Urban-area extraction from polarimetric SAR image using combination of target decomposition and orientation angle

摘要

The results of model-based target decomposition are the main features used to discriminate urban and non-urban area in polarimetric synthetic aperture radar (PolSAR) application. Traditional urban-area extraction methods based on model-based target decomposition usually misclassified ground-trunk structure as urban-area or misclassified rotated urban-area as forest. This paper introduces another feature named orientation angle to improve urban-area extraction scheme for the accurate mapping in urban by PolSAR image. The proposed method takes randomness of orientation angle into account for restriction of urban area first and, subsequently, implements rotation angle to improve results that oriented urban areas are recognized as double-bounce objects from volume scattering. ESAR L-band PolSAR data of the Oberpfaffenhofen Test Site Area was used to validate the proposed algorithm.