Video Decoder Motion Vector Prediction for Sub-Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in reducing redundancy in motion vector coding, particularly in efficiently predicting motion vectors for sub-blocks within a current block, which affects compression efficiency and bitrate requirements.
Innovation Solution
The proposed solution involves a method for video decoding that selects motion vectors from neighboring sub-blocks and uses motion vector predictors to derive motion vectors for sub-blocks within a current block, employing techniques like template matching and bilateral matching, and combining predictors to minimize the need for explicit coding of motion vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vectors for each sub-block are explicitly coded, then motion vector precision is improved, but bitrate increases
Solution Approach 1:
The patent uses motion vectors from neighboring blocks as predictors for current block sub-blocks. Instead of coding each motion vector independently, the decoder copies motion information from spatially adjacent blocks and temporally referenced blocks, significantly reducing the number of bits required while maintaining acceptable motion compensation accuracy.
Solution Approach 2:
The patent introduces motion vector predictors as an intermediary between reference block motion vectors and current block motion vectors. These predictors (such as merge candidates involving neighboring block motion vectors and temporal motion vectors) serve as intermediate representations that reduce the information needed to fully specify motion vectors while preserving essential motion information.
2Quantity of substance
If motion vector prediction from neighboring blocks is used, then bitrate is reduced, but motion vector accuracy may deteriorate
Solution Approach 1:
The patent combines multiple motion vector predictors (spatial neighboring block motion vectors, temporal motion vectors, and gradient-based predictors) into a merged prediction mechanism. The decoder can select from multiple candidate predictors or combine them using weighting factors, achieving better accuracy than any single predictor alone while maintaining compression efficiency.
Solution Approach 2:
The patent implements dynamic motion vector prediction where the prediction method and parameters are adapted based on local image characteristics. Gradient calculations and motion compensation parameters are dynamically adjusted according to the specific block content and motion patterns, allowing the system to optimize between compression and accuracy on a per-block basis.
Data Source
AI summary
Aspects of the disclosure provide methods and a decoder device for video decoding. In some embodiments, the decoder device for video decoding includes processing circuitry. The processing circuitry selects a motion vector of a neighboring sub-block that is a neighbor of a current block under reconstruction. The current block is in a coded picture that is a part of a coded video bitstream. The processing circuitry determines, from the motion vector, motion vector predictors for a plurality of sub-blocks included in the current block. The neighboring sub-block and the plurality of sub-blocks are in a first one of a single row and a single column of the coded picture.