Fisheye Camera Calibration Using Semi-Spherical 3D Pattern
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle vision systems using fisheye lens cameras face challenges in accurate calibration due to large radial distortion and misalignment between camera image and sensor planes, which affects the quality of wide-field-of-view imaging and object detection.
Innovation Solution
The system employs an enhanced calibration method using the OCamCalib model, extending it for fisheye lens cameras with intrinsic parameter calibration based on a Taylor polynomial expansion and 3D calibration points distributed on a semi-sphere, allowing for precise estimation of calibration parameters and reducing re-projection errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If fisheye lens cameras are used for wide-field-of-view imaging, then the field of view is expanded, but radial distortion increases and calibration accuracy deteriorates
Solution Approach 1:
The patent applies parameter changes by using Taylor polynomial expansion to model the radial distortion characteristics of fisheye lenses. This mathematical approach transforms the calibration problem into estimating polynomial coefficients that accurately represent the distortion across the wide field of view, thereby improving calibration accuracy while maintaining the expanded field of view capability
Solution Approach 2:
The patent transitions from traditional 2D planar calibration patterns to 3D calibration targets distributed in spatial volume. This dimensional change allows the calibration system to capture and correct radial distortion more effectively across the entire wide field of view by utilizing depth information and three-dimensional point distributions
2Device complexity
If traditional calibration methods are used, then the calibration process is simple, but re-projection errors increase and calibration accuracy deteriorates
Solution Approach 1:
The patent introduces an intermediary calibration target with known 3D point distributions as a mediator between the camera and the calibration process. This intermediary object serves as a reference that enables accurate estimation of intrinsic parameters and distortion coefficients, achieving high calibration accuracy while keeping the process manageable through structured point distributions
3Ease of manufacture
If misalignment between camera image and sensor planes is present, then manufacturing is easier, but imaging quality and object detection accuracy deteriorate
Solution Approach 1:
The patent applies preliminary action by performing calibration to determine the misalignment parameters between the image plane and sensor plane before actual imaging operations. By pre-characterizing the misalignment through calibration and incorporating these parameters into the imaging model, the system compensates for the manufacturing-induced misalignment, thereby maintaining high imaging quality and object detection accuracy despite easier manufacturing processes
Data Source
AI summary
A method for calibrating a camera of a vehicular vision system includes providing a camera and an image processor at the vehicle. A monoview noncoplanar three dimensional calibration pattern distributed on a semi-sphere is determined via processing of image data captured by the camera. Responsive to determination of the monoview noncoplanar three dimensional calibration pattern distributed on the semi-sphere, extrinsic parameters of the camera and intrinsic parameters of the camera are estimated. The system performs at least one of (i) a linear refinement of the estimated extrinsic parameters and intrinsic parameters and (ii) a non-linear refinement of the estimated extrinsic parameters and intrinsic parameters. Responsive to processing by the image processor of image data captured by the camera, the camera is calibrated at least in part by using the determined monoview noncoplanar three dimensional semi-spherical calibration pattern.