A Novel Hyperspectral Classification Method Based on C5.0 Decision Tree of Multiple Combined Classifiers

It is difficult for a single classifier to resolve the problem of high dimension in the hyperspectral image classification applications. Combination of multiple classifiers can make full use of the complementary of the existing classifiers, thus owns better classification performance. A novel multiple classifiers based on C5.0 decision tree is proposed. It reduces the hyperspectral dimension through wavelet-PCA transform algorithm firstly. Then three supervised classifiers, namely Minimum Distance, Maximum Likelihood and SVM, combined by C5.0 decision tree, are used to realize hyperspectral classification. Experiments based on AVIRIS hyperspectral image data show that higher classification accuracy may be achieved via the multiple combined classifiers than a single sub-classifier. The proposed method can reduce the dimension of features and improve the classification performance efficiently.

[1]  Farid Melgani,et al.  Toward an Optimal SVM Classification System for Hyperspectral Remote Sensing Images , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[2]  Gaurav Jain,et al.  An approach to text classification using dimensionality reduction and combination of classifiers , 2004, Proceedings of the 2004 IEEE International Conference on Information Reuse and Integration, 2004. IRI 2004..

[3]  Behnaz Bigdeli,et al.  A Multiple Classifier System for Classification of LIDAR Remote Sensing Data Using Multi-class SVM , 2010, MCS.

[4]  Chen-guang Dai,et al.  Support Vector Machine for Classification of Hyperspectral Remote Sensing Imagery , 2007, Fourth International Conference on Fuzzy Systems and Knowledge Discovery (FSKD 2007).