Unsupervised Unmixing of Hyperspectral Images Accounting for Endmember Variability

TL;DR

This paper introduces an unsupervised Bayesian method for hyperspectral image unmixing that models endmember variability and spatial information, estimating endmember means and covariances for improved analysis.

Contribution

It presents a novel Bayesian unmixing algorithm that accounts for endmember variability and spatial segmentation using Hamiltonian Monte Carlo sampling.

Findings

Effective estimation of endmember means and covariances

Improved unmixing performance on synthetic and real data

Quantification of material variability in scenes

Abstract

This paper presents an unsupervised Bayesian algorithm for hyperspectral image unmixing accounting for endmember variability. The pixels are modeled by a linear combination of endmembers weighted by their corresponding abundances. However, the endmembers are assumed random to take into account their variability in the image. An additive noise is also considered in the proposed model generalizing the normal compositional model. The proposed algorithm exploits the whole image to provide spectral and spatial information. It estimates both the mean and the covariance matrix of each endmember in the image. This allows the behavior of each material to be analyzed and its variability to be quantified in the scene. A spatial segmentation is also obtained based on the estimated abundances. In order to estimate the parameters associated with the proposed Bayesian model, we propose to use a Hamiltonian Monte Carlo algorithm. The performance of the resulting unmixing strategy is evaluated via simulations conducted on both synthetic and real data.