Vehicle Camera Calibration via Feature Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The quality of image stitching in vehicle surround view systems is poor due to offsets in camera positions and orientations after assembly, and existing calibration methods lack an effective online objective evaluation metric.
Innovation Solution
A dynamic camera calibration system that uses an image processor to determine and track matching features in overlapping fields of view, minimizing a cost function error to refine calibration results in a feedback loop, enhancing stitching quality and display of stitched images during vehicle maneuvers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If cameras are mounted on vehicle with standard assembly procedures, then installation efficiency is improved, but stitching quality deteriorates due to position and orientation offsets
Solution Approach 1:
The system performs preliminary calibration by capturing images of a calibration target before normal operation. This preliminary action establishes reference data that compensates for assembly variations, allowing efficient installation without requiring manual adjustment while maintaining stitching quality.
Solution Approach 2:
The system changes calibration parameters (intrinsic and extrinsic parameters) based on captured images and calculated metrics. By dynamically adjusting these parameters through the calibration process, the system compensates for position and orientation offsets while maintaining installation efficiency.
2Manufacturing precision
If offline calibration methods are used to improve stitching quality, then stitching quality is improved, but calibration time and complexity increase
Solution Approach 1:
The system performs self-calibration by automatically capturing images, calculating metrics, and adjusting parameters without requiring external intervention. The calibration target and automated metric calculation enable the system to calibrate itself, reducing both time and complexity compared to manual offline methods.
Solution Approach 2:
The system replaces complex manual calibration procedures with an automated computational approach. By using image processing algorithms and automated metric calculation, the system substitutes mechanical/manual adjustment processes with efficient computational methods that reduce calibration time while maintaining precision.
3Device complexity
If no objective evaluation metric is used, then system complexity is reduced, but stitching quality assessment becomes subjective and unreliable
Solution Approach 1:
The system introduces a calibration target as an intermediary object that enables objective measurement. The target provides known reference features that serve as a mediator between the cameras and the calibration process, allowing automated metric calculation without significantly increasing system complexity.
Solution Approach 2:
The calibration target uses distinct visual features (patterns, colors, or geometric shapes) that enable automated detection and metric calculation. These visual characteristics allow the system to objectively assess stitching quality through image processing without requiring complex additional hardware.
Data Source
AI summary
A camera calibration system for cameras of a vehicle includes a plurality of cameras disposed at a vehicle and having respective fields of view exterior of the vehicle, with the fields of view of two of the cameras overlapping. While the vehicle is moving, an image processor of a control processes image data captured by the two cameras to determine and track matching features in an overlapping region of the fields of view of the two cameras. The control, responsive to image processing of image data captured by the two cameras, determines motion of matching features and minimizes a cost function of the error in the matching of the features. The camera calibration system uses the determined minimized cost function to refine calibration results of the two cameras using a feedback loop.