Semi-supervised classification of hyperspectral data using spectral unmixing concepts

Spectral unmixing and classification have been widely used in the recent literature to analyze remotely sensed hyperspectral data. However, possible connections between semi-supervised classification and spectral unmixing concepts have been rarely investigated. In this work, we propose a new method to perform semi-supervised classification of hyperspectral images by exploiting the information retrieved with spectral unmixing. The proposed method integrates a well-established discriminative classifier (multinomial logistic regression) with different spectral unmixing chains, thus bridging the gap between unmixing and classification. Furthermore, the proposed method uses active learning when generating new unlabeled samples for classification. The proposed method is experimentally validated using real hyperspectral data sets, indicating that the combination of spectral unmixing and semi-supervised classification can lead to powerful new algorithms for hyperspectral data interpretation.