Double-side visual sensing and image processing of weld pool in P-GMAW

Vision is the most important method for a manual welder to monitor the weld process. To simulate this process, a double-side visual sensing system was established for observing weld pool in pulsed gas metal arc welding (P-GMAW). With the passive illumination, the filtering method, imaging time, imaging mechanism were investigated for both topside and backside weld pool in P-GMAW. The results suggest, with narrow band filter centered at 665 nm, clear topside weld pool image can be captured in base current period; and with neutral filter, clear backside weld pool image can be captured. It can be seen that the light source of double side imaging system mainly come from the radiation of weld pool itself and this mechanism ensures the imaging quality. Meanwhile, the image characters in different welding conditions were studied, and based on these characters, a series of image processing algorithm were proposed to extract the topside and backside weld pool boundary.