IDENTIFICATION OF CHEMICAL-CONSTITUENTS BY MULTIVARIATE NEAR-INFRARED SPECTRAL IMAGING

A near-infrared imaging spectroscopic system was tested to identify three main components of wheat (bran, gluten, starch). The system described below permitted the recording of images between 900 and 1900 nm by steps of 50 nm. A direct examination of images showed that starch was correctly identified at 1550 nm. However, such a direct study of images was not sufficient to characterize all the constituents. The images, therefore, were linearly combined by applying discriminant analyses. The more revelant wavelengths (950, 1450, 1500 nm) were determined using a stepwise discriminant analysis, and a mapping of the chemical constituents was obtained by applying a canonical discriminant analysis. In the segmented image, the percentages of well-classified pixels were 92 for bran, 95 for gluten and 99 for starch.