Hyperspectral Unmixing with Gaussian Mixture Model and Spatial Group Sparsity

Abstract

In recent years, endmember variability has received much attention in the field of hyperspectral unmixing. To solve the problem caused by the inaccuracy of the endmember signature, the endmembers are usually modeled to assume followed by a statistical distribution. However, those distribution-based methods only use the spectral information alone and do not fully exploit the possible local spatial correlation. When the pixels lie on the inhomogeneous region, the abundances of the neighboring pixels will not share the same prior constraints. Thus, in this paper, to achieve better abundance estimation performance, a method based on the Gaussian mixture model (GMM) and spatial group sparsity constraint is proposed. To fully exploit the group structure, we take the superpixel segmentation (SS) as preprocessing to generate the spatial groups. Then, we use GMM to model the endmember distribution, incorporating the spatial group sparsity as a mixed-norm regularization into the objective function. Finally, under the Bayesian framework, the conditional density function leads to a standard maximum a posteriori (MAP) problem, which can be solved using generalized expectation-maximization (GEM). Experiments on simulated and real hyperspectral data demonstrate that the proposed algorithm has higher unmixing precision compared with other state-of-the-art methods.

Gaussian manifold metric learning for hyperspectral image dimensionality reduction and classification

“…Due to the spatial homogeneity of HSI, we assume that each pixel neighborhood obeys a Gaussian distribution 32 – 34 Since hyperspectral remote sensing images have some spatial homogeneity, i.e., the probability of neighboring samples belonging to the same class is large, it is reasonable to approximate the simulation of pixel neighborhoods by Gaussian distribution. By calculating the mean μi and covariance matrix normalΣi of Xi, we obtain a multivariate Gaussian distribution model N(μi,Σi|Xi).…”

UTDN: An Unsupervised Two-Stream Dirichlet-Net for Hyperspectral Unmixing

“…In detail, all the parameters set in our experiments are as follows: the optimization of training loss is ADAM with 10e-4, the spatial-spectral balanced terms λ is set to 0.3, the batch size for training is set to 40, and the regularization term µ is set to 1. For the synthetic dataset experiments, the procedure of generating the abundances maps is followed [21]. The size is created with 60 × 60 pixels and constructed from five endmember classes, whose spectra are randomly selected from the ASTER spectral library [22].…”

A Novel Hyperspectral Unmixing Method based on Least Squares Twin Support Vector Machines

“…Statistical unmixing methods provide a natural framework for representing variability in endmembers (Jin et al, 2019). The core idea of these algorithms is to deduce a posteriori probability density from the prior distribution of endmembers and abundances (Zhou et al, 2018).…”