Abstract: An effective low cost vision system has been developed for automatic seam tracking of marrow gap (NG) TIG welding. The are region including the NG preparation is viewed by a CCD camera, and digitized into 256×256 pixels with 7-bit grey levels by the IBM-PC based frame grabber board. Information about the positions of sidewalls and tungsten electrode is extracted by analyzing of grey levels images in real time. Two methods of image analyzing have been developed, namely edge detection and template matching. Vision information is fed back to the welding rig controller for automatic seam tracking.

Key words: NGW, Seam tracker, Vision system