Quantitative holographic microwave imaging with a simulated calibration measurement

Abstract

Microwave imaging is a new technology that has recently been introduced into use. This imaging technology can be applicable in various fields, including medical examination and tumor detection, material characterization, foreign object detection, security applications, and many other areas, owing to their effectiveness in the near field and their non-ionizing properties. Microwave imaging is fundamentally based on solving the inverse scattering problem and numerous techniques for solution have been proposed and studied in the literature. Due to the ill-posed nature of the problem, the non-unique solution is the main issue and restricts our ability to find the exact parameters of objects in the results. Therefore, these techniques can be divided into two main categories based on the results obtained: qualitative and quantitative. While qualitative results provide only the contrast value of an object due to limited system definitions and prioritization of faster results, quantitative methods yield the actual value of the object by sacrificing the calculation speed and generalization in imaging calculations. The Holographic Imaging technique is one of the promising methods that offers quantitative and faster results but in return, it requires precise calibration procedures customized to the environment. In this thesis, the recently proposed Quantitative Microwave Holography (QMH) method is applied in dielectric imaging and the technique has improved by means of its calibration time drawback. The general QMH algorithm is more generalizable to object shapes compared to other quantitative techniques but not the object's actual values, and it needs to calibration to both environment and possible dielectric parameters space to get attainable results. This technique gives 3D quantitative results by basically using system response or system function obtained from the calibration procedure and using spectral reconstruction technique with the aid of linearization and holography. It is possible by inverting the convolution of this system function and the contrast function in the spectral domain after applying linearization to the famous non-linear scattering equation. The transfer function of the system is obtained by finding the scattering response of the point scatterer (PS) in each slant range position and this process is applied in each range space for the regular QMH method. Once the system response is determined, the object under test (OUT) response is obtained and the object function is resolved using regularization methods, such as Tiknhonov regularization, from the linear system where these two scattering data are related in the spectral domain by a product. The linear system that is used in the resolving step, is obtained using linearization approximations such as Born and Rytov. The Born and Rytov approximations are general methods that assume there are no internal scatterings or there are constant phase characteristics in objects respectively. Due to these assumptions, the object being imaged or CO being used has to be of limited size or have limited contrast values where these two requirements are drawbacks of QMH algorithm. All the scattering data used in reconstruction is obtained by subtracting the background information according to the Holography technique to obtain the difference data as in optical holography and this is the reason why we called this imaging method holographic imaging. In our proposed method, this time-consuming calibration procedure in the real world is replaced by a simulation environment. With this approach calibration procedure will be accelerated and some drawbacks of real measurement-based calibration procedures such as clutter noise from the background or issues with non-stable device characteristics will be eliminated by the ideal simulation. To utilize the simulation approach in such a precise calibration procedure, there has to be an identical simulation environment. Owing to the holography technique, stationary background properties are eliminated by subtracting operation and there is only the remaining modeling antenna system itself in the ideal case. The second part of this thesis consists of modeling the antenna system and this modeling is done by using Volume Integral Equation (VIE). In this method, the field of the identically modeled receiver antenna is used instead of the system Green function. However, this obtained field from the simulation environment has to be equated with the value in the real measurement setup. Therefore one calibration data measured from real real-world setup is used for normalization. In this way, only one measurement for all calibration procedures will be required and calibration procedure time will be equal to the processing time of computer speed.