Urban Characterization Based On Recent Incoherent Polarimetric Decompositions

This paper intends to characterize the urban area from PolSAR data using the incoherent Polarimetric Decomposition (PD) methods. These methods contribute intensively in interpreting the physical mechanisms of different land use and land covers. In this work, a comparison study is conducted using the existing PD methods in order to find which one has the capability to characterize better urban areas most accurately, regardless of how they are oriented relative to the radar line of sight (LOS). For this purpose, this study employed the three-component decomposition approach (FDD), the four component decomposition (Y4O), both without and with rotation of the coherency matrix, the six component decomposition (6SD), and the seven component decomposition (7SD). In order to assess the results, a PolSAR image acquired in L-band by RADARSAT-2 satellite over the San Francisco has been used.