Automatic Weld Bead Detection on Free-form Surface Parts for Remanufacturing Processes

A framework for automatic weld beads detection and estimation of nominal profile surface of free-form surface parts in the absence of supporting information from designing geometries targeting for repairing processes is presented in this paper. The parts are first digitized to obtain point cloud data, subsequently, an algorithm is used to identify workpiece coordinate system and the measured data is transformed to the new coordinate system. Subsequently, the point cloud data are converted into depth map images. B-spline fitting technique is used to estimate nominal profile surfaces of the parts and the weld beads on the surface are obtained by comparing measured data and the estimated nominal profile surface of the workpieces. A simulation data is used to validate the working principle of the framework. The results demonstrate that the proposed method is capable in estimating of nominal profile surfaces and identifying the defects.