Abstract: Analysis on the structured light image of weld seam was carried out since the welding seam tracking method with the structured light as an active lighting can be used to recognize the position of the weld seam with different groove shapes.The Beamlet transform was introduced to detect the line segment characteristic of the structured light in the image to identify the position of the weld seam effectively.The theory of Beamlet transform was introduced, and a line segment characteristic detection algorithm was developed for the structured light image of weld seam based on multiscale Beamlet transform.Test based on the maximal Beamlet statistics was carried out, Beamlet transform at a small scale to the structured light image of weld seam was done to identify the rough position of the structured light and reduce the searching range, and Beamlet transform at a large scale was done to identify the line segment characteristic of the structured light on the weld seam, then the weld seam position can be detected according to the slope variety of the line segments.The time to detect the position of the weld seam in a structured light image of weld seam is 260 ms by the algorithm.