Linear Feature Extraction for Hyperspectral Images Using Information Theoretic Learning

Linear Feature Extraction for Hyperspectral Images Using Information Theoretic Learning

in this paper, we propose a new linear feature extraction scheme for hyperspectral images. A modified Maximum relevance, Min redundancy (MRMD) is used as acriterion for linear feature extraction. Parzen density estimator and instantaneous entropy estimation are used for estimating mutual information. Using Instantaneous entropy estimatormitigates nonstationary behavior of the hyperspectral data and reduces computational cost. Based on proposed estimator andMRMD, an algorithm for linear feature extraction in hyperspectral images is designed that is less offended by Hueghsphenomenon and has less computation cost for applying to hyperspectral images. An ascent gradient algorithm is used for optimizing proposed criterion with respect to parameters of alinear transform. Preliminary results achieve better classification comparing the traditional methods.