Quantitative Near-Field Microwave Holography

Abstract

This thesis presents two quantitative holographic reconstruction techniques for the imaging of dielectric targets. The first method is a quasi-real-time holographic reconstruction technique, which is capable of imposing physically based constraints on the real and imaginary parts of the permittivity. The other method is a real-time holographic reconstruction technique that is faster than the constrained method but cannot accommodate constraints on the reconstructed permittivity in its current form. The goal of this thesis is to introduce both methods and recommend which is best. Microwave holography has been used by our research group to reconstruct images of a target’s shape and location from microwave scattering parameters. This thesis will demonstrate that holography can be extended to quantify the permittivity distribution in a region of interest. The problems presented in this thesis are generic and are meant to show that near-field quantitative holography is a valid approach for applications such as tissue imaging, baggage inspection, concealed weapon detection, etc. The holographic inversion is carried out in the spectral domain (Fourier space), which allows for the use of Fourier transform properties to expedite the algorithm. This differs from sensitivity-based imaging (another inversion method developed by Tu et al. (2015)) where the inversion is performed in real space and is unable to take advantage of the techniques proposed in this thesis to improve the speed of reconstruction. Mutual coupling is not taken into consideration in the forward model of scattering used here; however, this technique is meant to be viewed as a foundation for a more sophisticated reconstruction algorithm, like the iterative Born method, which can overcome such limitations. Iterative reconstruction methods require an accurate initial guess, which can be provided by the quantitative technique presented in this thesis. Moreover, this technique, implementing fast and efficient linearized inversion, can serve as a module, which is called repetitively by the iterative algorithm. Such a module will take the current estimate of the total field distribution inside the imaged volume as an input and will return an estimate of complex permittivity distribution.