Independent component discriminant analysis for hyperspectral image classification

Alberto Villa (1, 2) , Jon Atli Benediktsson (1) , Jocelyn Chanussot (3) , Christian Jutten (4)

1 University of Iceland [Reykjavik]
2 GIPSA-lab - Grenoble Images Parole Signal Automatique
3 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique
4 GIPSA-VIBS - GIPSA - Vision and Brain Signal Processing

In this paper, the use of Independent Component Discriminant Analysis (ICDA) for remote sensing classification is proposed. ICDA is a non-parametric method for discriminant analysis based on the application of a Bayesian classification rule on a signal composed by independent components. The method is based on the use of Independent Component Analysis (ICA) to choose a transform matrix so that the transformed components are as independent as possible. Then, a non parametric estimation of the density function is computed for each independent component. Finally, the Bayes rule is applied for classification assignment. The obtained results are compared with one of the most used classifier of hyperspectral images (Support Vector Machine) and show the comparative effectiveness of the proposed method.

Mots clés: Hyperspectral data; Independent Component Analysis; Bayesian Classification

Domaines: Traitement des images [eess.IV]

Alberto Villa, Jon Atli Benediktsson, Jocelyn Chanussot, Christian Jutten. Independent component discriminant analysis for hyperspectral image classification. WHISPERS 2010 - 2nd Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, Jun 2010, Reykjavik, Iceland. conference proceedings. ⟨hal-00578911⟩