An algorithm for camera calibration using a three-dimensional reference point

Abstract

A new algorithm for camera calibration using a three-dimensional (3D) reference point as the calibration target is described. The main idea of the method is that the relationship between two cameras' parameters, i.e. the pan angle θ (horizontal viewing angle) and the tilt angle ω (looking angle) and the world coordinates of a given 3D calibration point can be established by the principles of perspective projection and geometrical optics, and the solution of the transcendental equations for calibrating camera parameters can be then solved by an iterative algorithm. The experimental results show that using our method, the maximum relative errors of calibrated parameters θ and ω are 4.02 and 5.53%, respectively. The method has two important advantages: (1) calibration can be done even in the case that the calibration target is far away from the camera; (2) the shape of the calibration object is the simplest among currently available calibration methods.