Intra-Prediction Filtering for Acute-Angle Video Block Continuity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as HEVC/H.265, face issues with discontinuities in intra-prediction at acute angles due to the asymmetry of rectangular blocks, leading to inefficient use of reference samples and increased computational complexity.
Innovation Solution
Adaptive angular steps are introduced between directional intra-prediction modes based on filter length, reducing the number of modes in subsets to minimize discontinuities and maintain coding efficiency without increasing hardware or computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of directional intra-prediction modes is increased to improve prediction accuracy, then coding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by differentiating the angular step sizes based on the specific prediction mode being processed. Different directional modes use different angular steps (e.g., 2.25° for certain modes, 4.5° for others), allowing optimized prediction accuracy for each local region of the angular space without uniformly increasing complexity across all modes.
Solution Approach 2:
The patent introduces dynamic adaptation by making the angular step variable rather than fixed. The angular step is dynamically selected based on the prediction mode index and filter length, allowing the system to adapt the prediction granularity to the specific requirements of different directional modes.
2Device complexity
If fixed angular steps are used between directional intra-prediction modes, then device complexity is reduced, but discontinuities occur at acute angles
Solution Approach 1:
The patent resolves the discontinuity issue by applying different angular steps to different prediction modes. Acute angle modes use smaller angular steps to maintain continuity, while other modes use larger steps, creating local optimization for each region of the angular spectrum.
Solution Approach 2:
The patent changes the parameter of angular step size based on the prediction mode characteristics. By varying this key parameter across different modes rather than using a fixed value, the system eliminates discontinuities at acute angles while maintaining manageable complexity.
3Productivity
If the number of reference samples is increased to improve prediction accuracy, then coding efficiency is improved, but computational complexity increases
Solution Approach 1:
The patent changes the parameter of angular step size which indirectly controls the number and distribution of reference samples. By using variable angular steps, the system optimizes the reference sample utilization for each prediction mode, achieving better accuracy without uniformly increasing computational complexity.
Solution Approach 2:
The patent applies partial action by not uniformly increasing the number of reference samples across all modes. Instead, it selectively adjusts the angular step to optimize reference sample usage only where needed (e.g., for acute angle modes), avoiding excessive computational complexity in modes where larger steps are already used.
4Productivity
If rectangular blocks are used for video coding, then coding efficiency is improved, but asymmetry causes discontinuities in intra-prediction
Solution Approach 1:
The patent embraces the asymmetry of rectangular blocks by using different angular steps for different prediction modes. This asymmetric approach in mode configuration compensates for the geometric asymmetry of rectangular blocks, eliminating discontinuities while preserving the coding efficiency benefits of rectangular partitioning.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to the field of picture processing, for example still picture/image and/or video picture/image coding. In particular, the invention relates to a device and corresponding method for intra-predicting a prediction block of a video image. The device is configured to select a directional intra-prediction mode from a set of directional intra-prediction modes, wherein each directional intra-prediction mode corresponds to a different intra- prediction angle. Further, the device is configured to select a filter from a set of filters based on the selected directional intra-prediction mode. Further, the device is configured to determine, for a given prediction sample of the prediction block, a reference sample from a set of reference samples based on the selected directional intra-prediction mode, and apply the selected filter to the determined reference sample.