Unexploded ordnance detection by measuring object symmetry via linear prediction

Humanitarian demining has become a serious international issue nowadays due to the post-war buried unexploded ordnance (UXO) and land mines. In this research, a new technique based on linear prediction is proposed to discriminate UXO from non-UXO targets. The proposed technique takes into account the symmetry of the ground penetrating radar (GPR) signal returning from a buried object. We use the GPR data set collected by Battelle company and the Ohio State University from the Jefferson Proving Ground (JPG), U.S.A. in the experiments The results evaluated in terms of receiver operating characteristic (ROC) curves show that our proposed technique has a better detection performance than a traditional detection technique based on energy. The false alarm reductions at 100% detection by the baseline technique and the proposed technique are about 11% and 23%, respectively, compared to a situation where no technique is applied.