Time-domain nuclear magnetic resonance (NMR) is used for the characterization of wheat infested with Sitophilus zeamais. NMR parameters (T21, T22, P21, P22, and A21/22) were achieved by biexponential analysis combined with a discrete method. Sound wheat, S. zeamais, and S. zeamais/wheat binary mixture are all explored by this method to find the changes in the process of the number increase and growth stage. Based on different sets of NMR parameters, the classification and quantification of the stage and number are made by linear discriminant analysis and partial least squares regression with very high accuracy. The weight and moisture content information lack will make the misclassification rate increase but not by more than 7%. This NMR-based hidden-insect detection method, with fast and nondestructive advantages, was confirmed by X-ray and had a high potential to be equipped in the online analysis system.
Keywords: CPMG; LDA; PLS; S. zeamais; infested wheat; low-field 1H nuclear magnetic resonance.