Abstract

The methodologies for forming orthogonal vector representations of transformed infrared spectra of pure components in multicomponent systems are developed. Either the Gram-Schmidt orthogonalization method or factor analysis can be used to generate the vector representations; both give equivalent results when one vector representation per component is generated. The results of processing data in the Fourier domain using both orthogonalization methods are compared with those obtained by processing in the spectral domain for the following three systems: (1) simulated spectra; (2) real spectra with dissimilar features; and (3) real spectra with very similar features. In all cases, processing in the Fourier domain produces the optimum results while requiring a minimum amount of operator contributions.

Efficient computation of joint fractional Fourier domain signal representation

Lutfiye Durak, Ahmet Kemal Özdemir, and Orhan Arikan
J. Opt. Soc. Am. A 25(3) 765-772 (2008)

Matrix approach to quantitative refractive index analysis by Fourier domain optical coherence tomography

Peter H. Tomlins and Ruikang K. Wang
J. Opt. Soc. Am. A 23(8) 1897-1907 (2006)

Short-time Fourier transform and wavelet transform with Fourier-domain processing

F. T. S. Yu and Guowen Lu
Appl. Opt. 33(23) 5262-5270 (1994)

Identification of pure component spectra by independent component analysis in glucose prediction based on mid-infrared spectroscopy

Sangjoon Hahn and Gilwon Yoon
Appl. Opt. 45(32) 8374-8380 (2006)

Vector-subspace model for color representation

T. Jaaskelainen, J. Parkkinen, and S. Toyooka
J. Opt. Soc. Am. A 7(4) 725-730 (1990)