Spectral Variability in Hyperspectral Data Unmixing: A comprehensive review

The spectral signatures of the materials contained in hyperspectral images, also called endmembers (EM), can be significantly affected by variations atmospheric, illumination or environmental conditions typically occurring within an image. Traditional unmixing (SU) algorithms neglect variability endmembers, what propagates significant mismodeling errors throughout whole process and compromises quality its results. Therefore, large efforts have been recently dedicated to mitigate effects SU. This resulted development that incorporate different strategies allow EMs vary a image, using, for instance, sets known priori, Bayesian, parametric, local EM models. Each these approaches has characteristics underlying motivations. paper presents comprehensive literature review contextualizing both classic recent solve this problem. We give detailed evaluation sources their effect image spectra. Furthermore, we propose new taxonomy organizes existing works according practitioner's point view, based on necessary amount supervision computational cost they require. methods used construct libraries (which are required many SU techniques) observed as well library augmentation reduction. Finally, conclude with some discussions outline possible future directions field.