Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization

Abstract

This paper introduces a robust mixing model to describe hyperspectral data resulting from the mixture of several pure spectral signatures. This new model not only generalizes the commonly used linear mixing model, but also allows for possible nonlinear effects to be easily handled, relying on mild assumptions regarding these nonlinearities. The standard nonnegativity and sum-to-one constraints inherent to spectral unmixing are coupled with a group-sparse constraint imposed on the nonlinearity component. This results in a new form of robust nonnegative matrix factorization. The data fidelity term is expressed as a beta-divergence, a continuous family of dissimilarity measures that takes the squared Euclidean distance and the generalized Kullback-Leibler divergence as special cases. The penalized objective is minimized with a block-coordinate descent that involves majorization-minimization updates. Simulation results obtained on synthetic and real data show that the proposed strategy competes with state-of-the-art linear and nonlinear unmixing methods.

References in corpus (11)

• Nonlinear unmixing of hyperspectral images: models and algorithms
• Robust Principal Component Analysis?
• Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization
• A convex model for non-negative matrix factorization and dimensionality reduction on physical space
• Nonlinear spectral unmixing of hyperspectral images using Gaussian processes
• A Hierarchical Bayesian Framework for Constructing Sparsity-inducing Priors
• Minimum mean square distance estimation of a subspace
• Algorithms for nonnegative matrix factorization with the beta-divergence
• Robust Nonnegative Matrix Factorization via $L_1$ Norm Regularization
• On the unmixing of MEx/OMEGA hyperspectral data
• Automatic Relevance Determination in Nonnegative Matrix Factorization with the β-Divergence

Cited by in corpus (26)

• Hyperspectral unmixing with spectral variability using a perturbed linear mixing model
• Nonlinear hyperspectral unmixing with robust nonnegative matrix factorization
• Robust Volume Minimization-Based Matrix Factorization for Remote Sensing and Document Clustering
• Unsupervised Unmixing of Hyperspectral Images Accounting for Endmember Variability
• Hyperspectral Unmixing via Deep Autoencoder Networks for a Generalized Linear-Mixture/Nonlinear-Fluctuation Model
• Hyperspectral Unmixing in Presence of Endmember Variability, Nonlinearity or Mismodelling Effects
• Online Nonnegative Matrix Factorization with Outliers
• Robust Sparse Blind Source Separation
• Multi-Resolution Beta-Divergence NMF for Blind Spectral Unmixing
• Multiplicative Updates for NMF with $\beta$-Divergences under Disjoint Equality Constraints
• Majorization-minimization for Sparse Nonnegative Matrix Factorization with the $\beta$-divergence
• Learning Nonlinear Mixtures: Identifiability and Algorithm
• Adaptive Multi-Order Graph Regularized NMF with Dual Sparsity for Hyperspectral Unmixing
• Privacy-preserving Non-negative Matrix Factorization with Outliers
• Joint Majorization-Minimization for Nonnegative Matrix Factorization with the $\beta$-divergence
• Expanding boundaries of Gap Safe screening
• Adversarially-Trained Nonnegative Matrix Factorization
• Negative Binomial Matrix Factorization for Recommender Systems
• Fast Hyperspectral Unmixing in Presence of Nonlinearity or Mismodelling Effects
• Online Convex Matrix Factorization with Representative Regions
• Feature Weighted Non-negative Matrix Factorization
• Collaborative Filtering and Multi-Label Classification with Matrix Factorization
• Robust Fusion of Multi-Band Images with Different Spatial and Spectral Resolutions for Change Detection
• Sparse component separation from Poisson measurements
• Coupled dictionary learning for unsupervised change detection between multi-sensor remote sensing images
• Robust spectral unmixing of sparse multispectral Lidar waveforms using gamma Markov random fields