Stereo camera calibration based on accurate control points

This paper presents a high-precision stereo camera calibration technique combined the improved two-step method with accurate control points, in which the lens distortion of a camera is modeled by a combination of radial distortion and tangential distortion. Accurate control points on a virtual stereo calibration pattern are formed by an infrared LED moved with the probe of CMM over a certain distance. The proposed calibration procedure involves direct linear solutions for most of the calibration parameters with the control points near the center of images, followed by some iteration and optimization solutions for all parameters with all control points. The accuracy of reconstructing control points with nonlinear optimization solutions is improved by 0.11 mm as compared with linear solutions. Experimental results show that the proposed technique can fulfil the goals of high-precision calibration and measurement of cameras.