FUZZY CLUSTERS WITH VOLUME PROTOTYPES IN THE THEMATIC PROCESSING OF THE EARTH REMOTE SENSING DATA
Aleksey Aleksandrovich Buchnev, Valeriy Pavlovich Pyatkin
Abstract
The fuzzy clustering technology of the Earth remote sensing data, based on extended C-means and Gustafson-Kessel algorithms, is discussed. The algorithms extensions consist of clusters with volume prototypes construction and using of clusters similarity measure. The volume prototypes are less sensitive to a bias in the distribution of the data, and similar clusters are merged during clustering.