N
Nicolas Dobigeon
Researcher at University of Toulouse
Publications - 267
Citations - 13074
Nicolas Dobigeon is an academic researcher from University of Toulouse. The author has contributed to research in topics: Hyperspectral imaging & Gibbs sampling. The author has an hindex of 45, co-authored 253 publications receiving 11241 citations. Previous affiliations of Nicolas Dobigeon include University of Michigan & Institut Universitaire de France.
Papers
More filters
Journal ArticleDOI
Estimating the Number of Endmembers in Hyperspectral Images Using the Normal Compositional Model and a Hierarchical Bayesian Algorithm
TL;DR: This paper proposes to estimate the mixture coefficients of the Normal Compositional Model (referred to as abundances) as well as their number using a reversible jump Bayesian algorithm.
Journal ArticleDOI
Segmentation of Skin Lesions in 2-D and 3-D Ultrasound Images Using a Spatially Coherent Generalized Rayleigh Mixture Model
TL;DR: This paper addresses the problem of jointly estimating the statistical distribution and segmenting lesions in multiple-tissue high-frequency skin ultrasound images with a hybrid Metropolis-within-Gibbs sampler and a Markov chain Monte Carlo method.
Proceedings ArticleDOI
Bayesian fusion of hyperspectral and multispectral images
TL;DR: This paper presents a Bayesian fusion technique for multi-band images that combines a Gibbs sampling algorithm with a Hamiltonian Monte Carlo step to efficiently sample from a high-dimensional distribution.
Journal ArticleDOI
Joint Segmentation of Multivariate Astronomical Time Series: Bayesian Sampling With a Hierarchical Model
TL;DR: This paper approaches problems using Bayesian priors to represent relationships between signals with various degrees of certainty, and not just rigid constraints, by using a hierarchical Bayesian approach to a piecewise constant Poisson rate model.
Proceedings ArticleDOI
Unmixing hyperspectral images using the generalized bilinear model
TL;DR: A generalized bilinear model recently introduced for unmixing hyperspectral images is considered, and the positivity and sum-to-one constraints for the abundances are ensured by the proposed algorithms.