Motion Vector Prediction Using Co-located Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face inefficiencies in selecting motion vectors for pixel groups and subgroups, particularly in handling differing motion vectors within a single pixel group, which can lead to increased processing complexity and memory bandwidth consumption.
Innovation Solution
The proposed solution involves a motion vector prediction technique that selects default motion vectors for pixel groups without shifting them, using co-located prediction units from reference pictures, and scales these vectors based on temporal distances to assign motion vectors to pixel subgroups, allowing for efficient coding without requiring extensive shifting or increased memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If advanced motion vector predictor coding is used to handle differing motion vectors within pixel groups, then video coding quality is improved, but processing complexity and memory bandwidth consumption increase
Solution Approach 1:
The patent divides the pixel group into multiple pixel subgroups, each with its own motion vector. This segmentation allows different motion vectors to be assigned to different regions within the same pixel group, improving coding quality by accurately representing varying motion patterns while managing processing complexity through structured subdivision.
Solution Approach 2:
The patent applies different motion vector prediction strategies to different pixel subgroups based on their local characteristics. By selecting co-located prediction units from reference pictures and scaling them according to temporal distances, the system achieves local optimization of motion compensation accuracy without uniformly increasing complexity across the entire picture.
2Measurement precision
If advanced motion vector predictor coding is used to handle differing motion vectors within pixel groups, then video coding quality is improved, but memory bandwidth consumption increases
Solution Approach 1:
The patent performs preliminary selection of co-located prediction units from reference pictures before the actual motion compensation process. By pre-identifying and scaling appropriate motion vectors based on temporal distances, the system reduces the amount of data that needs to be fetched during decoding, thereby lowering memory bandwidth consumption while maintaining coding quality.
3Measurement precision
If pixel groups are shifted to select default motion vectors, then motion vector accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent copies motion vectors from co-located prediction units in reference pictures to the current picture. Instead of shifting pixel groups to find default motion vectors, the system creates copies of proven motion vectors from temporally correlated reference frames and scales them appropriately, achieving similar accuracy with reduced computational complexity.
Data Source
AI summary
A system includes motion vector prediction circuitry that generates motion vectors for pixel groups and pixel subgroups within the pixel groups. For a current pixel group within the currently processed picture, the motion vector prediction circuitry selects a target reference picture. Without first translating the current pixel group in accord with an offset vector, the motion vector prediction circuitry initiates generation of a default motion vector for the current pixel group. Responsive to the selected target reference picture, the motion vector prediction circuitry generates the offset vector. The motion vector prediction circuitry translates the current pixel group in accord with offset vector and determines motion vectors for pixel subgroups of the current pixel group based on the translated position of the current pixel group.