Hierarchical Bayesian sparse image reconstruction with application to MRFM
IEEE Transactions on Image Processing
Joint Bayesian endmember extraction and linear unmixing for hyperspectral imagery
IEEE Transactions on Signal Processing
Bayesian orthogonal component analysis for sparse representation
IEEE Transactions on Signal Processing
IEEE Transactions on Image Processing
Lattice independent component analysis for functional magnetic resonance imaging
Information Sciences: an International Journal
Objective Bayesian analysis for the normal compositional model
Computational Statistics & Data Analysis
Hybrid computational methods for hyperspectral image analysis
HAIS'12 Proceedings of the 7th international conference on Hybrid Artificial Intelligent Systems - Volume Part II
Hi-index | 0.01 |
This paper proposes a hierarchical Bayesian model that can be used for semi-supervised hyperspectral image unmixing. The model assumes that the pixel reflectances result from linear combinations of pure component spectra contaminated by an additive Gaussian noise. The abundance parameters appearing in this model satisfy positivity and additivity constraints. These constraints are naturally expressed in a Bayesian context by using appropriate abundance prior distributions. The posterior distributions of the unknown model parameters are then derived. A Gibbs sampler allows one to draw samples distributed according to the posteriors of interest and to estimate the unknown abundances. An extension of the algorithm is finally studied for mixtures with unknown numbers of spectral components belonging to a know library. The performance of the different unmixing strategies is evaluated via simulations conducted on synthetic and real data.