High precision occupies an extremely important position in the field of mechanical processing and laser measurement. Under these high precision requirements, image processing is widely applied and is especially demanding. As the essential pre-processing step in the process of image processing, the quality of image edge extraction directly affects the processing precision of the whole image, and then affects the final measurement or machining precision. The traditional edge detection method has the defects of no noise immunity, and cannot achieve high-activity processing of sophisticated image edge problems. Through the in-depth research and analysis of various knowledges concerning edge extraction, a novel anti-noise edge detector based on multi-structure elements morphology of different directions for binary, gray scale and color images is proposed in this paper. We get the final edge information by using eight morphological operations respectively and synthetic weighted method. It can remove the noise effectively while detecting complete edge information. Experimental results show that, comparing with conventional edge detection operations, the proposed method attains the outcome of eliminating the image noise and maintaining good edge effectively for simulated image.