Online visual quality inspection for weld seams

Arc welding is a widely used technology in almost all sectors of industrial production. Many task are automatically performed by robots. This paper presents a flexible vision-based quality management system to detect defects online during the weld process. The proposed method uses a single camera on the weld arms to observe the newly welded seam behind the weld head. Two phases, training and verification, are included in the system. The quality parameters are learned from a set of welded seams (training seams), selected by a weld engineer, which show sufficient quality, and by which the measurement reference and tolerance are formed. The verification process is conducted by frame-to-frame comparison of current work with the reference. The observed features are the position and the width of the seam as well as the light distribution in the vicinity of the arc.