Automatic distortion correction and signal reconstruction in GPR data in tunnel lining surveys

Ground-Penetrating Radar (GPR) is a key technology for ensuring the structural integrity of tunnel linings. However, data acquired in the commonly used Time mode is often distorted by the combined effects of data loss during wireless transmission and geometric distortion from non-uniform scanning. Existing reconstruction methods are generally unsuitable for this scenario as they typically require the locations of missing data to be known. To address this challenge, this paper proposes an integrated processing workflow. It first employs a two-stage module to automatically identify and regularize both types of distortion, resulting in a calibrated B-scan dataset that contains identified vacant traces. Subsequently, a rank-reduction interpolation algorithm, improved by a data-driven adaptive rank selection strategy, is utilized to perform high-fidelity reconstruction of these vacancies. Validation with both simulated and real GPR data demonstrates that the proposed method can effectively correct the distortions caused by data loss and non-uniform scanning, and successfully achieve high-fidelity data reconstruction through an automated, parameter-adaptive workflow, effectively eliminating the need for manual parameter tuning inherent in traditional rank-reduction methods.