Sensor Fusion of LIBS and Raman Spectra for Remote Detection of Explosives

Physical Sciences Inc. proposes a novel technique to subtract background spectra from the spectra of interest and to then combine the LIBS and Raman spectra for remote explosive detection. Our approach uses support vector machines with transductive inference for training. Based on our experience we anticipate a 95% correct identification of explosive vs. non-explosive material. Further, our calculations indicate we can easily perform the required sensor fusion in near real-time ( < 2 seconds) on a conventional computer workstation and in new proposed hardware we will be able to get this processing under 0.1 seconds.