Report on unmixing algorithms for hyperspectral data

A report on the development of new hyperspectral unmixing algorithms that go beyond the state-of-the-art in two respects. First, the inherent spatial resolution existing very often in hyperspectral images will be exploited and properly incorporated in the devised schemes. Second, various nonlinear mixing models will be investigated and unmixing algorithms adjusted to these models will be sought. The compressive sensing and Bayesian statistical frameworks will be used as the basis of our developments, enabling the design of sparsity aware statistical algorithms addressing the previously mentioned issues.