Hyperspectral Image Classification Using Spatial Features Extracted by Fuzzy C-Means and Dirichlet Mixture Model

Abstract

The spectral content of the small number of training data may not be enough for the classification of high-dimensional hyperspectral images. For this reason, spatial information is also exploited next to the spectral information. In this study, it is intended to classify hyperspectral images using spatial features extracted by fuzzy C-means (FCM) and Dirichlet Mixture Model (DMM). The contribution of the cascaded use of proposed methods are presented in the results section by tables.