Multi-modal Iterative Adaptive Processing (MIAP) performance in the discrimination mode for landmine detection

Due to the nature of landmine detection, a high detection probability (P d) is required to avoid casualties and injuries. However, high P d is often obtained at the price of extremely high false alarm rates. It is widely accepted that no single sensor technology has the ability to achieve the required detection rate while keeping acceptably low false alarm rates for all types of mines in all types of soil and with all types of false targets. Remarkable advances in sensor technology for landmine detection have made multi-sensor fusion an attractive alternative to single sensor detection techniques. Hence, multi-sensor fusion mine detection systems, which use complementary sensor technologies, are proposed. Previously we proposed a new multi-sensor fusion algorithm called Multi-modal Iterative Adaptive Processing (MIAP), which incorporates information from multiple sensors in an adaptive Bayesian decision framework and the identification capabilities of multiple sensors are utilized to modify the statistical models utilized by the mine detector. Simulation results demonstrate the improvement in performance obtained using the MIAP algorithm. In this paper, we assume a hand-held mine detection system utilizing both an electromagnetic induction sensor (EMI) and a ground-penetrating radar (GPR). The hand-held mine detection sensors are designed to have two modes of operations: search mode and discrimination mode. Search mode generates an initial causal detection on the suspected location; and discrimination mode confirms whether there is a mine. The MIAP algorithm is applied in the discrimination mode for hand-held mine detection. The performance of the detector is evaluated on a data set collected by the government, and the performance is compared with the other traditional fusion results.

Duke Scholars

Author Leslie M. Collins Electrical and Computer Engineering