Abstract
A popular approach for landcover classification in remotely sensed satellite images is clustering the pixels in the spectral domain into several fuzzy partitions. It has been observed that performance of the clustering algorithms deteriorate with more and more overlaps in the data sets. Motivated by this observation, in this article a two-stage fuzzy clustering algorithm is described that utilizes the concept of points having significant membership to multiple classes. The points situated in the overlapped regions of different clusters are first identified and excluded from consideration while clustering. Thereafter, these points are given class labels based on Support vector Machine classifier which is trained by the remaining points. The well known Fuzzy C-Means algorithm and some recently proposed genetic clustering schemes are utilized in the process. The effectiveness of the two-stage clustering technique has been demonstrated on IRS remote sensing satellite images of the cities of Bombay and Calcutta and compared with other well known clustering techniques. Also statistical significance test has been carried out to establish the statistical significance of the clustering results.