Autoassociative Neural Networks for Features Reduction of Hyper-Spectral Data

Fabio Del Frate, Giorgio Antonino Licciardi, Riccardo Duca

Abstract

The potential of neural networks has been applied to hyper-spectral data, and exploited either for classification purposes and for feature extraction and dimensionality reduction. For this task a topology named auto-associative neural network has been used. The processing scheme uses a neural network architecture consisting of two stages: the first stage reduces the dimension of the input vector, while the second stage performs the mapping from reduced input into land cover classification.

Index Terms—Land cover. Classification. Hyperspectral data. Autoassociative neural networks. Feature reduction

First Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS, 26-28 August 2009, Grenoble, France

© Authors, 2009

.

.