Diagonal Split Intra Prediction for 360-Degree Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding techniques face inefficiencies when dealing with 360-degree video compression, particularly due to the excessive pixel increase and distortion in equirectangular projection formats, which leads to increased data and encoding throughput, and existing intra prediction methods are not effectively operable on blocks with diagonal edges or textures.
Innovation Solution
The introduction of a 'split intra mode' in video encoding and decoding methods, where the predicted area is diagonally divided into two areas, allowing for separate directional or non-directional mode prediction for each section, enhancing spatial compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Weight of stationary object
If equirectangular projection is used for 360-degree video, then the image can be represented in 2D space, but pixel count increases excessively and distortion occurs
Solution Approach 1:
The patent divides the coding block into multiple sub-blocks along diagonal directions and applies different intra prediction modes to each sub-block. This segmentation allows tailored prediction for regions with different diagonal edge characteristics, improving compression efficiency without increasing data size
Solution Approach 2:
The patent applies different intra prediction modes (e.g., first diagonal direction mode, second diagonal direction mode, first non-diagonal mode, second non-diagonal mode) to different sub-blocks based on their specific diagonal edge characteristics. This local quality approach optimizes prediction accuracy for each region while maintaining overall compression efficiency
2Productivity
If existing intra prediction modes are applied to blocks with diagonal edges, then the prediction process is simple, but spatial compression efficiency is reduced
Solution Approach 1:
The coding block is segmented into multiple sub-blocks that can be independently predicted using appropriate intra prediction modes. This segmentation enables the system to handle diagonal edges effectively by applying specialized prediction modes to relevant regions, improving compression efficiency
Solution Approach 2:
The patent dynamically selects from multiple intra prediction modes (first diagonal direction, second diagonal direction, first non-diagonal, second non-diagonal) based on the specific characteristics of each sub-block. This dynamic adaptation allows optimal prediction for different diagonal edge orientations and textures while managing complexity through structured mode selection
3Reliability
If icosahedral or octahedron projection formats are used for 360-degree video, then distortion is reduced, but diagonal edges increase and existing intra modes become less effective
Solution Approach 1:
The patent segments coding blocks into sub-blocks aligned with diagonal directions and applies specialized intra prediction modes to each. This segmentation approach directly addresses the increased diagonal edges in icosahedral and octahedron projections by providing tailored prediction for each diagonal region
Solution Approach 2:
The patent applies different intra prediction modes to different sub-blocks based on their diagonal edge characteristics. This local quality approach enables effective prediction for blocks with diagonal edges while maintaining simplicity for regions without diagonal edges, thus improving overall compression efficiency for 360-degree video formats
Data Source
AI summary
The present disclosure describes techniques that can be considered in intra-predicting diagonal edges that may be present in a block to be encoded. In the new intra mode proposed in the present disclosure, a prediction area is divided into two areas in the diagonal direction, and the areas can be predicted in different directions.