Multiple‐component polarimetric decomposition with new volume scattering models for PolSAR urban areas

Abstract

Multiple-component model-based decompositions (MCSMs) of polarimetric synthetic aperture radar (PolSAR) data often exhibit overestimation of volume scattering power, which makes the oriented built-up areas show volume scattering rather than double-bounce scattering. Deorientation processing has been incorporated into the three- and four-component decomposition algorithms to overcome this limitation, where the coherency matrix is rotated to minimise the cross-polarised term. However, even with the deorientation, some urban areas with large orientation angles are still misjudged as vegetation. In this study, the performance of deorientation processing on the MCSM is discussed and then an improved polarimetric model-based decomposition method for PolSAR urban areas is proposed, which is inspired by Sato's decomposition method. Since the cross-polarised HV scattering component is caused not only by vegetation but also by oriented buildings, the volume scattering model of original multiple-component decomposition is extended to describe the HV scattering from these two different land covers. A general volume scattering model is adopted to describe the HV scattering from vegetated areas while the orientation angle of built-up areas is adaptively considered for modelling the HV scattering from oriented buildings. Experiments with the phased array type L-band synthetic aperture radar data demonstrate that the authors’ proposed method can get better decomposition results over urban areas than other methods.