An empirical research of multi-classifier fusion methods and diversity measure in remote sensing classification

In this paper, Multi-Classifier System (MCS) is applied to the automatic classification of remote sensing images, and some effective multi-classifier fusion methods with relatively high accuracy are proposed based on substantive experiments. The classification accuracy of MCS has been remarkably improved compared to single classifier with an average increment of 5%. In addition, a diversity measure named EPD is presented, and the paper proves that its ability in predicting the performance of classifiers combining can be used to assist the construction of multiple classifier systems.