An Adaptive Segmentation Method for Product Surface Inspection

Aiming at the specialty and practical requirement in on-line visual inspection of industrial product surface, an adaptive segmentation algorithm using bi-threshold selection method based on Otsu method is presented. For a gray image with two main peaks in its histogram, the target peak was extracted with two thresholds, which could segment the product surface from background and possibly existed defects. The second threshold was determined under the help of an artificial peak made in a way given in this paper. The experiment results indicate that the proposed method has low computation cost, fast speed and good segmentation performance