Innovative Alphabet‐Based Bayesian Compressive Sensing Technique for Imaging Targets with Arbitrary Shape

Anselmi, N. and Oliveri, G. and Hannan, M. and Salucci, M. and Massa, A. (2017) Innovative Alphabet‐Based Bayesian Compressive Sensing Technique for Imaging Targets with Arbitrary Shape. Technical Report. ELEDIA Research Center - University of Trento.

Text Innovative_Alphabet‐Based_Bayesian_Compressive_Sensing_Technique_for_Imaging_Targets_with_Arbitrary_Shape.v2.pdf Download (4MB)

Abstract

In this work an innovative two-dimensional (2D) microwave imaging technique exploiting Bayesian Compressive Sensing (BCS) and a wavelet-based alphabet for representing the problem unknowns is dealt with. The proposed approach is based on the generalization of the sparsity concept, extending the range of applicability of BCS-based inverse scattering (IS) techniques to objects with arbitrary shape and dimensions. A set of BCS reconstructions is performed considering different expansion bases in the alphabet, without the need for a-priori knowledge about the unknown scatterers. Then, the best reconstruction is recognized as that minimizing the number of non-null retrieved coefficients (i.e., the sparsest one). In order to verify the effectiveness of the proposed imaging technique, a set of representative numerical benchmarks is presented. Some comparisons with state-of-the-art IS techniques are presented, as well.

Actions (login required)

View Item