Image Segmentation of Wheat Bunt Teliospores Based on K-Means Clustering

In order to study the features of wheat bunt teliospores to be researched quantitatively through image analysis technique, the region of teliospores should be segmented beforehand. In view of the limitation of traditional segmentation methods, a kind of K-means clustering algorithm for colorized image segmentation was proposed according to the characteristics of wheat bunt. B component was chosen as segmentation clustering features, and the termination condition of iteration was the invariance of the sum of R, G and B components for the sum of distances between intra-class pixels to reach its local minimal. An experiment was conducted for wheat bunt teliospores segmentation and the results showed that the proposed algorithm was superior to other traditional methods since it was capable of filtering asymetrical background or impurities, reducing teliospore aherences and increasing the number of segmented teliospores.