Non-homogeneous hidden Markov chain models for wavelet-based hyperspectral image processing

We consider the use of non-homogeneous Markov chain (NHMC) models for wavelet transformations of hyperspectral signatures to generate features for signal processing purposes. Inspired by the use of hidden Markov trees for natural images, the NHMC model enables the characterization of absorption bands and other structural features of mineral spectra that are used by experts in tasks like classification and unmixing, primarily in an ad-hoc fashion. We show that NHMC models can successfully identify and capture the information in a spectral signature dataset that can be exploited by standard classification algorithms to identify and differentiate spectral families. We also identify several metrics that can help determine whether each spectral band is informative to classification in a multiscale fashion.