Using Fourier Transform near Infrared Spectroscopy in Determining Kernel Hardness, Protein and Moisture Content of Whole Wheat Flour

Abstract

Fourier transform near infrared (FT-NIR) spectroscopy was used successfully to determine the kernel hardness, protein and moisture content of a single sample set of whole wheat flour. The hardness calibration was derived by performing partial least square (PLS) regression on baseline corrected data and using the particle size index (PSI) test as reference method. Both the protein and moisture calibrations were derived by performing PLS regression on multiplicative scatter corrected (MSC) and baseline corrected data. In all cases, the calibration models were subsequently tested by means of independent validation procedures. Good calibration and prediction results were obtained for all three characteristics. Standard error of prediction ( SEP), root mean standard error of prediction ( RMSEP) and correlation coefficient ( r) values of 2.13%, 4.53% and 0.42, respectively, were obtained for the hardness calibration. Corresponding results for the protein and moisture calibrations were 0.51%, 1.16%, 0.81 and 0.15%, 0.38% and 0.85, respectively. The FT-NIR protein and moisture results of this study compare satisfactorily with results achieved in previous FT-NIR studies. No other FT-NIR wheat hardness results have been reported elsewhere, but results of this study compare well to those achieved in other NIR wheat hardness studies.