Adaptive multiscale sparse unmixing for hyperspectral remote sensing image

Sparse unmixing of hyperspectral images aims to separate the endmembers and estimate abundances mixed pixels. This approach is essential step for many applications involving images. The multi-scale spatial sparse algorithm (MUA) could achieve higher accuracy than state-of-the-art algorithms. regularization parameters, whose combinations markedly influence accuracy, are determined by manually searching in broad parameter space, leading time consuming. To settle this issue, adaptive (AMUA) proposed. Firstly, MUA model converted into a new version using maximum posteriori (MAP) system. Secondly, theories indicating that andnorms equivalent Laplacian multivariate Gaussian functions, respectively, applied explore strong connections among estimated noise variances. Finally, update parameters adaptively optimization process unmixing. Experimental results on both simulated data real show AMUA can substantially improve efficiency at cost negligible accuracy. And series sensitive experiments were undertook verify robustness algorithm.