Feature extraction from microwave data using backscatter coefficient

摘要

Earth has plethora of geographical features, each having different surface properties. These different properties are reflected in the images captured in various frequencies by several satellites. Synthetic Aperture Radar captures such an image into microwave frequency. Such an image is inherited by speckle noise. This paper evaluates performance of traditional filters to remove speckle noise as a preprocessing stage from a microwave image and extract feature from it. Backscatter coefficient, a characteristic property of microwave data is chosen as the feature. The paper basically aims to identify smoother surfaces like water bodies, particularly river and its tributaries from a microwave image.