Adaptive Intra Prediction Modes for Lower Video Coding Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards, such as HEVC and its extensions, face inefficiencies in intra prediction mode signaling, particularly in JEM, due to increased signaling overhead and lack of consideration for block shape and edge features, which can degrade coding performance.
Innovation Solution
Implement a variable number of intra modes based on block size and other factors, using look-up tables or closed-form expressions to convert between different sets of intra modes, and adapt intra mode signaling to optimize coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed large number of intra prediction modes are used for all blocks, then prediction accuracy is improved, but signaling overhead increases
Solution Approach 1:
The patent applies local quality by adapting the number of intra prediction modes to local block characteristics. Different block sizes and shapes receive different numbers of prediction modes from the look-up table, allowing each block to use an optimal number of modes tailored to its specific geometry rather than using a uniform large set for all blocks.
Solution Approach 2:
The patent changes the parameter of the number of intra prediction modes dynamically based on block characteristics. By using a look-up table that maps block size and shape to specific mode counts, the system adjusts the mode parameter adaptively, reducing the number of modes for blocks where fewer are sufficient and maintaining more modes for blocks that benefit from additional prediction directions.
2Loss of information
If intra mode signaling is simplified, then signaling overhead is reduced, but coding performance deteriorates
Solution Approach 1:
The patent applies local quality by tailoring the intra mode signaling to local block characteristics. The look-up table provides different numbers of most probable mode candidates based on block size and shape, ensuring that each block receives an appropriate level of signaling detail that matches its geometric properties and prediction needs.
Solution Approach 2:
The patent introduces dynamics by making the number of intra prediction modes variable rather than fixed. The system dynamically selects the number of modes from the look-up table based on the specific block being coded, allowing the signaling strategy to adapt flexibly to different block configurations while maintaining optimal coding performance.
3Loss of information
If intra prediction modes are restricted for certain asymmetric prediction units, then signaling overhead is reduced, but adaptability to different block shapes decreases
Solution Approach 1:
The patent applies local quality by providing different numbers of intra prediction modes for different block shapes through the look-up table. Rather than applying a uniform restriction to all asymmetric blocks, the system consults the look-up table to determine the appropriate number of modes based on the specific block's dimensions and aspect ratio, allowing customized adaptation to each block shape.
Solution Approach 2:
The patent introduces dynamics by making the intra mode selection adaptive to block shape characteristics. The look-up table enables the system to dynamically adjust the number of prediction modes based on the specific asymmetric prediction unit being coded, providing versatility across different block shapes while still reducing overhead compared to using all modes for every block.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A video coder determines a first block of the video data is intra mode coded; based on a first height and the first width of the first block, identifies a group of N available intra prediction modes for the first block of video data; selects from the group of N available intra prediction modes, a first intra prediction mode used to code the first block of the video data; and codes the first block using the first intra prediction mode. A video coder generates a first most probable mode (MPM) candidate list for the block; codes a first flag indicating an actual intra prediction mode used to code the block is not included in the first MPM candidate list; generates a second MPM candidate list by deriving at least one candidate intra prediction mode based on an intra prediction mode in the first MPM candidate list.