Abstract
In this paper, we propose a novel ensemble method for the classification of hyperspectral images. First, we run the mean-shift (MS) algorithm several times on the most relevant bands identified by the Markov Fisher Selector (MFS) algorithm to generate a set of different segmentation maps. Then in a second step, we label the regions in each MS map by applying the weighted-majority-voting rule (WMV) rule to a spectral-based classification map generated by the support vector machine (SVM) classifier. Finally, we fuse this set of spectral-spatial classification maps with an OWA operator. The determination of the associated weights is made using a Gaussian stress function. The results obtained on a hyperspectral image acquired by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor confirm the promising capabilities of the proposed classification system.