Surround-View Camera Online Calibration via Feature Point Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current surround-view camera systems in vehicles cannot recalibrate camera positions and orientations online, leading to misalignment issues due to changes in vehicle load, height, and orientation during operation, requiring offline recalibration at service centers.
Innovation Solution
A method for online calibration of surround-view cameras that identifies matching feature points in overlapping image areas, estimates camera parameters in world coordinates, and calculates vehicle pose to adjust camera positions and orientations dynamically while the vehicle is in use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera calibration is performed during vehicle manufacture using a checker-board pattern, then initial camera alignment is achieved, but the calibration becomes inaccurate over time due to changes in vehicle load, height, and orientation
Solution Approach 1:
The patent implements dynamic calibration by continuously tracking feature points in real-time video feeds from multiple cameras. The system adapts camera parameters on-the-fly based on detected feature point positions, allowing the calibration to adjust to changing vehicle conditions such as load variations, height changes, and orientation shifts, rather than relying on static pre-calibration values
Solution Approach 2:
The system employs feedback mechanisms by continuously monitoring feature point positions in overlapping camera fields of view and using this information to iteratively refine camera parameter estimates. The calibration process uses detected feature points to provide feedback on current alignment accuracy, which then drives adjustments to camera parameters to maintain optimal alignment
2Measurement precision
If offline recalibration at service centers is performed, then calibration accuracy is restored, but vehicle downtime and operational disruption increase
Solution Approach 1:
The patent enables continuous calibration by performing the calibration process online using real-time video feeds from the cameras during normal vehicle operation. The system continuously processes video frames, tracks feature points, and updates camera parameters without requiring the vehicle to be taken offline or to a service center, thereby maintaining uninterrupted vehicle operation
Solution Approach 2:
The system performs self-calibration by automatically detecting feature points in the environment, computing camera parameter adjustments, and applying corrections without requiring external intervention or specialized equipment. The vehicle's own camera system and processor are used to conduct the calibration, eliminating the need for service center resources
3Area of stationary object
If multiple cameras are used to provide surround-view coverage, then field of view is improved, but image alignment and stitching complexity increases
Solution Approach 1:
The patent introduces feature points as intermediaries to facilitate the alignment and stitching process. By detecting and tracking common feature points in the overlapping fields of view of multiple cameras, the system creates a reference framework that simplifies the computation of camera parameters and the subsequent stitching of images, making the process more manageable despite the increased number of cameras
Data Source
AI summary
A system and method for providing online calibration of a plurality of cameras in a surround-view camera system on a vehicle. The method provides consecutive images from each of the cameras in the surround-view system and identifies overlap image areas for adjacent cameras. The method identifies matching feature points in the overlap areas of the images and estimates camera parameters in world coordinates for each of the cameras. The method then estimates a vehicle pose of the vehicle in world coordinates and calculates the camera parameters in vehicle coordinates using the estimated camera parameters in the world coordinates and the estimated vehicle pose to provide the calibration.