Survey of geometric and statistical unmixing algorithms for hyperspectral images

Spectral mixture analysis (also called spectral unmixing) has been an alluring exploitation goal since the earliest days of imaging spectroscopy. No matter the spatial resolution, the spectral signatures collected in natural environments are invariably a mixture of the signatures of the various materials found within the spatial extent of the ground instantaneous field view of the imaging instrument. In this paper, we give a comprehensive enumeration of the unmixing methods used in practice, because of their implementation in widely used software packages, and those published in the literature. We have structured the review according to the basic computational approach followed by the algorithms, with particular attention to those based on the computational geometry formulation, and statistical approaches with a probabilistic foundation. The quantitative assessment of some available techniques in both categories provides an opportunity to review recent advances and to anticipate future developments.