3D stereo vision measurements for weed-crop discrimination

Abstract

The focus of this paper is weed-crop discrimination using 3D information from a binocular stereo vision system. Two methods of optical distance measurement are presented. Extrinsic camera calibration uses four circular calibration objects. The ground plane equation is defined by three points that corresponds to the circle centers of three circular objects placed on the ground plane. Finally, weed-crop discrimination is done by using colour and height thresholds. The function modules used in Matlab program are also presented.