CAMERA CALIBRATION METHOD FOR HIGH ACCURACY STEREOVISION

This paper presents a camera calibration method for far-distance, high accuracy stereo-vision. There are many general purpose methods found in the literature for solving the common issues of the camera calibration: intrinsic parameters estimation (focal length and principal point) and extrinsic parameters estimation (the absolute position and orientation of the cameras relative to the world coordinate system). Their results are acceptable for monocular or near range stereo vision but are improper for far range applications. For a high accuracy stereovision system the most critical parameters are the relative orientation of the two cameras positioned on the stereo-rig. Experiments proved that even a few seconds drift of one of the relative camera angles can lead to disastrous consequences in the whole stereo-vision process: incorrect epipolar lines estimations, and consequently lack of reconstructed 3D points. Therefore we propose an extrinsic parameters calibration method able to give very accurate extrinsic parameters of the stereo-rig. The main idea is to use a dedicated calibration scene, having the sizes comparable with the stereo-vision application’s working range, with some control points placed in known 3D positions. The extrinsic parameters are estimated by minimizing the image projection error of the 3D control points relative to their detected 2D image positions. The obtained results are allowing the reconstruction of 3D features up to 100 m with an error bellow 1%