Information-Theoretic Detection for Persitent Scatterers using COSMO-SkyMed SAR Data

The objective of this thesis is to derive a simple and effective detection scheme that can improve the capability of detecting multiple point-like targets, such as Persistent Scatterers (PSs), in the SAR tomography context. PSs techniques have proven to be powerful tools in monitoring man-made structures, especially buildings, with possible slow temporal deforma- tions. However, one of the main problem related to PSs techniques is the lack of high number of points within a single tomographic cell. The analysis of this problem can be done using a multiple hypothesis test to detect the presence of feasible multiple scatterers [1]. Although it is proven that the detection of multiple PS, in urban scenarios, is possible through several techniques [1], [2], the identification of more than two PSs in real scenarios still remains challenging from the point of view of the complexity aspect related to the used architecture scheme. In this thesis, this problem is framed in the context of the information theory and exploit the theoretical tool, developed in [3], to design of a one-stage adaptive architecture for multiple hypothesis testing problems for the PS detection in the context of SAR tomography. Moreover, we resort to compressive sensing approach for the estimation of the unknown parameters under each hypothesis. This architecture has been tested on both simulated as well as real recorded data. In addition, for the single and double scatterer case, this model has been tested in comparison with suitable counterparts on simulated data only. Finally, the early stage of this work has been discussed in [4], while the analytical model and the experimental discussion of this thesis are the subject of a submission to IEEE Transactions on Geoscience and Remote Sensing.