Global urban mapping using building density from polarimetric SAR images with POA correction

In this paper, an algorithm for urban mapping using building density from polarimetric synthetic aperture radar (PolSAR) images is proposed. Urban density estimation method applied in this study utilizes normalized combination of volume scattering power (P_v) and helix scattering power (P_c) derived by four-component decomposition method. However, this estimation method compresses the range of the scattering intensity between 0 and 1 because of normalization which is a part of POA correction process. The estimated index shows relative urban density which represents relative height of density within an image. Even if the values of estimated density are the same between different areas, they don´t always indicate the same level of actual density. Thus, in comparison of urban density between different cities, absolute scale of urban density corresponding to actual building density of each area is indispensable. The main part of this paper is estimating absolute urban density using polarimetric indices. In this study, T11, T22, and T33 which are the diagonal elements of the coherency matrix are used. Each of them represents different scattering process. Finally, urban mapping using the estimated absolute density over various cities is conducted.