Noise reduction algorithm of GPR data based on MMSE‐PDS

Abstract

AbstractGround‐penetrating radar (GPR) technology is widely used in tunnel engineering detection; however, various factors, such as environmental interference and low signal‐to‐noise ratio characteristics of the echo data, limit the detection accuracy. A noise and interference suppression algorithm based on improved singular value decomposition is proposed in this paper. Compared with traditional filtering methods, the proposed method has the advantages of thorough denoising, no clutter, efficient improvement of profile resolution and less dependence on parameters. The main features of the proposed algorithm are as follows. (1) Given the global characteristics of the noise disturbance on the signal space, the minimum mean square error estimation is employed to approximate the effective signal, introducing the correction factor to suppress the larger singular value from the noise output in the reconstructing process of the effective signal subspace and to eliminate the strong direct wave interference to avoid producing false signals. (2) A positive difference sequence search algorithm based on rank order variance as well as the method of selecting correction factors are proposed to improve the processing accuracy. In order to verify the design, the tunnel lining simulation model and the actual tunnel lining detection data are used. The results show good performance for noise and interference suppression, providing technical support for improving GPR data quality and tunnel detection accuracy.