Spectral Variability Augmented Sparse Unmixing of Hyperspectral Images

Spectral unmixing expresses the mixed pixels existing in hyperspectral images as product of endmembers and their corresponding fractional abundances, which has been widely used imagery analysis. However, endmember spectra even for from same material an image may include variability due to influence lighting conditions inherent properties materials within different pixels. Though in situ spectral library accommodate such by using multiple represent each kind material, performance improvement be restricted limited number material. Therefore, this article, is directly extracted considered transferable among first time. Furthermore, a further augment sparse synchronously performing endmember-based reconstruction variability-augmented model. By, respectively, imposing smoothness regularization over abundances coefficients, convex optimization-based augmented (SVASU) finally proposed, its convergence also analyzed. Experiments conducted synthetic real-world datasets demonstrate that proposed SVASU method not only significantly improves conventional library-based but outperforms several state-of-the-art algorithms.