Skip to Main Content
Normal FCM (Fuzzy C-Means Clustering) algorithm is based on the Euclidean distance, which is the isotropic clustering method. In fact, the distribution of remote sensing data pixels don't obey the isotropism or sphere distribution, and we can't get good results in actual usage with normal FCM algorithm. So we choose FCM algorithm based on Mahalanobis distance, which is the appropriate distance measurement method for recognition and is selected according to the actual geographical distribution of remote sensing pixels. Then, we can recognize the land cover types of remote sensing images in the Three Gorges Reservoir Area with FCM algorithm based on Mahalanobis distance. The results show that this classification method can distinguish almost every pixel well, which have the same spectra but not same one, and we make the classified images a significant change with this method.
Geoscience and Remote Sensing (IITA-GRS), 2010 Second IITA International Conference on (Volume:2 )
Date of Conference: 28-31 Aug. 2010