LP norm SAR tomography by iteratively reweighted least square: First results

Synthetic aperture radar tomography (TomoSAR) estimates the scene reflectivity along range, azimuth and elevation directions. Even if many works in recent literature deal with this topic, TomoSAR imaging remains a not easy procedure. In this work, the possibility to improve quality of imaging by a priori information is investigated experimentally; in particular we focus on urban scenario where targets of interest are point-like and radiometrically strong. Accordingly, we look for a sparse reflectivity function; this can be obtained minimizing the solution in an arbitrary L_p norm using the It-eratively Reweighted Least Square (IRLS) algorithm. Based on an experimental comparison among different choices for p, the conclusion drawn is that the usual choice p = 1 is the best trade-off between resolution and robustness to noise. Therefore, L₁ norm minimization by IRLS has been exploited to perform CS TomoSAR on real data, and we report in this paper first results obtained using COSMO-SKyMed data acquired over an area in Milan, Italy.