Attenuation of diffractions from reinforcement mesh in GPR data using derangement-based FX Cadzow filter

Abstract

The ballastless track is widely used in Chinese high-speed railways. The defect detection of it is an important task. Ground penetrating radar (GPR) is the main tool used for this task. In GPR data, the reflection from underground structures and the diffraction from potential defects (e.g. voids, small-scale cracks) are the target waves used for imaging. Since the ballastless track is reinforced with steel mesh, those target waves tend to be distorted or even overwhelmed by the undesired diffraction from reinforcing steel bars. In this case, some processes (e.g. layer tracking, defect recognition) that require high-quality data, are difficult to carry out after imaging. In order to provide high-quality data for these processes, we introduce a FX Cadzow filter with a derangement operator to attenuate undesired diffractions while keeping target waves intact. This method bases on the spatial coherence difference between target waves and the undesired diffraction. Firstly, we make target wave events relatively horizontal by dividing input data into small blocks. Secondly, the derangement operator destroys the spatial coherence of undesired diffractions by deranging each block of data. Finally, target waves are extracted by the FX Cadzow filter. We also present a 3D GPR model of the ballastless track and its parameters (e.g. relative permittivity, electric conductivity). Based on this model with and without steel bars, the software gprMax is employed to forward the simulated data. The simulated data and the field data have been used to testify the effectiveness of the proposed method.