A new approach for three-dimensional building extraction in high-resolution monoscopic SAR imagery

A novel approach is presented in this paper for three-dimensional building extraction from a single high-resolution amplitude SAR image. The proposed method, which addresses both building detection and building parameter estimation, relies on a maximum likelihood model inversion, assuming a parallelepipedic 3D building model. The performance of this new building extraction approach, which is validated on simulated data, is demonstrated on real SAR data over a semi-urban area.