This thesis presents an empirical study comparing the ability of multi-static and bi-static, handheld, ground penetrating radar (GPR) systems, using full wave inversion (FWI), to determine the properties of buried anti-personnel (AP) landmines. A major problem associated with humanitarian demining is the occurrence of many false positives during clearance operations. Therefore, a reduction of the false alarm rate (FAR) and/or increasing the probability of detection (POD) is a key research and technical objective. Sensor fusion has emerged as a technique that promises to significantly enhance landmine detection. This study considers a handheld, combined metal detector (MD) and GPR device, and quantifies the advantages of the use of antenna arrays. During demining operations with such systems, possible targets are detected using the MD and further categorised using the GPR, possibly excluding false positives. A system using FWI imaging techniques to estimate the subsurface parameters is considered in this work. A previous study of multi-static GPR FWI used simplistic, 2D far-field propagation models, despite the targets being 3D and within the near field. This novel study uses full 3D electromagnetic (EM) wave simulation of the antenna arrays and propagation through the air and ground. Full EM simulation allows the sensitivity of radio measurements to landmine characteristics to be determined. The number and configuration of antenna elements are very important and must be optimised, contrary to the 2D sensitivity studies in (Watson, Lionheart 2014, Watson 2016) which conclude that the degree (number of elements) of the multi-static system is not critical. A novel sensitivity analysis for tilted handheld GPR antennas is used to demonstrate the positive impact of tilted antenna orientation on detection performance. A time domain GPR and A-scan data, consistent with a commercial handheld system, the MINEHOUND, is used throughout the simulated experiments which are based on synthetic GPR measurements. Finally, this thesis introduces a novel method of optimising the FWI solution through feature extraction or estimation of the internal air void typically present in pressure activated mines, to distinguish mines from non-mine targets and reduce the incidence of false positives.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:762098 |
| Date | January 2018 |
| Creators | Sule, Suki Dauda |
| Contributors | Paulson, Kevin S. ; Riley, N. G. |
| Publisher | University of Hull |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hydra.hull.ac.uk/resources/hull:16567 |
Page generated in 0.002 seconds