Ground clutter suppression in GPR using framing based mean removal and subspace decomposition

Ground penetrating radar (GPR) is one of the effective tools for detecting subsurface objects. However, the existence of ground clutter negatively affects target signal extraction particularly for shallowly buried objects and/or weak target responses. Thus, the reliable clutter suppression method is needed for enhancing the target responses. This paper proposes a novel approach for ground clutter reduction by using framing based mean removal and a subspace decomposition method. In the first step, the GPR signal is processed by cross-correlation with a reflection from a flat-surface as the reference for determining the timing range of surface reflection. Then, the mean of the two neighboring traces in the predicted range is removed. The singular value decomposition and K-means algorithm are used to remove the residual noise according to the singular values. The simulation results indicate that the proposed method can effectively suppress the ground clutter without contamination of the target signal.