Wide Angle Intra Prediction for Non-Square Video Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional video coding technologies face inefficiencies in intra prediction for non-square blocks, as existing methods are optimized for square blocks and do not effectively utilize prediction directions that are less likely for non-square shapes, leading to suboptimal coding efficiency.
Innovation Solution
The proposed solution involves determining a range of reference samples based on the shape of a non-square block and using wide-angle intra prediction modes to adaptively replace conventional intra prediction modes, allowing for more effective intra prediction by considering the block's aspect ratio and optimizing the reference sample range for non-square blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional intra prediction modes optimized for square blocks are used for non-square blocks, then the device complexity remains low, but the coding efficiency deteriorates
Solution Approach 1:
The patent applies local quality by adapting the intra prediction mode selection to the specific geometry of each block. Different prediction modes are selected based on whether the block is square or non-square, and whether it is located at picture boundaries. This localized adaptation optimizes prediction accuracy for each block's specific characteristics rather than using a uniform approach for all blocks.
Solution Approach 2:
The patent introduces dynamic mode selection where the intra prediction mode is adapted based on block shape and position. The system dynamically chooses between conventional modes and wide-angle modes depending on the block's aspect ratio and location, allowing the prediction strategy to flexibly respond to different coding scenarios rather than being static.
2Productivity
If wide-angle intra prediction modes are added for non-square blocks, then the coding efficiency improves, but the device complexity increases
Solution Approach 1:
The patent changes the parameters of the prediction system by introducing wide-angle modes with different reference sample ranges and filtering characteristics. These parameter changes allow the system to better match the geometric characteristics of non-square blocks, improving prediction accuracy and subsequent entropy coding efficiency.
Solution Approach 2:
The patent segments the intra prediction space into different mode categories (conventional modes for square blocks, wide-angle modes for non-square blocks). This segmentation allows each category to be optimized independently for its specific use case, improving overall efficiency without requiring the system to handle all modes equally complexly.
3Measurement precision
If the reference sample range is extended for wide-angle prediction, then the prediction accuracy improves, but the bandwidth requirements increase
Solution Approach 1:
The patent applies partial action by extending the reference sample range only when necessary for wide-angle prediction modes, rather than always using the maximum range. The reference sample padding and filtering are applied selectively based on the prediction mode and block characteristics, achieving improved accuracy only where needed while minimizing the associated bitrate cost.
Data Source
AI summary
Aspects of the disclosure provide an apparatus for video coding. The apparatus includes circuitry configured to reconstruct one or more blocks neighboring a first block that is non-square, and determine a range of reference samples required for an intra prediction of the first block based on a shape of the first block. The range includes samples having coordinates of x=[−1], y=[−1 to (dimension1*2+m)], and x=[0 to (dimension2*2+n)], y=[−1]. Dimension1 denotes a length of a first side of the first block, dimension2 denotes a length of a second side of the first block that is perpendicular to the first side of the first block, and m and n are integers greater or equal to 0. The circuitry is further configured to perform the intra prediction based on the reference samples in the determined range.