Pre-warping Global Motion Compensation Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In vehicle image processing, camera motion between image frames complicates the determination of environmental awareness by introducing longer-than-average motion vectors, increasing processing difficulty and requiring additional computing resources, especially in applications with high speeds, high resolutions, or reduced frame rates.
Innovation Solution
A global motion model is used to remove camera-specific orientation and distortion, allowing for pre-warping of image frames to simplify optical flow estimation, reducing the search range and complexity of image processing algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera motion between image frames is accounted for in optical flow processing, then situational awareness accuracy is improved, but processing complexity and computing resources increase
Solution Approach 1:
The patent segments the motion compensation process into two distinct stages: global motion compensation (GMC) that handles camera-specific orientation and distortion, and local optical flow processing that handles residual motion. This segmentation allows each stage to focus on specific aspects of motion, reducing overall processing complexity while maintaining accuracy.
Solution Approach 2:
The patent applies preliminary global motion compensation to remove camera motion effects before performing optical flow processing. By pre-removing the dominant camera motion component, the subsequent optical flow algorithm only needs to handle residual local motion, significantly reducing search range and computational requirements.
2Measurement precision
If search range is increased to accommodate longer motion vectors from camera motion, then motion estimation accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary global motion compensation to remove camera-induced motion before optical flow processing. This preliminary action reduces the magnitude of motion vectors that the optical flow algorithm must handle, allowing for a smaller search range and reduced processing time while maintaining motion estimation accuracy.
3Measurement precision
If high image resolution is used to improve detection accuracy, then object detection precision is improved, but processing resources and time increase
Solution Approach 1:
The patent segments processing into global motion compensation (computationally efficient) and local optical flow (detailed analysis). This segmentation allows high-resolution images to be processed by first removing dominant camera motion with a efficient global model, then applying detailed optical flow only where needed, maintaining detection precision while improving overall processing efficiency.
4Quantity of substance
If frame rate is reduced to bandwidth for multiple camera inputs, then system bandwidth is improved, but motion compensation accuracy deteriorates
Solution Approach 1:
The patent changes the approach to motion compensation by using a global motion model that explicitly models camera orientation and distortion parameters. This parameter-based approach allows accurate motion compensation even at lower frame rates, as the global model can interpolate and extrapolate motion between frames more effectively than traditional optical flow methods.
Data Source
AI summary
This disclosure provides systems, methods, and devices for vehicle driving assistance systems that support image processing. In a first aspect, a method of image processing includes receiving image data comprising a first image frame and a second image frame; determining a global motion model corresponding to the first image frame and the second image frame; warping the first image frame based on the global motion model to determine a warped first image frame; and determining a local flow based on the warped first image frame and the second image frame. Other aspects and features are also claimed and described.