Detection and classification of multiple shallow-buried targets by a Ground Penetrating Radar

Date

2023-7-17

Author

Şık, Furkan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

227
views

109
downloads

Cite This

In this thesis, detection and classification of conducting and/or dielectric objects of different sizes, shapes and burial depths are investigated based on numerically simulated ultra-wideband Ground Penetrating Radar (GPR) data for different soil properties in flat or rough surface conditions. One dimensional A-scan time-domain signals, two-dimensional B-scan image data (in cross-track and/or down-track directions), as well as three dimensional C-scan image GPR data are processed altogether in single target or multiple target problems. The challenging scenario of extremely shallow-buried target detection is examined in detail paying special attention to undesired mutual coupling effects stemming from a collinear GPR antenna array. Signal processing algorithms used for the removal of direct coupling and ground-bounced signals are extensively studied and applied to various scenarios presented in the thesis on a comparative basis. Detection and classification of buried objects in a complex multiple-target scenario are investigated in detail where the targets with quite dissimilar geometries and different material compositions are placed very close to each other. Segmentation of energy maps is proposed and demonstrated to be useful to reveal the presence of low radar cross-section (RCS) targets that may be masked by closely located high RCS objects. For target classification, physics-based electromagnetic target features are extracted from the isolated A-Scan data by using the Page distribution (PD), a quadratic time-frequency distribution (TFD) technique. The method of principal component analysis (PCA) is implemented for feature fusion in the close vicinity of the detected target center to utilize multi-aspect electromagnetic scattered data for improved classification accuracy.

Subject Keywords

Ground Penetrating Radar (GPR), Shallow-buried GPR target detection, GPR target classification, Feature extraction by Page distribution in GPRs, Removal of direct-coupled and ground-bounced GPR signals

URI

https://hdl.handle.net/11511/104737

Collections

Graduate School of Natural and Applied Sciences, Thesis