Intra Prediction Mode Coding for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing HEVC standard's intra prediction mode coding is inefficient due to limited prediction angles and complexity, and the current mode coding does not accurately correspond to the actual probability distribution of available modes, leading to suboptimal compression efficiency.
Innovation Solution
The proposed solution involves selecting a subset of 33 angular intra prediction modes from a larger set, using context modeling to derive more than three Most Probable Modes (MPMs), and applying an N-tap interpolation filter with 1/32 pel accuracy for more accurate interpolation, allowing for increased prediction accuracy without increasing encoder complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a subset of 33 angular intra prediction modes is selected from 65 modes, then encoder complexity is reduced, but prediction accuracy may be compromised
Solution Approach 1:
The 65 angular intra prediction modes are segmented into two groups: 33 modes for standard encoding (reducing complexity) and the remaining 32 modes for enhanced prediction accuracy when needed. This segmentation allows the system to balance between computational load and prediction quality by selectively applying different mode sets based on content characteristics.
Solution Approach 2:
The patent changes the parameter of prediction mode quantity from 65 to 33 in the selected subset, while compensating for the reduction through improved interpolation accuracy (1/32 pel) and context modeling techniques. This parameter change resolves the contradiction by maintaining effective prediction capability with fewer modes through enhanced processing of the selected modes.
2Manufacturing precision
If 1/32 pel accuracy interpolation is applied, then prediction precision is improved, but computational complexity increases
Solution Approach 1:
Instead of applying 1/32 pel accuracy to all prediction modes and all blocks, the patent applies this high-precision interpolation selectively to the 33 selected angular modes and specific prediction units where it provides the most benefit. This partial application achieves the precision improvement while limiting the increase in computational complexity to only where necessary.
Solution Approach 2:
The patent applies different interpolation accuracies to different regions and prediction units based on their specific characteristics. High-precision 1/32 pel interpolation is applied locally to angular prediction modes where it most improves accuracy, while other areas may use standard interpolation. This local quality approach optimizes the balance between precision and complexity.
3Productivity
If context modeling with more than three Most Probable Modes is used, then compression efficiency is improved, but mode selection complexity increases
Solution Approach 1:
The patent performs preliminary analysis of neighboring blocks to pre-determine the Most Probable Modes (MPMs) before encoding the current block. By deriving more than three MPMs from contextual information of neighboring blocks in advance, the system prepares a prioritized list of candidate modes, reducing the complexity of mode selection during actual encoding while improving compression efficiency through better mode prediction.
Solution Approach 2:
The context modeling mechanism uses feedback from neighboring block encoding results to dynamically adjust and refine the Most Probable Modes list. The encoder analyzes the intra prediction modes used in neighboring blocks and uses this feedback information to construct an optimized MPM list for the current block, improving compression efficiency while managing complexity through intelligent reuse of contextual patterns.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A device for decoding video data includes one or more processors configured to decode syntax information that indicates a selected intra prediction mode for a block of video data from among a plurality of intra prediction modes. The plurality of intra prediction modes includes greater than 33 angular intra prediction modes. The angular intra prediction modes defined such that interpolation is performed in 1/32 pel accuracy. The one or more processors reconstruct the block of video data based on the selected intra prediction mode.