Artificial neural network ensemble for land cover classification

Ensemble of artificial neural networks (ANN) often has better performance than any of the single learned ANN in the ensemble. And the combination of remote sensing and geographic ancillary data is believed to offer improved accuracy in land cover classification. However, conventional statistical classifier such as maximum-likelihood classifier (MLC) is not suitable to the ancillary data. In this paper, ANN ensemble is introduced to research on land cover classification. Experimental results show that ANN ensemble has good generalization ability. And classification with combination of remote sensing and geographic ancillary data outperforms single remote sensing data in terms of accuracy. Multisource land cover classification based on ANN ensemble could gain higher classification accuracy.