-prediction modes based on block aspect ratio. By adjusting which modes are available and how they are indexed according to block geometry parameters, the system achieves higher prediction accuracy for rectangular blocks while managing signaling overhead through parameter-based mode selection rather than transmitting full mode information.
Solution Approach 2:
Instead of signaling the actual directional mode directly, the patent inverts the approach by signaling an aspect ratio category that indirectly determines the available modes. This inversion reduces signaling overhead by encoding block geometry characteristics rather than full prediction mode information, allowing the decoder to reconstruct the appropriate mode set.
3Adaptability or versatility
If mirroring procedure is applied to complement existing prediction modes, then the adaptability to rectangular blocks improves, but the computational complexity increases
Solution Approach 1:
The patent applies asymmetry by using mirroring procedures to generate complementary prediction modes for rectangular blocks. By reflecting existing symmetric prediction modes across block diagonals or axes, the system creates asymmetric mode sets that adapt to rectangular block geometries, improving versatility while keeping computational complexity manageable through reuse of base mode calculations.
Data Source
AI summary
An encoding apparatus is described for signaling an extension directional intra-prediction mode within a set of directional intra-prediction modes, the set of directional intra-prediction modes comprising predetermined directional intra-prediction modes and the extension directional intra-prediction mode. The encoding apparatus comprises a mode mapping unit selecting a predetermined directional intra-prediction mode, the selected predetermined directional intra-prediction mode being associated with an intra mode index, and mapping the extension directional intra-prediction mode onto the selected predetermined directional intra-prediction mode. A signaling unit generates a signaling indicator comprising at least one of the intra mode index and a flag value. An intra-prediction unit intra-predicts pixel values of pixels of a rectangular video coding block on the basis of the extension directional intra-prediction mode for providing a predicted rectangular video coding block. An encoding unit encodes the rectangular video coding block on the basis of the predicted rectangular video coding block.