Selective Motion Compensation for Global Motion Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression techniques face inefficiencies due to complex relationships between video quality, data usage, encoding complexity, sensitivity to errors, and the need for accurate motion representation, particularly in handling global and local motion in video frames.
Innovation Solution
Implementing a decoder that extracts a header with a list of reference frames for global motion compensation, determines a global motion model, and decodes blocks using this model to enhance compression efficiency by signaling global motion vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional motion compensation techniques are used, then video compression can be achieved, but compression efficiency is limited due to inability to accurately represent global motion
Solution Approach 1:
The patent segments motion representation into two distinct components: global motion (affecting the entire frame) and local motion (affecting specific regions). By separating these motion types and applying different compensation strategies to each, the system achieves more accurate overall motion representation, directly improving compression efficiency.
Solution Approach 2:
The patent introduces global motion parameters that operate at the frame level, adding a new dimension to traditional block-based motion compensation. This dimensional enhancement allows the system to capture large-scale motion patterns that affect entire frames, complementing the existing local motion compensation and significantly improving compression performance.
2Measurement precision
If more motion vector information is transmitted to improve motion accuracy, then prediction accuracy improves, but data usage increases
Solution Approach 1:
The patent merges global motion parameters with local motion vectors into a unified motion compensation framework. By combining these two types of motion information, the system achieves higher prediction accuracy while avoiding the need to transmit separate, redundant motion data for every block, thus controlling data usage.
Solution Approach 2:
The global motion parameters serve multiple functions simultaneously: they provide a baseline motion model for the entire frame, reduce the magnitude of local motion vectors needed, and enable more accurate prediction across different regions. This multi-functionality improves prediction accuracy without proportionally increasing data usage.
3Manufacturing precision
If complex encoding algorithms are used to improve motion compensation accuracy, then video quality improves, but encoding complexity increases
Solution Approach 1:
The patent divides the encoding process into two stages: global motion parameter estimation (applied once per frame) and local motion compensation (applied to individual blocks). This segmentation allows complex operations to be performed efficiently at the frame level, reducing overall encoding complexity while maintaining high video quality.
Solution Approach 2:
The patent performs global motion estimation as a preliminary step before block-based motion compensation. By pre-calculating the global motion model, the system simplifies subsequent local motion compensation operations, reducing the computational burden during the main encoding process while improving overall motion compensation accuracy.
Data Source
AI summary
Systems and methods are provided for encoding and decoding a video signal in a bitstream including a current coded picture, header information for the current coding unit including a flag indicating whether motion compensation prediction is enabled and that each motion vector of the current coding unit is to be determined relative to a reference picture in the list based on a motion vector of a neighboring coding unit. The methods include determining and applying a motion model when motion compensation prediction is enabled and using intra prediction to decode blocks not encoded with motion compensation prediction.