Demirağ, YiğitBütün, BayramÖzbay, Ekmel2018-04-122018-04-12201797815106092110277-786Xhttp://hdl.handle.net/11693/37559Date of Conference:Proceedings of SPIE, Next-Generation Spectroscopic Technologies XConference name: 10–11 April 2017In this study, we present a classification algorithm for terahertz time-domain spectroscopy systems (THz-TDS) that can be trained to identify most commonly used explosives (C4, HMX, RDX, PETN, TNT, composition-B and blackpowder) and some non-explosive samples (lactose, sucrose, PABA). Our procedure can be used in any THz-TDS system that detects either transmission or reflection spectra at room conditions. After preprocessing the signal in low THz regime (0.1-3 THz), our algorithm takes advantages of a latent space transformation based on principle component analysis in order to classify explosives with low false alarm rate. © 2017 SPIE.EnglishAlgorithmClassificationExplosiveSpectroscopyTerahertzConference Paper10.1117/12.2264953