Adaptive MPM Lists for Multiple-Reference-Line Intra Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods, such as those in HEVC, limit the accuracy of multiple reference line (MRL) intra prediction by restricting the number of Most Probable Modes (MPMs) in the MPM list, which affects the efficiency of intra prediction in video coding.
Innovation Solution
The solution involves separating the MPM list for non-line 0 most probable modes (Non-LOMPMs) specific to MRL intra prediction from the common MPM list for the closest reference line 0, and deriving Non-LOMPMs based on neighboring block prediction directions to enhance the MPM list for MRL intra prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the MPM list for MRL intra prediction is restricted to a small number of modes, then the coding complexity is reduced, but the prediction accuracy deteriorates
Solution Approach 1:
The patent divides the MPM list into two separate lists: one for reference line 0 and another for non-zero reference lines. This segmentation allows each list to be optimized independently, with the non-zero reference line list containing more modes specific to MRL prediction, thereby improving prediction accuracy without excessively increasing overall coding complexity.
Solution Approach 2:
The patent extends the MPM list in the dimensional sense by adding more prediction modes to the non-zero reference line list. Instead of keeping a single limited list, it creates an expanded list that includes additional angular modes specifically for MRL intra prediction, thus improving accuracy by adding more options in the mode selection space.
2Measurement precision
If the MPM list is expanded to include more modes, then the prediction accuracy is improved, but the bit rate increases
Solution Approach 1:
By segmenting the MPM list into reference line 0 specific modes and non-zero reference line specific modes, the patent enables more efficient entropy coding. The decoder can selectively use modes from the appropriate list based on the reference line index, reducing the search space and allowing for more compact bit rate representation of the expanded mode set.
Solution Approach 2:
The patent applies different MPM list configurations to different reference lines. Reference line 0 uses one set of modes while non-zero reference lines use an expanded set. This local differentiation allows the system to provide enhanced prediction accuracy where needed (for MRL modes) while maintaining efficient coding for standard cases.
3Device complexity
If the same MPM list is used for all reference lines, then the device complexity is reduced, but the adaptability to different reference lines deteriorates
Solution Approach 1:
The patent segments the MPM list into reference line-specific subsets, creating dedicated mode lists for reference line 0 and non-zero reference lines. This segmentation enables the system to adapt to the specific characteristics of each reference line type while maintaining manageable complexity through structured organization.
Solution Approach 2:
The patent introduces dynamic adaptability by selecting different MPM lists based on the reference line index. The system dynamically switches between the reference line 0 MPM list and the non-zero reference line MPM list, allowing it to adapt to different reference line characteristics without requiring a completely different system for each case.
Data Source
Figure 1A~1B
Figure 1C
Figure 1D
AI summary
A method of controlling intra prediction for decoding or encoding of a video sequence, is performed by at least one processor and includes identifying whether a first intra prediction mode of an above neighboring block of a prediction block is different than a second intra prediction mode of a left neighboring block of the prediction block. The method further includes, based on the first intra prediction mode being identified to be not different than the second intra prediction mode, obtaining a most probable mode (MPM) list including the first intra prediction mode, first immediate modes immediate to the first intra prediction mode and one among second immediate modes respectively immediate to the first immediate modes, and performing the intra prediction on multiple reference lines of the prediction block, using the obtained MPM list, the multiple reference lines being without a first reference line closest to the prediction block.