Improved substance identification by suppression of multiple-reflection-induced spectral noise

Terahertz (THz) spectroscopy is a powerful technique for identification of dangerous materials like drugs and explosives, especially when hidden under optically opaque concealments. Multiple-reflections within such sample systems lead to additional spectral features which could be misinterpreted as artificial material absorption bands. We show that combinations of different chemometric tools help reducing such misinterpretations. However, an angle-depending detection scheme leads to a much more powerful artifact suppression. We received experimental data which are in good agreement with our numerical predictions. Hence, the overall false alarm rate of our pattern-recognition algorithm is significantly reduced.