Correntropy Maximization via ADMM - Application to Robust Hyperspectral Unmixing

Fei Zhu (1) , Abderrahim Halimi (2) , Paul Honeine (3) , Badong Chen (4) , Nanning Zheng (4)

1 TJU - Tianjin University
2 HWU - Heriot-Watt University [Edinburgh]
3 DocApp - LITIS - Equipe Apprentissage
4 Xjtu - Xi'an Jiaotong University

In hyperspectral images, some spectral bands suffer from low signal-to-noise ratio due to noisy acquisition and atmospheric effects, thus requiring robust techniques for the unmixing problem. This paper presents a robust supervised spectral unmixing approach for hyperspectral images. The robustness is achieved by writing the unmixing problem as the maximization of the correntropy criterion subject to the most commonly used constraints. Two unmixing problems are derived: the first problem considers the fully constrained unmixing, with both the nonnegativity and sum-to-one constraints, while the second one deals with the nonnegativity and the sparsity promoting of the abundances. The corresponding optimization problems are solved using an alternating direction method of multipliers (ADMM) approach. Experiments on synthetic and real hyperspectral images validate the performance of the proposed algorithms for different scenarios, demonstrating that the correntropy-based unmixing with ADMM is particularly robust against highly noisy outlier bands.

Mots clés: machine learning; hyperspectral; geophysical image processing; hyperspectral imaging; learning (artificial intelligence); maximum entropy methods; remote sensing; correntropy maximization; ADMM; alternating direction method of multipliers; hyperspectral images; robust supervised spectral unmixing; fully constrained unmixing; cuprite mining image; optimization; robustness; kernel; noise measurement; convex functions; alternating direction method of multipliers (ADMM); correntropy; hyperspectral image; maximum correntropy estimation; unmixing problem

Domaines: Apprentissage [cs.LG]; Traitement du signal et de l'image [eess.SP]

Fei Zhu, Abderrahim Halimi, Paul Honeine, Badong Chen, Nanning Zheng. Correntropy Maximization via ADMM - Application to Robust Hyperspectral Unmixing. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55 (9), pp.1-12. ⟨hal-01965043⟩