Identification of mixed substances using a random forest regressor to classify THz absorbance spectra
MetadataShow full item record
We report on the development and application of a random forest regressor that not only identifies but also estimates the relative concentrations of substances (one explosive and two simulants), both in one-substance and two-substance samples. Performance of the regressor is quantified using Receiver Operating Characteristics and the performance is contrasted with that of a simple Spectral Angle Mapping technique that worked well on single-substance samples [1-3].
Rheenen, Arthur Dirk van; Aurdal, Lars; Nystad, Helle Emilia; Haakestad, Magnus W.. Identification of mixed substances using a random forest regressor to classify THz absorbance spectra. Proceedings of SPIE, the International Society for Optical Engineering 2018