Improving detection of low-metallic content landmines using EMI data

EMI, or metal detector, sensors are used extensively to detect landmines. There is often a significant amount of metallic clutter present in the environment, thus EMI sensors that utilize traditional detection algorithms based solely on metal content suffer from large false alarm rates. The focus of this paper is on performance improvements that have been obtained using Bayesian-based statistical algorithms designed to detect low-metallic content landmines in highly cluttered environments. The Joint UXO Coordination Office (JUXOCO) is sponsoring a series of experiments designed to establish a performance baseline for a variety of sensors for the problem of detecting low-metallic content mines. Algorithm developers are blind to the ground truth. This paper describes the results of several experiments performed in conjunction with this effort. Statistical algorithms have been applied specifically to the problem of detection of low-metal mines, and dramatic performance improvements have been observed with respect of the baseline performance. The tradeoffs associated with the amount of spatial information used as well as the incorporation of prior information are explored, along with the performance of alternative approaches. Approaches to classification of the mine-type are also discussed and the performance of such classifiers is presented.