Multiple Classifier System for Remote Sensing Images Classification

A new multiple classifier system (MCS) is proposed to address the land cover classification problem with remote sensing images. The system considers both effectiveness and efficiency by combining the classifiers pruning and ensemble at the same time, which is realized by transferring these tasks to an optimization problem. Experimental results show that proposed MCS can successfully classify the remote sensing images with high accuracy as well as reduce the computation cost. Besides, the given system generally outperforms the individual classifiers and majority vote scheme applied on all classifiers on different datasets. According to the experiment results, we conclude that our MCS is a promising method for remote sensing images classification problem, especially when the feature is not sufficient.