Blind source separation applied to spectral unmixing: comparing different measures of nongaussianity

Authors:
Cesar F. Caiafa;Emanuele Salerno;Araceli N. Proto
Affiliations:
Laboratorio de Sistemas Complejos, Facultad de Ingeniería, UBA, Capital Federal, Argentina;Istituto di Scienza e Tecnologie dell'Informazione, CNR, Pisa, Italy;Laboratorio de Sistemas Complejos, Facultad de Ingeniería, UBA, Capital Federal, Argentina and Comisión de Investigaciones Científicas de la Prov. de Buenos Aires, Capital Federal, ...
Venue:
KES'07/WIRN'07 Proceedings of the 11th international conference, KES 2007 and XVII Italian workshop on neural networks conference on Knowledge-based intelligent information and engineering systems: Part III
Year:
2007

Citing 4
Cited 0

The Minimum Entropy and Cumulants Based Contrast Functions for Blind Source Extraction

IWANN '01 Proceedings of the 6th International Work-Conference on Artificial and Natural Neural Networks: Bio-inspired Applications of Connectionism-Part II
Separation of statistically dependent sources using an L2-distance non-Gaussianity measure

Signal Processing - Special section: Distributed source coding
Blind spectral unmixing by local maximization of non-Gaussianity

Signal Processing
Adaptive blind deconvolution of linear channels using Renyi's entropy with Parzen window estimation

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

We report some of our results of a particular blind source separation technique applied to spectral unmixing of remote-sensed hyperspectral images. Different nongaussianity measures are introduced in the learning procedure, and the results are compared to assess their relative efficiencies, with respect to both the output signal-to-interference ratio and the overall computational complexity. This study has been conducted on both simulated and real data sets, and the first results show that skewness is a powerful and unexpensive tool to extract the typical sources that characterize remote-sensed images.