Spatiotemporal Motion Vector Prediction Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding technologies, such as HEVC, face challenges in optimizing motion vector prediction to enhance compression efficiency and reduce complexity, particularly in constructing spatiotemporal motion vector predictors for improved coding performance.
Innovation Solution
The method involves obtaining spatiotemporal motion vector prediction candidates by combining spatial and temporal motion vectors from neighboring blocks, modifying values based on these candidates, and including them in an ordered set for encoding or decoding, which reduces the number of tests required and improves coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If traditional motion vector prediction methods are used, then coding performance is maintained, but compression efficiency is insufficient and complexity is high
Solution Approach 1:
The current block is divided into sub-blocks, and motion vector prediction is performed separately for each sub-block. This segmentation allows the use of multiple spatial locations (top-left, top-right, bottom-left, bottom-right) to derive prediction candidates, improving compression efficiency by better capturing local motion variations while keeping each sub-block's prediction process simple
Solution Approach 2:
The patent extends traditional 2D spatial motion vector prediction by incorporating temporal dimension through reference pictures. Spatiotemporal motion vector candidates are derived by combining spatial neighbors with temporal information from reference frames, adding a time dimension to the prediction process and improving compression efficiency without significantly increasing complexity
2Loss of energy
If more spatiotemporal motion vector prediction candidates are constructed, then compression efficiency improves, but computational overhead increases
Solution Approach 1:
The patent derives motion vector prediction candidates for all four corners of sub-blocks (top-left, top-right, bottom-left, bottom-right), which is more than traditional methods that typically use only one or two locations. This excessive action in deriving candidates from multiple spatial positions improves compression efficiency by providing better prediction accuracy, while the systematic approach keeps computational overhead manageable
Solution Approach 2:
Motion vector prediction candidates are derived in advance during the encoding process and stored in a candidate list before final selection. This preliminary action allows the decoder to reconstruct the same candidates without additional computation, reducing bitrate while the systematic derivation process keeps computational overhead controlled
Data Source
AI summary
In general, encoding or decoding a picture part can involve determining a spatiotemporal motion vector predictor (STMVP) candidate for a merge mode of operation from spatial motion vector candidates taken from spatial positions around a current coding unit (CU) and a temporal motion vector candidate, where at least one embodiment involves determining the STMVP candidate based on considering at most two spatial positions and based on an average of at least two of the spatial and temporal motion vector candidates.