%A L. Poli
%A G. Oliveri
%A A. Massa
%I University of Trento
%D 2015
%X This report is aimed to show a comparison between the probabilistic inversion methods based on single-task and multi-task Compressive Sensing (CS) strategies recast in a Bayesian framework and a deterministic technique of the state-of-the-art (a conjugate gradient-based method). The results show in particular the better capabilities of the multi-task CS method to take advantage of multi-frequency data when dealing with small-size scatterer in inverse scattering problems. The efficiency and robustness of such a method is validated considering different sparse-scatterer scenarios and different value of signal-to-noise ratio on the data.
%K Compressive Sensing, Inverse Scattering, Interval Analysis
%L elediasc12668
%T Performance comparison between multi-frequency deterministic and probabilistic approaches