Coke Photomicrograph Segmentation Based on an Improved Mean Shift Method

In the view of characteristics for coke micrograph, a segmentation algorithm combining mean shift and edge confidence, is proposed. Firstly, the edge confidence of image pixels is calculated, and with the edge confidence the weighting function of mean shift algorithm is computed, the sampling points of feature space are weighted in order to improve the accuracy of detected modes. Secondly, coke image is segmented preliminarily by iterating the weighted mean shift vector. Because that the number of clusters in initial segmentation is larger than that of the actual clusters, which may result in over-segmentation, the combining conditions are set by the spatial distance and the average value of the edge confidence. The coke photomicrograph is finally segmented with the new combining conditions. Experimental results show that with the proposed algorithm the segmentation among different optical textures of coke is more reasonable and effective.