Multi-task Bayesian Compressive Sensing for microwave imaging exploiting the minimum-norm current formulation

Poli, L. and Oliveri, G. and Viani, F. and Massa, A. (2014) Multi-task Bayesian Compressive Sensing for microwave imaging exploiting the minimum-norm current formulation. Technical Report. University of Trento.

Preview

Text Multi-task Bayesian Compressive Sensing for microwave imaging exploiting the minimum-norm current formulation.pdf Download (18MB) | Preview

Abstract

In this report, an innovative three-step contrast-source probabilistic technique is proposed for the reconstruction of image 2D-sparse dielectric profiles. Within the formulation of the inverse scattering problem, such an approach combines (i) a SVD-based step to retrieve the minimum-norm currents, (ii) a probabilistic reformulation of the inverse scattering problem in terms of the real and imaginary parts of the sparse contrast currents (iii) a multi-task BCS strategy for properly correlating the unknown variables (real and imaginary parts of contrast source coefficients). An enhanced version of the multi-task implementation that takes into account the correlations real and imaginary parts of contrast source coefficients related to different views is also investigated through a wide set of numerical experiments.

Actions (login required)

View Item