Accounting for spectral variability in hyperspectral unmixing using beta endmember distributions

Two approaches based on the Beta Compositional Model are presented in this thesis for hyperspectral spatial-spectral unmixing (i.e., finding the proportions of each endmember in a hyperspectral image). The two approaches considered are using either Quadratic Programming (QP) optimization or a Metropolis-Hastings (MH) sampler in order to incorporate endmember spectral variability during unmixing. The QP approach determines the proportion values by minimizing the difference between the mean of Beta approximation to the convex combination of Beta endmember distributions, while the MH sampling method takes both the mean and variance into consideration.