Motion Vector Coding Using Long-Term HMVP Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face challenges in efficiently utilizing input motion vector data for motion compensation, particularly in handling long-term motion vector predictors, which can lead to inefficiencies in compression and decoding processes.
Innovation Solution
The implementation of a system that includes a processor to obtain and add long-term motion vector predictor candidates to a history-based motion vector prediction list, enabling their use in video encoding and decoding, and allowing for exemption from certain pruning rules, thereby enhancing the utilization of input motion vector data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If long-term motion vector predictor candidates are added to the HMVP list, then motion compensation efficiency is improved, but device complexity increases
Solution Approach 1:
The patent pre-computes and stores long-term motion vector predictor candidates from previously decoded pictures before they are needed for current block decoding. This preliminary action populates the HMVP list in advance with high-quality candidates, improving motion compensation efficiency without adding computational complexity during the actual decoding process.
Solution Approach 2:
The HMVP list is designed to serve multiple purposes: it stores both short-term and long-term motion vector predictors, and can be used for different block types and prediction modes. This multi-functionality allows a single data structure to handle diverse motion compensation scenarios, improving efficiency without proportionally increasing complexity.
2Productivity
If long-term motion vector predictors are utilized, then compression performance is enhanced, but loss of information increases
Solution Approach 1:
The patent changes the temporal parameter of motion vector predictors by introducing long-term candidates from pictures far in the past, rather than only using recent short-term candidates. This parameter change allows better matching for certain video content patterns, improving compression performance while the redundancy check prevents information loss by eliminating duplicate candidates.
Solution Approach 2:
The patent implements a redundancy check mechanism that identifies and discards duplicate motion vector candidates in the HMVP list before they are used for prediction. This prevents information loss by ensuring only unique, valuable candidates are stored, while the ability to recover and reuse these candidates across different blocks improves overall compression performance.
3Adaptability or versatility
If long-term motion vector predictor candidates are obtained from video data at multiple levels, then adaptability is improved, but processing time increases
Solution Approach 1:
The patent segments the video data into hierarchical levels (picture level, slice level, block level) and obtains long-term motion vector candidates at each level independently. This segmentation allows the system to adapt to different video content characteristics at appropriate granularities while distributing the processing workload, preventing any single stage from becoming a time-consuming bottleneck.
Solution Approach 2:
The patent performs preliminary processing to identify and store potential long-term motion vector candidates during the initial parsing of video data at different hierarchical levels. This preliminary action prepares candidate pools in advance, reducing the processing time required during actual motion compensation operations by having pre-organized candidate data ready for selection.
Data Source
AI summary
Systems, methods, and instrumentalities are disclosed for performing motion compensation with long-term history-based motion vector predictor candidate(s). A device configured for performing such motion compensation may include a processor. The processor may be configured to obtain a long-term motion vector predictor candidate. The processor may also be configured to add the long-term motion vector predictor candidate to a history-based motion vector prediction (HMVP) list for a current block. The processor may also be configured to decode the current block using the HMVP list having the long-term motion vector prediction candidate.