Abstract:The images under the HSI color space distribution of two dimensional color component scattered were not tight and existing clustering center calculation error,it was difficult to accurately separate background and debris using two dimensional color component,and the segmented ferrographic images still contained many tiny debris which were not needed. The image segmentation by K-means Clustering and the between-cluster variance method was put forward. Spherical particles,cutting particles,severe sliding particles,red oxide particles,fatigue particles were selected to segment by clustering and Ostu method under Lab color space’s two dimensional color component,and segmented images were dealt with three-dimensional mathematical morphology method. The results show that the proposed method can achieve effective segmentation between tiny debris and target debris,and can obtain a more complete color wear particle images,it provides effective basis for identification of the color parameters of debris.