Fuzzy Markov chains approach to feature selection for high dimensional remote sensing data

Advances in sensor technology for Earth observation make it possible to collect multispectral data in much high dimensionality. For example, the NASA/JPL Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) generates image data in more than 220 spectral bands simultaneously. For such high dimensionality, the appropriate selection of features has a significant effect on the cost and accuracy of an automated classifier. In this paper, a feature selection method using fuzzy Markov chains is proposed. It has been shown that the fuzzy Markov chain is a robust system with respect to small perturbations of the transition matrix, which is not the case for the classical probability Markov chains. In this paper, classical and fuzzy Markov chain approaches are applied to the problem of feature selection for high dimensionality.