LAUSR

Ground penetrating radar is an effective nondestructive method for exploring subsurface object information by exploiting the differences in electromagnetic characteristics. However, this task is negatively affected by the existence of ground clutter and noise especially if the object is weak or/and shallowly buried. Therefore, this paper proposes a novel method for suppressing the clutter and background noise simultaneously in both flat and rough surfaces. First, the ground clutter is removed mainly by applying a simplified least square fitting background method, which remains the residual random noise signal. The remaining signal is then decomposed by singular value decomposition, which assumes that the decomposed signal contains four main components including strong target, weak target, very weak target, and accumulated noise signals. The powered singular values and their differences are clustered by K-means to extract the target signal components. The simulation results indicate that the proposed method is able to enhance the target signal with satisfactory results under both flat and rough surfaces as well as in a high-level background noise. Besides, this method also shows its superiority to the latest existing proposed methods.