The weld image edge-detection algorithm combined with Canny operator and mathematical morphology

Edge-detection is the important link in weld image processing, according to the problem of noise reduction and edge details preservation in weld image edge detection, this paper presents a weld image edge-detection algorithm combined with Canny operator and mathematical morphology. Pseudo-edges of Canny edge-detection Algorithm is eliminated by edge-detection operator based on the mathematical morphology, and weld center curve is obtained by local-region scanning. Relative to based on Sobel operator and Prewitt operator, this algorithm inhibits the interference of noise on the edge and extracts the information of weld edge accurately. The standard error of weld center curve detected by this algorithm and actual weld center curve is decreased. Based on the advantages of both keeping Canny operator accurate location and high signal-noise ratio, this algorithm improves the performance of the Canny operator in terms of inhibition of pseudo-edges and is suitable for the process of image processing in welding seam tracking based on vision sensor.