Temporal Motion Vector Prediction via Spatial Neighbor Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques, such as H.265/HEVC, face challenges in accurately predicting temporal motion vectors due to differences in moving objects between collocated blocks in different pictures, leading to suboptimal compression efficiency.
Innovation Solution
The method involves selecting a candidate block from neighboring blocks in a previously decoded picture, based on a motion vector predictor, to perform temporal motion vector prediction for the current block, allowing for more accurate prediction even when collocated blocks do not contain the same moving object.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If temporal motion vector prediction is performed using collocated blocks from previously decoded pictures, then compression efficiency is improved, but prediction accuracy deteriorates when collocated blocks do not contain the same moving object
Solution Approach 1:
The patent introduces an intermediary selection process that uses spatial neighboring blocks as mediators to identify the most suitable candidate block from the previously decoded picture. Instead of directly using the collocated block, the method examines motion vectors of spatial neighbors to determine which candidate block best matches the current block's motion characteristics, thereby improving prediction accuracy while maintaining compression efficiency
Solution Approach 2:
The patent changes the selection criterion from fixed collocated block position to dynamic candidate block selection based on motion vector analysis. By evaluating motion vectors of spatial neighboring blocks and selecting the candidate block with the most compatible motion characteristics, the method adapts the prediction process to actual motion patterns, resolving the contradiction between compression efficiency and prediction accuracy
2Measurement precision
If motion vectors are coded directly without prediction, then prediction accuracy is maintained, but bitrate increases significantly
Solution Approach 1:
The patent implements a feedback mechanism where motion vectors from spatial neighboring blocks are used to inform the selection of temporal motion vector candidates. This feedback loop allows the system to adaptively choose the best candidate based on local motion patterns, achieving both accurate motion vector prediction and efficient bitrate utilization through entropy coding of the selected candidate index
Data Source
AI summary
A method of video decoding for a decoder includes acquiring a current picture from a coded video bitstream. The method further includes selecting a first neighboring block from a plurality of neighboring blocks of a current block in the current picture. The method further includes determining, based on a motion vector predictor of the selected first neighboring block, a candidate block that is included in a previously decoded picture and that is one of (i) neighboring to a collocated block of the current block and (ii) within the collocated block. The method further includes performing temporal motion vector prediction for the current block based on a motion vector predictor of the candidate block.