Work Package 1 (WP1)
Coherent scatterer characterization for displacement monitoring
Person in charge : Jean-Marie Nicolas
Florence Tupin, LTCI-Télécom ParisTech
Marie-Pierre Doin, LG-ENS
SAR images are based on the interaction between eletromagnetic waves and Earth ground. It is why it is very important to understand the nature of this interaction. In order to simplify this short description, there are two main cases :
- A unique target can be viewed inside the pixel. In this case, the value of the pixel depends only on the RCS (Radar Cross Section) of this target. If this target can be viewed at different acquisition dates, and if its response is stable, it will be called "coherent scatterer" (CS).
- A lot of targets can be viewed inside the pixel. In this case, the pixel is the result of a coherent summation called "speckle".
The nature of Coherent Scatterers is at present time partially understood. For example, a corner reflector can be viewed as a perfect CS. At the same time, a large rock on a glacier is not a CS as the ice melts under this rock so that its adjusting varies between two radar acquisitions. To cope with this lack of understanding of the wave interaction, we propose image processing tools in order to find CS candidates inside a stack of images and to experimentally understand the time variations of these CS candidates, both in amplitude and in phase. We aim to develop two specific tools :
- MAGNICOS ("Magnifying Coherent Scatterers"),This interactive tool will be designed to detect and track coherent scatterers at different levels:
- between two images acquired in interferometric mode, with moderate temporal baselines : CS identified in both images will be tracked and matched using the acquisition configurations. The output of this stage will consist in accurately co-registered interferometric information over the CS candidates.
- between data sets acquired at different dates and with potentially different trajectories: a technique, looking for CS that "survived" the important time-lag between acquisitions, will be developed, based on geometrical processing and average phase behaviors over stable reference points.
This tool requires accurate resampling of the data. As these data are complex ones (a real part and an imaginary part), resampling has to cope with specific problems as phase vortex so that a theoretical step devoted to these mathematical objects is required.
- SYTER-HP to process RAW data in a common geometric representation. Based on the SYTER concept (SYTER is a SAR procesor in time domain developed at LTCI), it will be devoted to CSK data.