MPM Reordering for Intra Prediction in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding methods face inefficiencies in determining the most probable intra prediction mode for blocks, which affects compression efficiency and decoding accuracy.
Innovation Solution
The method involves accessing neighboring blocks to determine a most probable mode (MPM) set for a given block, including intra prediction modes from adjacent blocks, and encoding or decoding an index corresponding to the determined intra prediction mode within this set.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional intra prediction mode determination methods are used, then the encoding process is simpler, but compression efficiency and decoding accuracy deteriorate
Solution Approach 1:
The patent applies preliminary action by determining the MPM set before encoding the intra prediction mode. The encoder and decoder both pre-establish the MPM set using the same rules based on neighboring block modes, so that when the mode index is transmitted, the decoder already has the correct mapping context. This preliminary setup enables efficient compression without requiring complex real-time calculations during decoding.
Solution Approach 2:
The MPM set acts as an intermediary between the raw intra prediction mode values and the compressed bitstream. Instead of directly encoding the actual prediction mode, the system encodes an index into the MPM set, which the decoder then maps back to the actual mode using its own constructed MPM set. This intermediary mechanism enables lossless compression while maintaining decoding accuracy.
2Measurement precision
If more neighboring blocks are considered for MPM set, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent applies local quality by considering the spatial locality of neighboring blocks. The MPM set is constructed using only the intra prediction modes of immediately adjacent blocks (above and left neighbors), which are locally correlated with the current block. This localized approach captures the essential spatial redundancy without the complexity of analyzing distant or global block patterns.
Solution Approach 2:
The patent uses partial action by selecting a limited subset of neighboring blocks (specifically the above and left neighbors) to construct the MPM set, rather than analyzing all possible neighboring blocks. This partial consideration of neighbors provides sufficient prediction accuracy while avoiding the excessive computational complexity that would result from examining every surrounding block.
3Reliability
If MPM set is determined using multiple neighbor blocks, then intra prediction mode selection improves, but encoding overhead increases
Solution Approach 1:
The patent applies copying by having the decoder reconstruct the same MPM set that the encoder used, using identical rules based on the decoded neighboring block modes. This copying of the MPM set structure ensures that both encoder and decoder operate with the same reference framework, enabling accurate mode reconstruction without transmitting the full MPM set data, thus minimizing bitstream overhead.
4Productivity
If conventional encoding methods are used, then processing speed is faster, but compression efficiency deteriorates
Solution Approach 1:
By pre-determining the MPM set using simple rules based on neighboring block modes, the system avoids complex real-time optimization during encoding and decoding. The preliminary construction of the MPM set with fixed rules enables both encoder and decoder to work efficiently with minimal computational overhead, maintaining high processing speed while achieving improved compression through better mode selection.
Data Source
AI summary
To construct an MPM set for encoding or decoding an intra prediction mode, a plurality of neighbor blocks may be used. In one implementations, two above-intermediate neighbor blocks, such as those specified by sample locations at a distance equal to the height or half of the width may be used for a flat rectangular block, or two left-intermediate neighbor blocks, such as those specified by sample locations at a distance equal to the width or half of the height may be used for a tall rectangular block. The order of searching the neighbor blocks can be based on the current block's shape, or based on the distance between a neighbor's intra prediction mode and a vertical or horizontal intra prediction mode. Both the encoder and decoder follow the same MPM set construction process. Hence, no additional syntax elements are needed to signal the neighbor blocks and the search order.