Semi-Supervised FCM and SVM in Co-Training Framework for the Classification of Hyperspectral Images

摘要

Collection of labeled samples is very hard, time-taking and costly for the Remote sensing community. Hyperpectral image classification faces various problems due to availability of few numbers of labeled samples. In the recent years, semi-supervised classification methods are used in many ways to solve the problem of labeled samples for the hyperspectral image classification. In this Article, semi supervised fuzzy c-means (FCM) and support vector machine (SVM) are used in co-training framework for the hyperspectral image classification. The proposed technique assumes the spectral bands as first view and extracted spatial features as second view for the co-training process. The experiments have been performed on hyperspectral image data set show that proposed technique is effective than traditional co-training technique.