A study on two-sided linear prediction approach for land mine detection

Abstract

Ground penetrating radar (GPR) is a widely used tool for land mine detection. However, land mine detection still remains a difficult task because of the changing conditions and the strong reflection of the ground. In this paper, a generalized two-sided linear prediction model is used to estimate the background. By effectively removing the background components from the GPR signal, the residual energy is found to be more reliable to generate the test statistic for detection. Results based on real GPR data show that the proposed method is able to not only remove the ground bounce and background signal but also suppress the response from some clutter objects.