Bi-predictive Motion Vector Prediction Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding techniques, such as those in the HEVC standard, face challenges in efficiently encoding and decoding bi-predictive video blocks due to limitations in motion vector prediction, which affects compression efficiency.
Innovation Solution
The implementation of adaptive motion vector prediction (AMVP) techniques that allow one motion vector of a bi-predictive video block to be used as a candidate predictor for another, often using a scaled version of the first motion vector to improve compression by expanding the set of candidate predictors for the second motion vector.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If traditional motion vector prediction methods are used in bi-predictive inter mode, then the encoding process is simple, but compression efficiency is limited due to insufficient candidate predictors
Solution Approach 1:
The patent combines the first motion vector and its scaled version into a unified candidate predictor set for the second motion vector. By merging these two sources of prediction candidates, the system expands the available predictors without requiring entirely separate prediction mechanisms, thus improving compression efficiency while controlling complexity.
Solution Approach 2:
The patent applies scaling operations to transform the first motion vector into a scaled version that serves as an additional candidate predictor. This parameter transformation (scaling) generates new prediction candidates from existing motion vectors, enriching the predictor set and improving compression performance without fundamentally changing the prediction framework.
2Measurement precision
If the set of candidate motion vector predictors is expanded to include scaled versions of first motion vector, then prediction accuracy improves, but encoding complexity increases
Solution Approach 1:
The patent performs scaling of the first motion vector in advance during the candidate predictor generation phase, before the actual motion vector prediction occurs. This preliminary preparation of scaled candidates ensures that when prediction is needed, accurate predictors are already available, improving prediction accuracy without adding complexity during the critical prediction execution phase.
3Productivity
If more candidate predictors are used for the second motion vector, then coding efficiency improves, but the data required for encoding increases
Solution Approach 1:
The patent creates a scaled copy of the first motion vector to serve as an additional candidate predictor for the second motion vector. This copying approach generates diverse prediction candidates from existing data without requiring entirely new information, thereby improving coding efficiency while minimizing the increase in encoding data requirements.
Data Source
AI summary
This disclosure describes video coding techniques applicable to a bi-predictive inter mode that uses adaptive motion vector prediction (AMVP). In particular, this disclosure describes techniques for encoding or decoding video data in which AMVP is used to code a first motion vector and a second motion vector associated with a current video block that is coded in the bi-predictive inter mode. More specifically, this disclosure describes techniques in which one motion vector of a bi-predictive video block can be used to define a candidate motion vector predictor (MVP) for predicting another motion vector of the bi-predictive video block. In many examples, a scaled version of the first motion vector of the bi-predictive video block is used as the MVP for the second motion vector of the bi-predictive video block. By defining an additional MVP candidate for the second motion vector of a bi-predictive video block, improved compression may be achieved.