GPR detection of underground void based on low-rank and sparse representation

Underground void detection is important due to its effects on subsidence higher risk. Ground penetrating radar (GPR) is a nondestructive geophysical electromagnetic method and has been widely used in underground void detection or cavities locator. However, because of heavy clutter conditions, the void response is usually obscured. In this paper, an imaging-followed-removal algorithm based on low-rank and sparse representation is presented to suppress the clutter and obtain the void geometrical properties. The proposed algorithm firstly focus the sparsity of the raw data using Frequency-Wavenumber migration imaging method, and then removal the clutter by the robust principle component analysis (RPCA) method. Simulation results of air-filled void, air-water-filled void, and void with pipeline in its side show that our proposed algorithm can remove the dutter efficiently and then the void information can be easily obtained, which is beneficial to further identification.