Phase unwrapping for DEM generation as an inverse problem

A traditional way to add short wavelength information on an existing Digital Elevation Model (DEM) is to produce a geocoded residue interferogram, i.e. removing the synthetic fringes calculated from the input DEM. The phase unwrapped product is then simply added to the original DEM in its geographic geometry. Our different approach consists in four steps: (1) we first build a topographic interferogram in the Synthetic Aperture Radar (SAR) natural geometry. (2) We then compute the SAR orthorectification consisting in calculating independently each pixel location on the ellipsoid (latitude, longitude and elevation). In this step, we use an iterative search algorithm to retrieve the correct elevation within an SRTM DEM. (3) The elevation matrix is used as a priori information in the least-squares phase unwrapping which is performed in the SAR geometry. (4) At last, the geometry matrices are used to resample the elevations in a cartographic reference system. The outcome forms the fusion of the interferogram and the information available in the DEM and sparse GPS measures. This approach is a classical least-squares inversion with a priori information. We finally discuss results over the Athens-Corinth region (Greece) comparing them with a high resolution SPOT DEM.