An ADMM-Based Approach Associated with a Linear Mixing Model Multiplicatively Tuned to Deal with Spectral Variability in Hyperspectral Unmixing

In this work, a hyperspectral unmixing approach, based on an Alternating Direction Method of Multipliers (ADMM), is proposed. This approach is designed for a linear mixing model in which the spectral variability phenomenon is spectrally modeled in a multiplicative manner. The study evaluates the performance of the designed ADMM-based algorithm against the gradient-descent-based algorithms previously proposed. Experiments, using synthetic hyperspectral data, are carried out and the obtained results show that the overall performance of the proposed approach is satisfactory and the proposed algorithm, globally, outperforms the tested approaches from the literature.