Target Identification from High Resolution Remote Sensing Image by Combining Multiple Classifiers
Target identification from high resolution remote sensing image is a common task for many applications. In order to improve the performance of target identification, multiple classifier combination is used to a QuickBird high resolution image, and some key techniques including selection and design of member classifier, classifier combination algorithm and target identification methods are investigated. After constructing a classifier ensemble composed of five members: maximum likelihood classifier (MLC), minimum distance classifier (MDC), Mahalanobis distance classifier (MHA), decision tree classifier (DTC) and support vector machine (SVM), double fault measure is used to select three classifiers for further combination. MLC, DTC and MHA are selected, and their independence and diversity are evaluated. Different classifier combination strategies are experimented to extract sports field and buildings from QuickBird image. The results show that multiple classifier combination can improve the performance of image classification and target identification, and the accuracy is affected by many factors.
Keywordsmultiple classifier combination high resolution remote sensing target identification hierarchical classifier system classifier selection
Unable to display preview. Download preview PDF.
- 1.Shi, W.Z., Zhu, C.Q.: Road Feature Extraction from Remotely Sensed Image: Review and Prospects. Acta Geodaetica Et Cartographica Sinica 30(3), 257–361 (2001)Google Scholar
- 3.Bo, Y.C., Wang, J.F.: Combining Multiple Classifiers for Thematic Classification of Remotely Sensed Data. Journal of Remote Sensing (5), 555–564 (2005)Google Scholar
- 7.Xie, Z.X., Yu, D.R., Hu, Q.H.: Study on sensor fault tolerance with multi-classifiers of SVM fusion. Journal of Harbin Engineering University 27(supp.), 389–293 (2006)Google Scholar
- 8.Lv, Y., Shi, P.F., Zhao, Y.M.: Voting Principle for Combination of Multiple Classifiers. Journal of Shanghai Jiao Tong University 34(5), 680–684 (2000)Google Scholar
- 10.Lin, H., Li, J.P., Mo, D.K.: Information Identification on QuickBird Image. Journal of Image and Graphics 10(10), 1504–1511 (2005)Google Scholar
- 11.Mou, F.Y., Zhu, B.Q., He, H.Z.: Multiple Features Based Analysis of High-resolution Remotely Sensed Imager. Bulletin of Surveying and Mapping 10, 4–7 (2004)Google Scholar
- 12.Kang, H.J., Doermann, D.: Selection of Classifiers for the Construction of Multiple Classifier Systems. In: Proceedings of the Eight International Conference on Document Analysis and Recognition, pp. 263–268 (2005)Google Scholar
- 13.Liu, C.P., Dai, J.F., Zhong, W., et al.: Multi-source Remote Sensing Information Based on Fuzzy Evidence Theory. Pattern Recognition and Artificial Intelligence 6, 213–218 (2003)Google Scholar
- 14.Congahon, R.G., Green, K.: Assessing the Accuracy of Remotely Sensed Data: Principles and Practices. Lewis Publishers (1999)Google Scholar
- 15.Kuncheva, L.I., Whitaker, C.J., Shipp, C.A.: Is Independence Good for Combining Classifiers? In: Proceedings of ICPR 2000, vol. 2, pp. 168–171 (2000)Google Scholar
- 19.Giancinto, G., Roli, F.: Design of Effective Neural Network Ensembles for Image Classification Purposes. Image Vision Computer (19), 697–705 (2001)Google Scholar