Mikrodalga ile görüntülemede nitel yöntemler

Microwave imaging (MWI) emerges as a novel technology that aims to extract physical properties of inaccessible objects from the scattered electric field measurements. MWI covers a very wide range of applications which includes but not limited to nondestructive testing (NDT), subsurface imaging, through wall imaging, biological imaging. The most important factor that tends the scientists to imaging with microwaves is non-ionizing nature of MWI when interacted with biological tissues. Therefore, MWI can be regarded as a healthy alternative of current imaging technologies, which are mostly based on ionizing radiation. Inverse scattering theory provides a group of highly theoretical approaches, known as qualitative method. These methods are based on inverting an integral equation for each point over a reconstruction domain to determine only the shape and the position of unknown scatterers without requiring any a-priori information. Two well-known representatives of qualitative inverse scattering methods are (i) linear sampling method (LSM) and (ii) factorization method (FM). These two methods are in fact quite similar in formulation as well as performance. These qualitative approaches are also usable in the MWI, where the inverse problem can be casted into two integral equations. Traditional solution approaches for MWI are based on non-linear or linear optimization methods. These methods recasts the nonlinear scattering problem in form of a minimization problem. Additionally, they can utilize Born approximation to linearize the problem and then the cost function is minimized via one of the canonical optimization approaches such as conjugate gradient method, newton's method. In other words, these methods model the physical scattering mechanism to determine electrical properties of dielectric objects and attempts to minimize a cost functional by using a canonical minimization procedure. In contrast to these approaches, the qualitative inverse scattering methods uses linearity of the scattering problem or the duality principle to reach their final aim, which is obtaining the support of the scatterer. Thus, their modest goals and linear nature make these qualitative approaches easier to implement and more efficient in use of computational resources. In contrast to all these attractive features, both LSM and FM are rarely investigated from an engineering perspective due to their mathematical background. Hence, analyzing these methods from an engineering perspective and making these methods applicable in real world imaging scenarios is an important problem. In the first part of the thesis, we analyze the problem of imaging buried targets under a rough surface for a two dimensional transverse magnetic scattering scenario. In fact, imaging of buried targets under the rough ground is a challenging inverse scattering problem with many applications in engineering such as land mine detection and remote sensing of archaeological artefacts. Conventional technology that uses microwaves for subsurface sensing of the underground is ground penetrating radar (GPR) which generates radargrams that require further interpretation by experts. Also, various quantitative inverse scattering methods are existent in the literature to provide additional information on the morphological and electrical properties of buried obstacles. Furthermore, the surface roughness, which is a critical factor that determines the limits of subsurface imaging, are only considered in a very few studies. In this context, qualitative inverse scattering methods, which are almost exclusively used for reconstructing the shape of inaccessible targets from the scattered field measurements, are particularly interesting since such methods can image multiple objects without requiring a-priori knowledge. Thus, we present a qualitative imaging method for subsurface sensing under a rough surface. The method relies on FM, where the aim is to retrieve the shape of unknown dielectric objects embedded inside a dielectric body whose closed boundary and electrical parameters are known a priori. Results show the stability and accuracy of the proposed method under very realistic conditions. In the second part of this thesis, the problem of using the qualitative methods in the real world measurement scenario is addressed. In the real world applications imaging systems extensively incorporate vector network analyzers (VNAs) instead of implementing additional modules to perform microwave measurements. This is particularly driven by availability of high performance VNAs. While there are alternative ways of reconstructing an image in MWI, imaging methods are naturally formulated in terms of scattered electric field vectors whereas measured scattering parameters (S-parameters) are only auxiliary quantities. Consequently, an intermediate step is required for experimental setups where measured S-parameters are mapped into scattered electric field. This is mostly handled by comparing simulated electric fields against measured S-parameters. As a better alternative to the canonical calibration procedures, we develop novel qualitative microwave imaging algorithms, which uses the measured S-parameters directly. Obtained experimental results prove the accuracy and the stability of the presented method. In the third and last part of the thesis, we consider the problem of using qualitative imaging methods in real world concealed target imaging scenarios. Real world concealed target detection can have different applications ranging from medical imaging to subsurface sensing, as mentioned above. Main challenge for such inverse problems is that the solution procedures are expected to capture the electrical parameters (relative dielectric constant 𝜖𝑟, conductivity 𝜎) of whole medium, which includes the buried objects. Up to date, many quantitative techniques are developed to obtain the complete electrical parameter distribution of a medium. However, if we take a glance at these formulations, we can see that they involve a considerable computational burden. Being contradictory to quantitative techniques, qualitative inversion methods, which aim to recover only the shape of the scatterers, have relatively simple formulations and require lower computational resources. In contrast to such obvious advantages, qualitative inversion techniques are rarely employed in buried obstacle detection, since these methods have strong a priori knowledge requirements in their original form. In particular, to be able to detect the shape of an inclusion by means of these methods, we must supply these two a priori pieces of information: (i) the dielectric parameters of the surrounding medium and (ii) the scattered field when there is no buried object inside the surrounding medium. It is obvious that fulfilling such strong conditions altogether is of a serious issue in any imaging problem. To this end, we propose a strategy to overcome the a priori knowledge requirement on the dielectric parameters of the surrounding medium. Results, which are obtained from real experiments performed in an anechoic chamber, confirm the accuracy and the stability of the proposed formulations.