FAST FUZZY CLUSTERING

This paper presents a multistage random sampling fuzzy c-means-based c lustering algorithm, which significantly reduces the computation time required to partition a data set into c classes. A series of subsets o f the full data set are used to create initial cluster centers in orde r to provide an approximation to the final cluster centers. The qualit y of the final partitions is equivalent to those created by fuzzy c-me ans. The speed-up is normally a factor of 2-3 times, which is especial ly significant for high-dimensional spaces and large data sets. Exampl es of the improved speed of the algorithm in two multi-spectral domain s, magnetic resonance image segmentation and satellite image segmentat ion, are given. The results are compared with fuzzy c-means in terms o f both the time required and the final resulting partition. Significan t speedup is shown in each example presented in the paper. Further, th e convergence properties of fuzzy c-means are preserved. (C) 1998 Else vier Science B.V.