Abrasive Segmentation of Multiple Diamond Images Based on Secondary Morphological Reconstruction

Abstract

Faced with a lot of noise problems in diamond abrasive image segmentation under complex background, a segmentation method of diamond abrasive images is proposed based on secondary morphological reconstruction. The method of histogram equalization was firstly used to pre-process the original image, which aim is to improve the image by adjusting the global contrast, and an appropriate structural element is constructed to carry out to determine abrasive size by using the morphological smoothing operation; Secondly, the first morphological reconstruction is carried out to further virtualize the complex background and clear garget abrasives by using the geodesic corrosion operation; Then, the second morphological reconstruction is carried out by calculating geodesic expansion operation. Combined with the local extreme value processing of the image, each target abrasive area can be obtained. Finally, the segmented target abrasive area is used to extract the multiple abrasives from complex background. It is found that the complex background noise can be readily eliminated by the proposed method, and the abrasives in the image can be well segmented. The results show that segmentation boundary is clear, without background noise, and the segmentation region is significant. The extracted diamond abrasives can be used for advanced image analysis such as abrasive characteristics and abrasive identification.