Abstract: In light of such problems as low accuracy and low efficiency resulted from the traditional manual size detection method for E-type circlips,an improved detection method based on machine vision was proposed herein. Firstly, the grayscale image was passed through a bilateral filter to attenuate the noise. The measurement zone was quickly and accurately determined through shape matching and a priori knowledge,while the image pyramid hierarchical search strategy was used to improve the shape matching and locating efficiency. The edge contours were extracted from the measurement zone down to the subpixel level using a subpixel edge detection algorithm,and the inner and outer diameter edges and opening edges were obtained. Finally,circle fitting was performed using Iterative reweighted least squares,and the minimum distance among the convex polygons was calculated using the rotation method,which completes the size measurement. Results of experiments carried out using the proposed method indicate high measurement accuracy,high measurement efficiency,and good stability,all of which fulfill the criteria for industrial automation.

Key words: machine vision, shape matching, circlip, size detection