Sparse semi-supervised hyperspectral unmixing using a novel iterative Bayesian inference algorithm

In this paper a novel hierarchical Bayesian model for sparse semi-supervised hyperspectral unmixing is presented. Adopting the sparsity hypothesis and taking into account the convex constraints of the estimation problem, suitable priors are selected for the model parameters. Then, a new low-complexity, iterative conditional expectations algorithm is developed to perform Bayesian inference. The proposed method converges fast to a sparse solution, which offers improved estimation accuracy. The theoretical results presented in the paper are fully verified by simulations both on synthetic and real hyperspectral data.