The 3D position recognition of multiple coils using ellipse fitting with probabilistic edge detection with stereo cameras

The development of unmanned crane is an important issue of Factory Automation in steel industry. This paper focus on unmanned crane for picking up coils on a trailer although positions and diameters of coils is not fixed. In order to pick up those coils, unmanned crane should know exact 3D coil position. Therefore, in order to find out exact 3D coil’s centers, this paper propose the system that recognize 3D position of coil center based on ellipse detection and 3D reconstruction method using stereo vision system. In details, the probabilistic ellipse oriented edge detection method and ellipse fitting method using RANSAC algorithm are proposed. Then, resolving correspondence problem using epipolar geometry and 3D reconstruction using calibrated stereo vision method are used to complete our work. Experiments show the effectiveness of the proposed method.