Study on the segmentation of abrasive grains in diamond tools

Segmentation of abrasive images of diamond tools is the most important part of computer vision, which is the basis for detecting the state of diamond tools, such as the characteristics of diamond number, distribution, size and state. As diamond images have rich irregular background textures and reflective features, it is a major challenge to detect and segment these diamond images. The present work demonstrates an invalid edge elimination method for such images to segment and mat the diamond grains, which mainly consists of histogram, edge detection, and morphology. Based on the method, the part of diamond in the original images can be accurately and effectively segmented and matted. Then, the method is also used to segment and mat the diamond grains in other images of different diamond tools, and it also has a good segmenting and matting effect. Finally, the segmented and extracted diamond grains are used to analyse the diamond number, density and its wear state, the result indicates that the proposed method can be used in diamond tool detection.