A comprehensive review of data processing and target recognition methods for ground penetrating radar underground pipeline B-scan data

Underground pipelines hold a crucial role in modern urban infrastructure, and Ground Penetrating Radar (GPR) has been increasingly favored as a non-destructive tool for their detection and monitoring. However, the complex and varied urban underground environment, as well as the dense distribution of various pipeline materials, present significant challenges in the interpretation of GPR signals and the recognition of targets. This study provided a comprehensive review of the current state-of-the-art in data processing and target recognition methods for GPR underground pipeline B-scan data. The unique features and characteristics of GPR pipeline B-scan data were initially examined, including the impact of pipeline materials, scanning methods, and electromagnetic wave frequencies. Traditional signal processing techniques, such as filtering, wavelet transform, and empirical mode decomposition, as well as emerging machine learning and deep learning-based methods for denoising, feature extraction, and target recognition, were systematically reviewed. The advantages and limitations of these approaches in practical applications were analyzed in detail. Feasible research directions were proposed to address the current challenges and further enhance the effectiveness of GPR technology in underground pipeline detection and management. This review serves as a valuable reference for researchers and practitioners in the fields of GPR, machine learning, and underground infrastructure monitoring.