Unsupervised fuzzy clustering with multi-center clusters

A new unsupervised fuzzy clustering algorithm is provided in this paper to cluster the data patterns without a priori information about the number of clusters. The initial guesses of the locations of the cluster centers or the initial guesses of the membership values are not necessary. With the minimization of a new objective function, cluster centers are generated one by one. Related centers are defined to belong to the same cluster. Multi-centers are adopted to represent the non-spherical shape of clusters. Thus, the clustering algorithm with multi-center clusters can handle non-traditional curved clusters. The proposed algorithm is tested on different data sets with a variety of cluster shapes, cluster densities, and number of points in each cluster. Also, the results are compared with some other clustering algorithms to show the effectiveness of the algorithm. Moreover, the designed unsupervised fuzzy clustering algorithm is applied to cluster the pixels in a color image to show the efficiency of the algorithm. (C) 2002 Elsevier Science B.V. All rights reserved.