EXTENDED MORPHOLOGICAL PROFILES WITH DUALITY FOR HYPERSPECTRAL IMAGE CLASSIFICATION

In this paper, the Extended Morphological Profile with duality is proposed and studied for hyperspectral image classification, by which, the shape noise is reduced and thus better classification accuracy is obtained compared to the conventional Extended Morphological Profile technique. Moreover, the integration of a linear filtering technique and Support Vector Machine based classifier is also used for classification of the hyperspectral images and it has been shown that the classification accuracy with urban data set is further improved. The classification map is also improved after filtering, in particular where data set is more congested. The shadows are more clearly shown in classification map after filtering.