Automated Detection of Fingerprint Traces of High Explosives Using Ultraviolet Raman Spectroscopy

Ultraviolet (UV) resonance Raman spectroscopy is a promising technique for the detection of trace explosives. For real-world applications, it is necessary to develop data evaluation algorithms that automatically recognize the spectral features of explosives in a sample spectrum. We have developed a robust algorithm that can tolerate high levels of fluorescence background. We successfully demonstrated the detection of traces of ANFO and TNT explosives at surface coverage levels of 55 mu g/cm(2) in a blind test experiment. The sensitivity and selectivity is discussed in terms of receiver operating characteristics (ROC) curves.