Automated Remote Sensing Image Classification Method Based on FCM and SVM

An automated remote sensing image classification method combining FCM(Fuzzy c-Means) clustering algorithm with SVMs(Support Vector Machines) is proposed. The proposed new method aims to resolve the problem that training samples need to be chosen manually when used supervised classification method such as SVM, and compared with unsupervised classification method, it has higher classification accuracy. In the working flow of the new method, FCM algorithm was used to clustering original data firstly, and then according to the membership matrix of every pixel with each class and the size of each clustered region, some mixed pixel as labeled samples were chosen to train SVM classifier. The experimental results shown that the proposed method has the higher efficiencies and accuracies in the classification of Landsat TM data.