A Multi-Classifier Combined Decision Tree Hierarchical Classification Method

To improve the accuracy of remote-sensing image classification, proposed a multiclassifier combined decision tree hierarchical classification method. First we using decision tree algorithm to generate an initial decision tree (IDT), and then edit the rules of IDT, connect the IDT with multi-classifier to generate a hybrid decision tree (HDT). During the classification procedure, land categories are obtained seriatim according to the HDT, subsequently the corresponding region of new-obtained category will be masked out, so that the classification process becomes more and more easy. This method can avoid the cross interference when various categories classify in the same time. Accuracy analysis of experimental results shows that compared with traditional classification methods, this method can greatly improve the accuracy of classification.