Pattern recognition of amphetamines FT-IR spectra with modified phase – input Fourier correlation
Pages 70-76
Alin C. Teusdea, Mirela Praisler
Abstract
This paper presents a pattern recognition analysis over a database of GC-FTIR spectra of pure amphetamines (3745-605cm-1). The analyzed spectra are organized in three classes: stimulant amphetamines (15 spectra), hallucinogenic amphetamines (7 spectra) and non-amphetamines (135 spectra).
The correlation process is based on amplitude sine modulated fringe-adjusted phase-input joint transform correlation method (FA-sinePiJTC). In this method, the correlation is done between a pair of phase-transformed amphetamines spectra. The proposed phase-input algorithm consists of two steps: the amplitude sine modulation step and the adaptive filtering step.
The results show that all the correlations coefficients corresponding to the 157×157 correlation matrix are above the detection efficiency coefficient threshold value of 1,2000. In conclusion, the pattern recognition method based on modified phase-input Fourier correlation is adequate for the discrimination of the GC- FTIR spectra of amphetamines.