Title: A new filter for dimensionality reduction and classification of hyperspectral images using GLCM features and mutual information

Abstract: Dimensionality reduction is an important preprocessing step of the hyperspectral images classification (HSI), it is inevitable task. Some methods use feature selection or extraction algorithms based on spectral and spatial information. In this paper, we introduce a new methodology for dimensionality reduction and classification of HSI taking into account both spectral and spatial information based on mutual information. We characterise the spatial information by the texture features extracted from the grey level cooccurrence matrix (GLCM); we use Homogeneity, Contrast, Correlation and Energy. For classification, we use support vector machine (SVM). The experiments are performed on three well-known hyperspectral benchmark datasets. The proposed algorithm is compared with the state of the art methods. The obtained results of this fusion show that our method outperforms the other approaches by increasing the classification accuracy in a good timing. This method may be improved for more performance
Keywords: hyperspectral images; classification; spectral and spatial features; grey level cooccurrence matrix; GLCM; mutual information; support vector machine; SVM.