Edge detection method of mechanical parts based on circular Laplace algorithm

DAI Qiao-sen, HE Yi-bin, CHEN Yu-chen, LIU Xiang, HE Su-xun, WU Lin-hui, DU Wei, WANG Qiang
(College of Mechanical and Electrical Engineering, Wuhan Institute of Technology, Wuhan 430073, China)

Abstract: Aiming at the problem of discontinuous and unclear edges in the traditional part edge detection method, the anti-noise, edge continuity, and integrity of edge details of the image edge detection algorithm were studied. The efficiency of the traditional Laplacian algorithm was analyzed and studied. Based on the advantages, a new algorithm was adopted, namely the circular Laplacian algorithm. For the collected pictures of gears and watch parts, an experiment of edge contour extraction was designed. The peripheral Laplace algorithm was used to sharpen the edges of the picture to obtain the preliminary edge of the part. Then the mathematical morphology algorithm was used to denoise the extracted edge image to obtain a part edge detection image with clear edges and complete contours. The experimental results show that the circular Laplacian algorithm can retain the high-efficiency characteristics of the traditional algorithm while detecting the complete details of the edge of the part, and the morphological algorithm which is used for the noise in the image can effectively remove the invalid information in the image.

Key words: edge detection; circular Laplace algorithm; rotation invariance; mathematical morphology