Motion Vector Predictor Selection for Higher Video Coding Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing MPEG-4AVC/H.264 standard for moving picture compression coding faces inefficiencies in motion vector prediction, leading to poor prediction precision and coding efficiency due to the limited use of neighboring block motion vectors.
Innovation Solution
A moving picture coding and decoding technology that generates multiple motion vector predictor candidates by predicting from neighboring blocks within the same or different pictures, prioritizing candidates based on reference list and picture compatibility, and selects the optimal predictor to reduce code size and improve coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vector prediction is performed using limited neighboring blocks in MPEG-4AVC/H.264, then device complexity is reduced, but prediction precision deteriorates
Solution Approach 1:
The patent segments the motion vector prediction process into multiple stages: generating multiple candidate motion vectors from different neighboring blocks (A, B, C, D), evaluating each candidate's prediction precision, and selectively using the best candidate. This segmentation allows the system to achieve high prediction precision through multiple candidates while managing device complexity by using systematic evaluation criteria and early termination when optimal candidates are found.
2Measurement precision
If multiple motion vector predictors are derived from neighboring blocks, then prediction precision is improved, but code size increases
Solution Approach 1:
The patent applies partial action by deriving motion vector predictors from only the most relevant neighboring blocks (A, B, C, D) rather than all possible blocks. The system generates multiple candidates but selectively processes only those that meet specific criteria, avoiding the excessive computation and code size that would result from considering all neighboring blocks. This partial approach achieves good prediction precision while controlling code size.
Solution Approach 2:
The patent changes parameters by introducing multiple motion vector candidates (mvA, mvB, mvC, mvD) with different derivation methods and evaluation criteria. The system dynamically selects which candidates to use based on prediction precision requirements, transforming the fixed single-predictor approach into a flexible multi-parameter system that adapts to different coding scenarios, thereby improving precision without proportionally increasing code size.
3Productivity
If motion vectors are coded without prediction from neighboring blocks, then coding complexity is reduced, but coding efficiency deteriorates
Solution Approach 1:
The patent implements feedback by evaluating the prediction precision of multiple motion vector candidates and using this evaluation to guide the selection process. The system calculates prediction errors for each candidate, compares them against thresholds, and selectively applies motion vector prediction based on the feedback from these precision measurements. This feedback mechanism ensures coding efficiency is improved only when prediction actually benefits the current block, avoiding unnecessary complexity.
Data Source
AI summary
A motion vector predictor candidate generation unit derives a plurality of motion vector predictor candidates by prediction from first coded prediction blocks neighboring a prediction block subject to coding within the same picture as the prediction block subject to coding, or from a second coded prediction block located at the same position as or in the neighborhood of the prediction block subject to coding in a picture different from that of the prediction block subject to coding, and adds the derived motion vector predictor candidates in a motion vector predictor candidate. The motion vector predictor candidate generation unit determines, for the purpose of obtaining a predetermined number of motion vector predictor candidates, which of first coded prediction blocks provides the motion vector from which to derive the motion vector predictor candidate, organizing the blocks in the order of priority.