Three Camera Lens Distortion Correction Models and Its Application

Abstract

Due to the manufacture errors and assembly errors of camera, there are different degrees and types of lens distortion seriously affects the use of camera. Therefore, the camera calibration is the necessary work in photogrammetry. Self-calibration technology is convenient and flexible, which is widely used in camera calibration and has become an important research direction in photogrammetry and computer vision. After long-term development and research, self-calibration has become an important technical core in camera calibration, and the establishment of self-calibration model is the key to ensure high-precision photogrammetry. Three common lens distortion correction models–Brown model, quadratic polynomial model and binary Fourier series model are mainly discussed in this paper. The nonlinear least squares iterative method is used to solve the collinearity equation with additional system parameters. The parameters of the distortion correction model are obtained while solving the elements of interior orientation in the image. The compensation effects of the three models on lens distortion are compared by experiments, and the reliability of the solution of model parameters is verified. The experiment shows that the maximum residual of image point and the norm of residual vector are the smallest when the Brown model is used to compensate the lens distortion. The fitting accuracy of the model for the observed value of image point is the highest, and its accuracy can reach 1$0^{-6}$m.