Exploring Self-learning for spatial-spectral classification of remote sensing images

The process of acquiring training samples in the area of remote sensing is expensive and also time consuming. The Semi-supervised classification technique has been explored to address the problems involving classification using limited labeled data. Self learning with support vector machine has been popularly used for remote sensing image data classification. However, as per the studies the Self-learning with support vector machine algorithm has not been able to achieve good accuracy. In this paper, semi-supervised fuzzy c-means (SFCM) is used with support vector machine (SVM) to improve the accuracy of self-learning. Spatial information is also integrated by applying probability filtering methods. The experimental results on two publically available images with labeled pixels exhibit that the classification accuracy of remotely sensed images has been improved by using SFCM with SVM in Self-learning semi-supervised framework.