Multimodal Trace Explosives Detection with Feature-Based Fusion

Physical Sciences Inc. proposes to develop a multimodal system for long-range standoff trace explosives detection and mapping with sensitivity in the microgram to nanogram range. The technology will rely on data fusion between multiple sensors, including standoff explosives mapping and follow on compound identification. A probabilistic detection approach is used through fusing the outputs of different modalities in the spatial and spectral domains. The Phase II effort will encompass development assembly and characterization of a prototype standoff, multimodal trace explosives detection system utilizing a high-spatial resolution single-band imager, an active long-wave infrared diffuse reflectance spectrometer, and a commercially available close range Raman spectrometer for explosives mapping. The Phase II effort will focus on demonstrating the feasibility of the proposed approach by characterizing prototype performance with outdoor field data representative of vehicle interrogation checkpoints, with surface loadings in the microgram per square centimeter sensitivity range. An improved detection capability will be demonstrated through the use of innovative data fusion, detection, and spatial filtering algorithms. Completion of the Phase II effort will result in a demonstration of the standoff explosives detection system at a government test event.