Ensemble Learning Based Classification of High Resolution SAR Images

Abstract

The improvements in SAR systems having a very wide range of applications lead to obtain higher resolution images and consequently to increase the need for classification of these images. In this paper, approximate sparse multinomial logistic regression (ASMLR) based bagging method as a ensemble learning method is proposed for classification of SAR images. Based on the classification of real SAR images, it is shown that obtained better classification results with proposed method compared to widely used classifiers.