Enhancements for Persistent Scatterer Interferometry with High Resolution SAR

Kurzfassung

With the new class of meter-resolution SAR satellites like TerraSAR-X, Radarsat-2 or Cosmo SkyMed, new potentials and new challenges concerning the identification and motion estimation of PS arise. First inspections of TerraSAR-X interferograms show that the high resolution should lead to a markedly increase of PS in terms of spatial density. On the other hand, the resolution gives also rise to prominent scattering areas (instead of point scatterers) in numerous cases. In the presented study, we analyze the specific scattering effects with real TerraSAR-X images and propose adapted processing strategies to exploit the new class of data. Concerning the identification of persistent scattering areas, we developed a differential-geometric blob detection scheme which was inspired by similar techniques known from optical image sequence or video processing. Based on local contrast and shape criteria, this algorithm extracts compact image areas that remain approximately constant over a subset of images. In addition, numerous geometric and radiometric attributes are calculated from the extracted image primitives, which allow the final classification of PS areas. Not only the appearance of PS areas but the higher density of PS in general seems to call for adapted motion estimation algorithms. To better exploit these effects we employ models for coupled motions, i.e. dependencies between motions of neighbouring PS. It is in particular interesting to analyze the behaviour of the estimation accuracy based on the type of dependency and the amount of noise. We exemplify the potential of this approach by numerical quality assessment based on various simulations.