Using bidimensional empirical mode decomposition method to identification buried objects from GPR B-scan image

Abstract

In order to successfully identify the subsurface targets amidst the surrounding clutter, it is necessary to locate and distinguish the genuine target reflections from spurious reflections. In this paper, we propose a novel Bidimensional Empirical Mode Decomposition (BEMD) system to identify buried objects from ground penetrating radar (GPR) images. The entire process can be subdivided into four steps. First the image is decomposed by the BEMD to extract the Intrinsic Mode Functions (IMFs) of the B-scan image. All these IMFs can be expressed as gradual single-frequency signals that enhance the physical meaning of instantaneous frequencies and instantaneous amplitudes. Then the IMF component which reflects more target information is selected for detection of further hyperbolas. After the extraction, background clutter is to a large extent removed while keeping the target signal. To find the best-fitting hyperbolas, object position of vertex is estimated by maximum point estimation method and velocity is estimated by minimum entropy method. Finally, the location of reflection hyperbolas is extracted. Applications of this method over simulated image and experimental data show its effectiveness in the detection of hyperbolas.