Image Clustering Based on Extreme K-means Algorithm

Classification is an important step for automatic recognition. This paper present a new Classification method based on our Improved K-means clustering algorithm. As it is known that the performance of the traditional k-means algorithm largely depends on the choice of the initial centers, and the algorithm generally uses random procedures to get them. In order to improve the efficiency of the k-means algorithm, a good selection method of clustering starting centers is proposed in this paper. The proposed algorithm determines an initial scale for each cluster of patterns, and calculate initial clustering centers according to the norm of the points. Experiments results show that the proposed algorithm provides good performance of clustering.