Urban Area Extraction Using X-Band Fully Polarimetric SAR Imagery

Abstract

Here, we present a method for extracting urban areas from X-band fully polarimetric synthetic aperture radar (SAR) data by reducing the effects of polarization orientation angle (POA). The proposed classifier performs two classifications utilizing characteristics of X-band scattering in land cover. One classification uses total power of scattering and volume scattering derived by using four-component decomposition methods with correction for the POA effect. The other classification uses polarimetric coherence between SHH} and SVV}. These two results are intersected and final urban areas are extracted after postclassification processing. We applied the proposed method to airborne X-band fully polarimetric SAR data of the polarimetric and interferometric airborne synthetic aperture radar system, developed by the National Institute of Information and Communications Technology, Japan. Validation of the results for three Japanese urban areas shows that the proposed method provides an acceptable overall accuracy of approximately 80%-90% at a 100-m spatial scale. It is also shown that texture-based classifiers using single polarimetric data have accuracy limitations when applied to extracted urban areas where the POA of objects is not uniform.