Techniques for focusing pulse -echo ground -penetrating radar images of subsurface regions with unknown permittivities through planar and non-planar boundaries
Abstract
Pulse-echo ground-penetrating radar (GPR) imaging algorithms typically rely on the assumption that the area of interest may be approximately modeled as a homogeneous background material containing scatterers and clutter. Iterative algorithms require an estimate of a wave's propagation velocity in that medium, and noniterative algorithms require that the velocity estimate be accurate for the resulting image to be a faithful rendering of the scene. In this study, we revisit the time-of-flight velocity estimation method and adapt it to account for an air gap which may exist between the radar antenna aperture and the air-ground interface. We also propose a new method for estimating the background wave velocity based on the plane-to-plane backpropagation (PPB) imaging algorithm. The subsurface imaging problem is further complicated when the ground's surface is non-planar and the CPR measurements are acquired some distance above that boundary, because Fourier-based imaging techniques generally require the air-ground interface to be planar. When Fourier techniques are used to image through a non-planar surface, several effects will be evident in the resulting images, namely, the images of the buried targets will spread, fade, and will appear to shift from their actual locations. This thesis describes the effects of an incorrectly placed air-ground interface on GPR images created with the PPB method. To compensate for a non-planar air-ground interface, the wideband time-domain synthetic aperture method is reviewed and reformulated, using Snell's law of refraction, to account for measurements taken some distance above an undulating interface. A modification of the PPB algorithm that accounts for a non-planar interface is also formulated, and illustrative examples of both imaging methods are given to validate the algorithms' performance for surface variations that are relatively smooth with respect to the radar waveform wavelengths. Additionally, a method for inferring the surface topography from the radar returns is described.
Degree
Ph.D.
Advisors
Bell, Purdue University.
Subject Area
Electrical engineering|Geotechnology
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.