Feature Extraction for Underground Object Reconstruction from Ground Penetrating Radar (GPR) data

Abstract

Ground Penetrating Radar (GPR) is very beneficial for underground object scanning and detection. It utilises radar pulses as the signal, hence it able to penetrate surfaces in obtaining the underneath information without disturbing and destructing the ground. However, its radargram output in hyperbolic signal are very challenging to be analysed. Thus, suitable algorithm has to be designed and developed to interpret the data. This work highlights on the usage of drop-flow algorithm in detecting important features of the hyperbolic signal. Previous study has shown that these features is promising in understanding and further, reconstructing the GPR data. Results show that the features extracted from the hyperbolic signal able to be identified for further processing, which is necessary for visualization purpose.