Classification of nutritional components in cereal products based on their relative content of total dietary fiber, fat, protein and sugar was accomplished using two spectroscopic methods, Fourier transform Raman (FT-Raman) and near-infrared (NIR) spectroscopy. FT-Raman spectra (202-3399 cm-1) and NIR spectra (400-2500 nm) of one hundred twenty ground cereal samples were collected and utilized for chemometric analysis. Classification of the samples according to high and low content of each component was attempted using soft independent modeling of class analogy (SIMCA) and discriminant partial least squares (PLS) regression. The selection of x-variables (frequency regions) and various preprocessing of the spectral data were investigated to improve the accuracy.
source: http://www.ars.usda.gov/research/publications/publications.htm?seq_no_115=179721