Weld quality assessment using an edge detection algorithm

Heat input during the welding process and subsequent re-cooling changes the microstructure, hardness, toughness, and cracking susceptibility in heat affected zone (HAZ). Weld quality of a weldment largely depends on the area of HAZ. Determination of exact area of the HAZ by manual stereological methods and conventional visual inspection techniques is a difficult task. These techniques of evaluation are based on approximating the complex shape of HAZ as combination of simplified shapes such as rectangles, triangles etc. In this paper, a filtering scheme based on morphology, thresholding and edge detection is implemented on image of weldments to assess quality of the weld. The HAZ of mild steel specimens welded at different welding parameters by Metal Active Gas/Gas Metal Arc Welding process were compared and presented.