TY  - RPRT
KW  - Inverse Scattering (IS)
KW  -  Bayesian Compressive Sensing (BCS)
KW  -  Microwave Imaging
KW  -  First Order Born Approximation
PB  - ELEDIA Research Center - University of Trento
AV  - public
ID  - elediasc12741
A1  - Anselmi, N.
A1  - Oliveri, G.
A1  - Hannan, M.
A1  - Salucci, M.
A1  - Massa, A.
N2  - A key requirement to be satisfied when exploiting Compressive Sensing (CS) methods in inverse scattering (IS) problems is that the unknowns (e.g., the contrast function or the equivalent sources) are sparse with respect to the considered expansion basis. State-of-the-art CS-based microwave imaging techniques typically consider single-resolution pixel-based representations, limiting their domain of applicability to the retrieval of few and isolated pixels within the investigated domain. Within this framework, this work is aimed at extending the range of applicability of CS-based approaches to the retrieval of unknown scatterers having arbitrary shape and dimensions. Since in real applications no a-priori information about the investigation domain is available, the idea is to retrieve a set of "candidate" solutions by executing several CS inversions using different expansion bases (e.g., pixel, Haar wavelets, Meyer wavelets, ...). Following the CS paradigm, the "best" solution can then be identified as the sparsest one, i.e., the solution with the lowest number of non-zero retrieved coefficients. A preliminary numerical validation of the proposed alphabet-based CS microwave imaging technique is given. Some numerical comparisons with competitive state-of-the-art inverse scattering techniques is shown, as well.
M1  - technical_report
UR  - http://www.eledia.org/students-reports/741/
Y1  - 2016///
TI  - Free?Space Microwave Imaging through Alphabet?Based Bayesian Compressive Sensing
ER  -