Method of detecting weld deviation based on the image feature of molten pool's edge

Through the research and analysis of the image of all-position pipeline gas metal arc welding (GMAW) pool, a phenomenon that the root welding molten pool image tip edge is overlapped with the weld groove center has been found and verified. Based on this discovery, a kind of method for detecting the welding deviation has been developed. The principle of this method is to detect the tip position of the root weld pool's CCD image as the weld groove center position coordinate value by the way of median filtering, wavelet transform, region segmentation and searching algorithm. Then the welding deviation can be calculated by the difference of weld groove center and wire center coordinate value. Experiments show that this method can obtain the welding deviation from the root weld pool image in real time and can provide a reliable basis for the automatic seam tracking control of welding robot.