Inter Prediction Merge Lists for Parallel Motion Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in reducing complexity and improving encoding/decoding efficiency, particularly in high-resolution and high-quality image data transmission and storage.
Innovation Solution
An inter prediction method that generates a merging candidate list for video coding, selectively using first and second merging candidate lists based on block location and size, allowing parallel processing and sharing of motion information within a merging candidate sharing unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a single merging candidate list is used for all prediction units, then the encoding process is simple, but the encoding efficiency is limited
Solution Approach 1:
The patent divides the merging candidate list generation into two separate lists: a first merging candidate list for prediction units not belonging to a merging candidate sharing unit, and a second merging candidate list for prediction units belonging to a merging candidate sharing unit. This segmentation allows different processing strategies for different types of prediction units, improving overall encoding efficiency while managing complexity through structured division.
Solution Approach 2:
The patent dynamically selects which merging candidate list to use based on whether the current prediction unit belongs to a merging candidate sharing unit. This dynamic adaptation allows the encoding process to switch between different candidate list generation modes, optimizing performance for different block types and locations without requiring a completely complex system design.
2Device complexity
If motion information is derived independently for each prediction unit, then the accuracy is high, but the computational complexity increases
Solution Approach 1:
The patent merges motion information from multiple sources including spatial neighboring blocks, temporal reference blocks, and merged candidates into a unified second merging candidate list. This combining approach allows prediction units within a sharing unit to benefit from aggregated motion information, reducing computational complexity while maintaining accuracy through diversified candidate sources.
Solution Approach 2:
The patent copies motion information from reference blocks and neighboring blocks into the merging candidate lists. By replicating proven motion vectors from similar blocks, the system reduces the need for complex independent motion estimation for each prediction unit, thereby lowering computational complexity while preserving accuracy through reliable copied data.
3Reliability
If high-resolution image data is transmitted without compression, then the image quality is maintained, but the transmission cost and storage cost increase
Solution Approach 1:
The patent changes the parameter representation by using merging candidate lists to encode motion information more efficiently. Instead of transmitting full motion vectors for every prediction unit, the system uses indices to reference pre-computed merging candidates, significantly reducing the bit rate required for transmission while maintaining image quality through accurate motion compensation.
Solution Approach 2:
The patent discards redundant motion information by using merging candidates that are shared across multiple prediction units. Rather than transmitting identical motion data for each block, the system transmits a reference index that points to the shared candidate, effectively discarding redundant data during transmission and recovering it at the decoder side through the reference mechanism.
Data Source
AI summary
An inter predication method is provided. The method includes: generating a merging candidate list of a block to be predicted and corresponding to a current prediction unit (PU); deriving motion information of the block to be predicted, on the basis of one of a plurality of merging candidates constituting the merging candidate list; and performing a prediction on the block to be predicted, on the basis of the derived motion information, to generate a prediction block corresponding to the current PU.