Automatic asphalt layer interface detection and thickness determination from ground-penetrating radar data

Abstract

Ground-penetrating radar (GPR) has been widely used in asphalt pavement layer thickness measurement. It requires the detection of pavement layer interfaces, which could be accomplished via manual inspection. This may be subjective and cause inaccurate thickness measurements due to the field data noise. Existing layer interface detection methods for thickness determination usually require customized programming and/or manual inputs for execution, which could hinder real-time applications. This article proposes an automatic method for asphalt pavement layer interface detection and thickness determination using GPR. The effects of in-situ noise levels on thickness measurement accuracy were discussed. The Canny edge detector was used to estimate the two-way travel time (TWTT) range. The wavelet transform was used for noise cancellation. Layer interface reflections were identified using the matched filter method for thickness determination. Vibration effect during the continuous GPR survey was mitigated using the empirical mode decomposition (EMD) method. A simulation study was conducted. The proposed method provided more accurate measurements than manual inspection for thickness values from 10.16 cm (4 in.) to 2.54 cm (1 in.). GPR field test was performed, and results were compared to field core thickness measurements. Results show that GPR-measured thickness errors are 6.5 % and 1.6 % using manual inspection and the proposed method, respectively. The proposed method is automatic and can be implemented using existing programming functions, which may enable real-time asphalt pavement thickness determination during GPR surveys.