Analysis of the Statistical Correlation in the Context of Multi-Look SAR Tomography

Synthetic Aperture Radar Tomography successfully exploits the coherent processing principle, allowing for a precise long-term monitoring of the analyzed scenes. A challenging aspect is represented by the detection of the points which offer reliable estimates of the scene's parameters. Multi-looking processing-averaging of multiple points contributions-has proved to significantly improve the performances of the detection process. The derivation of the likelihood ratios exploited for the identification of stable targets implies the statistical modelling of the averaged looks. The analysis conducted in this work on a real dataset of acquisitions confirms that the assumed uncorrelated model of the data vectors is the most appropriate one.