Pixel-Based Classification of SAR Images Using Features

Abstract

The classification of synthetic aperture radar (SAR) images is an important task in the correct interpretation of the land-covers. In this study, the pixel-based classification was performed by extracting attributes from SAR images. Histogram of oriented gradients, local binary patterns, Gabor filters, morphological profiles and gray level co-occurrence matrix methods are used as feature extraction. Bayesian Gaussian mixture model was used as classifier. The tests carried on real SAR images show that the features extracted by morphological profile method give better results than the features extracted by the other methods.