Automatic Extraction of Hyperbolic Signatures in Ground Penetrating Radar Images

Deterioration of concrete structures has become a widespread problem with high repair costs. The corrosion of rebar is one of the major causes. GPR has potential for rebar corrosion detection. But the rapid survey generates a large amount of data, which require an automatic approach for effective data processing and information extraction. This paper proposes an automatic process to effectively extract the rebar reflection in the radargram image and estimate the concrete condition above the rebar. The process uses template matching to locate the hyperbola position, image processing to extract hyperbolic region, and algebraic fitting to rapidly estimate hyperbola parameters. The estimated parameters can be used to calculate the wave propagation velocity and relative permittivity in concrete above the rebar, which can be used to further evaluate the concrete condition. The effectiveness of the proposed method is validated using experiment testing.