Grouping of Persistent Scatterers in high-resolution SAR data of urban scenes

Persistent Scatterer Interferometry (PSI) is a technique to simultaneously estimate surface deformation and 3D structure from stacks of SAR images. It was proposed first about one decade ago to monitor preferably urban areas, where in general the highest numbers of PS are found. At that time no high-resolution satellite SAR data were available. Instead, for example, stacks of ERS imagery were used providing ground range resolution of about 25 m. In data of such kind only the strongest PS can be detected, which are usually caused by corner reflectors built by orthogonal building and road planes of considerable size, whereas smaller structures causing weaker ones signal are averaged by clutter or mutually interfere with others in the same resolution cell. Thus, if any, only a few or even just one single PS are found per building. The advent of a new senor generation of systems like TerraSAR-X and COSMO-Skymed in 2007 led to a significant improvement of spatial resolution of about one order of magnitude. This comes along with a dramatic rise of PS density: In some cases tens to hundreds are detected at large buildings, which offers the possibility to monitor even individual urban objects.In addition, especially at building façades the distribution of those PS is often quite regular. A reason for that is the usually rectilinear arrangement of façade structures inducing PS like windows or balconies. Those patterns contain a lot of information about the objects under investigation, which is mostly ignored in current PSI processing schemes. For example, consider a regular structure of windows on a certain façade of a multi-story building. Assuming the same kind of structure generates one single PS at each window, the phase centers of all scatterers caused by windows of each floor share the same height. This means, we may benefit from such kind of redundancy, for instance, to improve the height estimate by averaging over PS having the same elevation.In this work, we first discuss the regular appearance of PS at urban façades for an urban test scene in TerraSAR-X spotlight mode data. Then, we show how PS analysis could benefit by exploitation of the redundancy due to repetitive patterns of man-made objects. Finally, we propose a PS grouping scheme based on a production system and discuss first results achieved for the test area.