A image processing and features extraction method for structured light image of welding seam

A method of image processing and features extraction was presented for structured light images of welding seams, by which the seam features images are detected with the disturbances such as reflection. The profiles of grooves were obtained by detecting the peaks of intensity curve of column pixels, and a criterion of temporal and spatial continuity of laser stripe was adopted to judge validness of points. According to the profiles of welding groove, the center of gravity of the groove was selected and extracted as feature point. Experimental results of welding seam tracking testified the effectiveness and reliability of the method.