Calibration of Fisheye Camera With Colinear Constraint of the Main Optical Axis and Epipolar Line Orthogonality-Transverse Axis

Fisheye camera has widely been applied in many fields for its advantages in the ultrawide field of view (FOV). However, the images taken with fisheye cameras expose severe image distortion, resulting in limits in the application of high-accuracy measurement. For this reason, this article presents a novel calibration model using a stereo pair of fisheye images with a colinear constraint of the main optical axis and epipolar line orthogonality-transverse axis. The relationship between the fisheye image and the perspective image point coordinates is first established through the spherical equidistant and the perspective projection function, and then the calibration model is developed by both the derived collinear constraint model and distortion parameters including radial and tangential distortion. The accuracy of the proposed model is verified through an indoor 3-D calibration field with 300 artificial ground control points (GCPs). Experimental results indicate that the root mean square error (RMSE) of the mean reprojection error in the X- and Y-directions are respectively reduced to 1.6 and 1.8 pixels, and the RMSEs of the space intersection accuracy of the checkpoints in the X-, Y-, and Z-directions are respectively reduced to 2.01, 2.03, and 2.32 mm. Compared with the Schwalbe model and Scaramuzza model, the proposed method can meet the requirements of a fisheye camera calibration accuracy, and improve the calibration accuracy. It is proven suitable for complex calibration scenes with colinear constraint of the main optical axis and epipolar line orthogonality-transverse axis.