Shared Candidate List for Parallel Video Inter-Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression techniques face challenges in efficiently handling high-resolution, high-quality video data, leading to increased costs for transmission and storage due to the large amount of data involved, particularly in inter-prediction processes where complexity arises from generating multiple merging candidate lists for prediction blocks.
Innovation Solution
A method and apparatus for performing inter-prediction in a parallel fashion by generating a single candidate list for prediction blocks, which includes deriving spatial and temporal merging candidates based on pixel position and size, and determining whether parallel merge processing can be applied to reduce complexity and improve processing speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple merging candidate lists are generated for each prediction block in inter-prediction, then prediction accuracy is improved, but processing complexity and time increase significantly
Solution Approach 1:
The patent combines multiple merging candidate lists into a single shared candidate list that is reused across multiple prediction blocks. Instead of generating separate candidate lists for each prediction block, the system creates one consolidated list containing merging candidates from spatial and temporal references, which is then shared among all prediction blocks in the parallel processing architecture. This reduces the total number of candidate lists generated while maintaining prediction accuracy through the use of parallel processing.
Solution Approach 2:
The patent segments the processing into two distinct phases: first, generating a single merging candidate list in parallel for multiple prediction blocks simultaneously; second, selecting the appropriate merging candidate from this shared list for each prediction block. This segmentation allows the computationally intensive candidate generation to be performed once in parallel, rather than repeatedly for each block, thereby reducing overall processing complexity.
2Measurement precision
If multiple merging candidate lists are generated for each prediction block, then prediction quality is enhanced, but encoding and decoding time increase
Solution Approach 1:
The patent performs preliminary action by generating the merging candidate list in advance through parallel processing before the actual prediction blocks are processed. The single shared candidate list is prepared beforehand and made available for all prediction blocks, eliminating the need to regenerate candidate lists for each block. This preliminary preparation significantly reduces encoding and decoding time while maintaining prediction quality.
Solution Approach 2:
By merging the candidate generation process across multiple prediction blocks into a single parallel operation, the system generates one consolidated candidate list that serves all blocks. This combining approach reduces the total processing time compared to generating separate candidate lists sequentially for each prediction block, while still providing high-quality merging candidates for accurate prediction.
3Productivity
If a single candidate list is shared among multiple prediction blocks, then processing speed improves, but candidate selection complexity increases
Solution Approach 1:
The patent applies local quality by allowing each prediction block to select merging candidates based on its specific characteristics and requirements from the shared candidate list. While the candidate list is shared globally, the selection process is localized to each prediction block's context, enabling efficient selection without requiring complex global optimization. Each block evaluates candidates from the shared list according to its local prediction needs.
Data Source
AI summary
The present invention relates to a method and device for sharing a candidate list. A method of generating a merging candidate list for a predictive block may include: producing, on the basis of a coding block including a predictive block on which a parallel merging process is performed, at least one of a spatial merging candidate and a temporal merging candidate of the predictive block; and generating a single merging candidate list for the coding block on the basis of the produced merging candidate. Thus, it is possible to increase processing speeds for coding and decoding by performing inter-picture prediction in parallel on a plurality of predictive blocks.