Author : Shivakumar.B.R, and Pallavi.M

The conventional hard clustering method restricts each point of data set to exclusively just one cluster. As a consequence, with this approach the segmentation results are often very crispy, i.e., each pixel of the image belongs to exactly just one class. However, in many real situations, for images, issues such as limited spatial resolution, poor contrast, overlapping intensities, and noise and intensity in-homogeneities variation make this hard (crisp) segmentation a difficult task. The fuzzy classifier makes use of spatial features extracted from a multispectral data, and a classification image is generated using maximum likelihood classification. Fuzzy cluster analysis is performed by allowing gradual memberships, thus offering the opportunity to deal with data that belong to more than one cluster at the same time. Most fuzzy clustering algorithms are objective function based. They determine an optimal classification by minimizing an objective function. In objective function based clustering usually each cluster is represented by a cluster prototype. A case study is presented on different Fuzzy classification methods by varying the parameters and a comparison is done as to find which method gives higher accuracy and Kappa value. Two classification methods are used here. They are: Maximum Likelihood Classifier and Mahalanobis Distance Classifier. The data considered contains both vegetation and water bodies in equal proportion. The proposed approach decreases the number of misclassifications between the Sea Water and River Water classes and the number of misclassifications between the Hilly Vegetation and Plain Land Vegetation classes raising the overall accuracy to above 80%.

Keywords: Fuzzy C-Means (FCM), Fuzzy Supervised Classification, Maximum Likelihood Classification, Mahalanobis Distance Classification.

Article published in International Journal of Current  Engineering  and Technology, Vol.3,No.2 (June- 2013)