ABSTRACT:

Time-domain terahertz spectrometry is a promising technology for identifying substances covered by optically opaque dielectric materials. However, the performance of algorithms analyzing this data has been weak, especially with a low signal-to-noise ratio. We have developed a new approach for analyzing this data that is robust against the effects of scattering and variable power spectral density. Identification of chemicals with over 97% accuracy is achieved with a false positive rate of only 0.3% on a dataset giving the literature methods much difficulty.