Accurate calibration of stereo cameras for machine vision

Camera calibration is an important task for machine vision, whose goal is to obtain the internal and external parameters of each camera. With these parameters, the 3D positions of a scene point, which is identified and matched in two stereo images, can be determined by the triangulation theory. This paper presents a new accurate estimation of CCD camera parameters for machine vision. We present a fast technique to estimate the camera center with special arrangement of calibration target and the camera model is aimed at efficient computation of camera parameters considering lens distortion. Built on strict geometry constraint, our calibration method has compensated the error for distortion cased by circular features on calibration target, which gets over the relativity influence of every unknown parameters of traditional calibration way and make the error distributed among the constraint relation of parameters, in order to guarantee the accuracy and consistency of calibration results. Experimental results are provided to show that the calibration accuracy is high