With the development of high resolution (HR) synthetic aperture radar (SAR) imaging

the technology of building 3D information extraction from SAR images is becoming an important issue in radar remote sensing.In order to retrieve building height information from single HR SAR image

a model-based height estimation approach is proposed.An orthographic projection model is established to achieve the quick generation of the model hypothesis.A segmentation-based likelihood function is derived to examine the consistency between the model hypothesis and the actual SAR image.The hypothesis test procedure

which is iterated with the simulated annealing algorithm

is designed to optimize the hypothesis from the high dimensional model space.The experimental results using both of the simulated and actual airborne SAR images show the effectiveness of the proposed method.