AMVP-Merge Motion Prediction with SbTMVP for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in predicting motion vectors, particularly in inter-predicted blocks, leading to suboptimal compression and decoding performance.
Innovation Solution
Implementing a subblock-based temporal motion vector prediction (SbTMVP) mode that utilizes a bi-directional predictor and merge predictor to enhance motion vector prediction accuracy, specifically using a displacement vector to derive motion information from collocated blocks in a reference picture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subblock-based temporal motion vector prediction (SbTMVP) mode is implemented to predict motion vectors at subblock level, then motion vector prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The current block is divided into multiple subblocks, and motion vectors are predicted separately for each subblock using SbTMVP. This segmentation allows for more precise motion compensation by capturing local motion variations within the block, thereby improving prediction accuracy without requiring a complete redesign of the prediction framework.
Solution Approach 2:
The patent performs preliminary construction of candidate lists (AMVP and merge candidate lists) before final motion vector prediction. By pre-processing and organizing potential motion vector candidates in advance, the system reduces the computational burden during actual prediction, balancing accuracy improvement with complexity management.
2Measurement precision
If bi-directional predictor with AMVP and merge predictors is used, then prediction accuracy is improved, but data transmission overhead increases
Solution Approach 1:
The patent combines AMVP predictor and merge predictor into a unified bi-directional prediction framework. By merging these two prediction methods, the system leverages the strengths of both approaches to improve prediction accuracy while managing the bitstream overhead through efficient candidate selection and signaling mechanisms.
Solution Approach 2:
Different prediction methods (AMVP and merge) are applied to different directions or subblocks based on local characteristics. This allows the system to adaptively select the most appropriate prediction method for each region, improving overall prediction accuracy while minimizing the overhead associated with transmitting multiple full candidate lists.
3Reliability
If multiple candidate lists (AMVP and merge) are constructed, then prediction reliability is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary construction of both AMVP and merge candidate lists during the encoding/decoding process. By preparing these candidate lists in advance before final prediction, the system ensures reliable prediction outcomes while optimizing processing time through efficient organization and early elimination of unlikely candidates.
Solution Approach 2:
The system uses feedback mechanisms to evaluate and select the best prediction candidates from the constructed lists. By implementing selection criteria and evaluation metrics, the system can reliably choose optimal motion vectors while avoiding exhaustive search through all candidates, thus balancing reliability with processing efficiency.
Data Source
AI summary
Processing circuitry receives a coded video bitstream comprising coded information of a current block in a current picture, the coded information is indicative of an AMVP-merge mode, the AMVP-merge mode uses a bi-directional predictor including an AMVP predictor in a first direction and a merge predictor in a second direction. The processing circuitry constructs an AMVP candidate list that includes a plurality of AMVP candidates for the AMVP predictor and constructs a merge candidate list that includes a plurality of merge candidates for the merge predictor. The merge candidate list includes an SbTMVP candidate. The processing circuitry determines the AMVP predictor from the AMVP candidate list, determines the merge predictor from the merge candidate list, and reconstructs the current block according to a combination of a first prediction according to the AMVP predictor in the first direction and a second prediction according to the merge predictor in the second direction.