Multi-parameter Analysis of Corn Using Near-infrared Reflectance Spectroscopy and Chemometrics

The objective of the paper is to develop a multi-parameter chemometric model to determine the starch, oil, moisture and protein contents of corn using Near-Infrared Reflectance Spectroscopy (NIRS) data. The model is developed using Partial-Least Squares Regression (PLSR) algorithm in LabVIEW 2013. The corn data set used in the study was obtained from http://www.eigenvector.com/data/Corn where it is freely available for download. The study data set consisted of NIR absorbance spectra for 80 different corn samples along with the true values of moisture, protein, oil and starch contents of these samples obtained using laboratory analysis. The spectral band considered is 1100-2498 nm at 2 nm intervals (a total of 700 wavelengths). Out of this, 60 samples were used for calibration and the parameters for the remaining 20 samples were predicted. The following correlation coefficients were achieved between the actual value and the predicted value (Moisture - 0.999, Starch - 0.984, Oil - 0.989, Protein-0.967). The results of this study show that the proposed method determines the parameters of interest accurately.