Material identification on martian hyperspectral images using bayesian source separation

Identification of materials in a planetological scene observed by an imaging spectrometer is a common problem in remote sensing. Usually the pixel size is larger than the typical size of material change over planets, leading to a linear spatial mixing. We propose here an unsupervised approach based on source separation methods to estimate the pure spectra of the components present in the observed scene and their abundances in each pixel. Previous studies have shown that this approach is interesting for Martian ices. This method assumes the positivity of both the pure spectra and the mixing coefficients. We propose here to apply this technique to detect Martian minerals and we show that adding the sum-to-one constraint (or additivity constraint) on the abundance vectors allows one to improve the estimation performance.