Spatial Geometric Partitioning for Accurate Low-Bitrate Intra Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards such as VVC and AVS3 require improvements in the implementation of Spatial Geometric Partitioning Mode (SGPM) and Angular Weighted Prediction (AWP) to enhance video compression efficiency.
Innovation Solution
Implementing extensions to SGPM with regression, fusion with multiple intra prediction modes, adaptive blending area sizes, and conditional matrix-based intra prediction, along with improvements to AWP, to optimize motion prediction in video coding systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If SGPM partitions a coding block into two parts with intra prediction, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies segmentation by partitioning the coding block into two distinct parts using a partitioning line, where each part is predicted using different intra prediction modes. This allows the encoder to capture complex boundaries and edges more accurately while managing computational complexity through selective application of the technique.
Solution Approach 2:
The patent introduces dynamic elements by allowing adaptive selection of partitioning modes, prediction modes for each partition, and blending area sizes based on content characteristics. The encoder can dynamically adjust the blending area size and select from multiple prediction modes to optimize between accuracy and complexity for different video regions.
2Productivity
If multiple prediction modes and blending areas are used in SGPM, then video compression efficiency is improved, but encoding complexity increases
Solution Approach 1:
The patent changes parameters by introducing multiple intra prediction modes (planar, angular, vertical, horizontal) and variable blending area sizes that can be adjusted based on the video content. These parameter variations enable better compression efficiency by adapting to different texture and edge characteristics while providing a framework to manage encoding complexity.
Solution Approach 2:
The patent creates a universal framework that can handle various prediction scenarios using the same SGPM structure. The multi-functional design allows the same partitioning and blending mechanism to work with different prediction modes and area sizes, providing versatility across different video content types without requiring separate specialized algorithms.
3Stability of the object's composition
If blending area size is increased for SGPM, then prediction smoothness is improved, but processing overhead increases
Solution Approach 1:
The patent makes the blending area size dynamic rather than fixed, allowing it to be adjusted based on the specific video content and partitioning configuration. This dynamic adjustment enables smooth transitions in regions that benefit from blending while avoiding unnecessary processing overhead in regions where small or zero blending areas are sufficient.
Solution Approach 2:
The patent applies local quality by allowing different blending area sizes for different partitions and regions within the coding block. Areas with complex boundaries or edges can have larger blending areas for smoother transitions, while homogeneous regions can use smaller blending areas, optimizing the balance between smoothness and processing overhead locally.
Data Source
AI summary
A Versatile Video Coding (“VVC”) and later standard encoder and a VVC and later standard decoder are provided, and a third-generation Audio and Video coding standard (“AVS3”) and later standard encoder and an AVS3 and later standard decoder are provided configuring one or more processors of a computing system to perform spatial geometric partitioning, including extension of regression spatial geometric partitioning mode (“SGPM”) to intra prediction; fusion of SGPM with multiple intra prediction modes; adaptive blending area size for SGPM; conditional matrix-based intra prediction for SGPM; and implementing any or all of the preceding for angular weighted prediction (“AWP”).