Topographic mapping: a new approach

A new scheme for reconstructing scene topography using both interferometric phase and backscattering intensity is presented. The authors use the framework of Bayesian theory and look for the scene topography which best explains the observed data. They iteratively improve the estimation changing the topography and simulate, in the Lambertian approximation, a SAR image to be compared to the real one. The fidelity to the observed interferogram is expressed in terms of the phase-difference between the simulated and the observed phase. The regularity of the scene topography is expressed as a Gibbs random field. This allows the authors to obtain an energy to be minimized. The solution of this optimization problem is obtained by a relaxation method