Through-the-wall microwave imaging with sampling methods

Microwave imaging is a promising technology and basically aims to retrieve electrical parameters such as permittivity and conductivity of the unapproachable objects with their shape and location. Microwave imaging systems extract information of materials by processing their scattered electric field data or scattering parameters. Due to non ionized characteristic of microwaves, they harmlessly penetrate to human body parts. This feature allows microwaves to find numerous applications in medical imaging. Alongside medical applications, microwave imaging applications also include through the wall imaging, subsurface imaging, non destructive testing and concealed weapon detection. The problem in microwave imaging separates into two groups direct and inverse problems. In electromagnetics, in direct problem the scattered field is calculated by knowing features of scatterers. Mostly the scattered field cannot be calculated analytically, numerical methods such as physical optics, modal techniques or integral equation methods must be performed for this purpose. On the other hand, in inverse problems, characteristics of scatterers are unknown and these parameters are reconstructed by set of measured scattered field outside the investigated domain. Countless inversion methods which are split into two parts, namely qualitative and quantitative methods, are developed for performing inversion. Qualitative methods include but not limited to linear sampling method, factorization method, orthogonality sampling method and singular sources method. The methods in that category aim to reconstruct shape and location of scatterers very fast with respect to other counterparts. Despite qualitative methods, quantitative methods retrieve electrical parameters of objects in investigated domain by minimizing the cost function. The most well known quantitative methods are Born iterative method, distorted Born iterative method and contrast source inversion method. The initial value of contrast must be find for these methods, Born approximation and back propagation are applied for this purpose. The thesis under study considers through the wall microwave imaging application. In the first part of the thesis, steps to solve direct problem are introduced. The derivation of Green's function of the wall is developed analytically for homogeneous planar layered media. The wall considered as infinitely long and rectangular solid at two axes. The well established method of moments is used to calculate scattered electric field at a straight line. The illumination is done by Hertzian electric dipoles in front of the wall and multi frequency case is considered. Secondly, two of qualitative inversion algorithms, namely linear sampling and factorization methods are presented with their explicit formulations. These methods are highly ill posed and regularization techniques must be implemented for finding approximate solution of real problem. Linear sampling and factorization methods are applied to different scenarios. Perfect electric conductor and dielectric scatterers are involved in simulation setups. The reconstruction process is performed by using proper Green's function. The comparison of linear sampling and factorization methods can be seen from numerical results and also their achievements in through the wall imaging can be clearly commented.