Improvements to an explosives detection algorithm based on active neutron interrogation using statistical modeling

Abstract

Earlier efforts have identified an algorithm that uses active neutron interrogation to find explosives hidden in cargo containers. This algorithm uses flags, in the form of specific mathematical manipulations of the exiting neutron and photon radiation at different angles, to classify the cargo type, search for hidden explosives, and minimize certain false positives due to cargo heterogeneities. Statistical modeling software has now been applied to the previously-identified flags in an effort to improve the detection algorithm. The new detection models have shown accurate results exceeding 95 % for simplified screening scenarios 80–90 % when more realistic conditions are considered.