ANN Classification of OMIS Hyperspectral Remotely Sensed Imagery: Experiments and Analysis

Abstract

In order to experiment the performance of some popular ANN algorithms to OMIS (Operational Modular Imaging Spectrometer) hyperspectral image, three widely used ANNs, including Back Propagation Neural Network (BPNN), Radial Basis Function Neural Network (RBFNN), Fuzzy ARTMAP network and their improvements, are employed and compared. It is concluded that ANN classifiers perform much better than traditional classifiers such as SAM, MLC and MDC, and RBFNN outperforms BPNN and Fuzzy ARTMAP in terms of classification accuracy. It is also concluded that dimensionality reduction by PCA can be effectively used to feature extraction for hyperspectral image classification.

5 Figures and Tables

Cite this paper

@article{Du2008ANNCO, title={ANN Classification of OMIS Hyperspectral Remotely Sensed Imagery: Experiments and Analysis}, author={Peijun Du and Kun Tan and Wei Zhang and Zhigang Yan}, journal={2008 Congress on Image and Signal Processing}, year={2008}, volume={4}, pages={692-696} }