Theoretical And Experimental Research Of Autofocus For Subsurface Imaging Based On Back-projection Algorithm

Surface Penetrating Radar(SPR) is an effective equipment for location, detection, and recognition of the subsurface targets, and is widely used in civil and military applications. Subsurface Penetrating imaging(SPRI), an important signal processing technique in SPR, greatly improves image resolution and makes it easier for data interpretation such as target identification and classification. Generally, the dielectric properties of the medium should be prior information for SPRI, but this is usually unknown under the practical conditions. Therefore, the uncertainty of dielectric properties would result in the failure of subsurface imaging. Fortunately, autofocusing technique, which can refocus the back scattering energy without the accurate estimation of dielectric properties, offers a practical and flexible solution to this problem.Currently, few works on the autofocusing technique of subsurface imaging has been done, especially for Ground Penetrating Radar(GPR) that is much more seriously influenced by medium. Besides, there are not enough researches on the imaging mechanism and its availability of the autofocusing technique, which limits its application in GPR. Above all, the thesis will focus on the methodology and applications of the autofocusing technique, including its basic principles, derivation of focusing function, and performance evaluation, and so on.The autofocusing technique of subsurface targets employs the basic principles of autofocusing used in optical imaging. By adjusting parameters iteratively according to the focusing metric, autofocusing technique achieves a goal of imaging with free prior medium parameters. Chapter 2 reviewed the autofocusing techniques in optical and SAR/ISAR imaging firstly, and illustrated models and implementations of subsurface imaging autofocus, especially for GPR. Considering that focusing function plays a key role in the autofocusing technique, some indicators used for evaluating the performance of focusing function are also proposed in this chapter.Features varying with parameter errors are the basis of autofocusing implementation, thus, chapter 3 mainly studied the morphological characteristics of GPR images. The superposition model of point spread function(PSF) was used to explain the phenomenon of diffusion side lobes in GPR’s BP images. Moreover, precise mathematical expression of diffusion side lobes is derived. The given analysis indicates that the image shows a ‘defocusing- focusing- defocusing’ phenomenon with the variation of the unknown parameter.Due to the importance of focusing function, which is employed to evaluate focusing metric of image, chapter 4 focused on theoretical analysis and experiments of focusing function. Current focusing functions and their problems were illustrated firstly, and then some explorations were given. Further more, a novel focusing function was derived based on the PSF model. At last, some experiments were carried out to verify the aforementioned conclusions. What’s more, a time-domain pulse compression technology based on the principle of ‘signal matching-pulse detection-waveform substitution’ is introduced to enhance the experimental data, in which homographic filtering technology was used to solve the unknown matching signal source, one of the innovation points of this thesis.