Rapid determination of protein, starch and moisture content in wheat flour by near-infrared hyperspectral imaging

Abstract

In this study, near-infrared (NIR) hyperspectral imaging (HSI) (969–2174 nm) technology was investigated for the determination of protein, starch and moisture content in 77 different varieties of wheat flour. The NIR-HSI system combined with effective wavelengths (EWs) algorithms was applied to obtain spectral information with wheat flour. Then, four regression models based on the original spectral information and the EWs were established to determine the relationship between the spectrum and detection index. Five EWs algorithms were applied to select EWs to optimise the models. The coefficient of determination and root mean square error for prediction of the obtained optimum models were 0.9859 and 1.1580 g/100 g for protein, 0.9243 and 0.2068 g/100 g for starch, and 0.8646 and 2.1669 g/100 g for moisture, respectively. The visualisation of the protein, starch and moisture content was achieved using the optimal models. The result indicated that the NIR-HSI technology is a promising approach for predicting the protein, starch and moisture content of wheat flour effectively and accurately. It also provided a reference for non-destructive and rapid prediction of other chemical compositions in foxtail millet.