Welding seam detection between cylinder and plane using point cloud data

Welding seam detection method for cylinder and plane using point clouds is proposed. Steel used for welding is generally covered with mill scale on its surface, and its color is black so that the image processing is difficult. Furthermore, the measured point cloud with an angle has many measurement errors. In the proposed method, the point clouds are acquired separately at an angle to capture the plane and the side of the cylinder perpendicularly to reduce noise in the depth sensor. The spatial equation of the plane is derived from the point cloud data of the plane, and the centerline and radius of the cylinder are derived from the point cloud of the cylinder’s side by fitting the cylinder model. The welding line between the cylinder and the plane is derived by an ideal intersection line from the models of the plane and the cylinder. Experimental results using the proposed method show that the welding line can be obtained with good accuracy.