Unsupervised classification based on fuzzy c-means with uncertainty analysis

Abstract

Fuzzy c-means (FCM) is a widely used unsupervised classifier for remote sensing images. This letter presents an uncertainty analysis-based FCM (UAFCM) classification method. The uncertainty in this letter refers to the discriminative ability of class attributes in fuzzy classification on a per-pixel basis. UAFCM is performed by analysing the uncertainty in FCM classification result and reclassifying the pixels with large uncertainty. Specifically, the uncertainty in FCM classification is measured by entropy and a proposed square error-based criterion. A threshold value is then determined to recognize the pixels with large uncertainty, which are reclassified with spatial connectivity subsequently. Experiments on three remote sensing images show that the proposed UAFCM consistently obtains more accurate classification results than does FCM, and hence provides an effective new unsupervised classification method for remote sensing images.

Acknowledgements

This work was supported by Research Grants Council, Hong Kong (PolyU 5249/12E) and the Hong Kong Polytechnic University (Project No.: 1-ZV4F, 1-ZVBA, G-U753, G-UA35, G-YK75, G-YJ75, G-YZ26, and H-ZG77). The authors would like to thank Professor D. Landgrebe, Purdue University, West Lafayette, IN, USA, for providing the FLC1 and HYDICE data set, Professor Paolo Gamba of the University of Pavia for providing the ROSIS data, and Dr. Lefei Zhang of the Wuhan University for providing the ground truth of the HYDICE data set. The reviewers’ valuable suggestions and comments have greatly improved this manuscript.