Abstract:In order to segment the wear particle chain in the ferrography image effectively and improve the accuracy of wear particle feature extraction,a wear particle chain segmentation method combining the color and shape features of the ferrography image was proposed by analyzing the characteristics of different regions of the ferrography wear particle chain image.Firstly,the image of abrasive chain was pre segmented based on morphological operation,and the abnormal large abrasive was extracted and the adhesion part of abrasive chain was disconnected.Then,the wear particle chain was segmented by the method of marked watershed and gray clustering.Finally,considering the difference of color information of different abrasive chains,the gray clustering was improved by combining the adaptive threshold method to realize the adaptive segmentation of abrasive chains.The results show that,compared with traditional segmentation methods,the proposed method can effectively avoid over segmentation and under segmentation,which has better segmentation effect,high applicability,and the operation is simple.