Variety Identification of Chinese Cabbage Seeds Using Visible and Near-Infrared Spectroscopy

Visible and near-infrared reflectance spectroscopy was applied to identify varieties of Chinese cabbage seeds. Chemometrics was used to establish the identification models from a total of 120 samples, 20 samples from each of the six varieties. Soft independent modeling of a class analogy (SIMCA) models were established based on principal component analysis, and a good identification result of about 94% was achieved based on the calibration set of 40 samples. Partial least-squares discriminant analysis (PLS-DA) and least-squares support vector machine (LS-SVM) were used to further improve the correct answer rate. A correct answer rate higher than 97% was reached by LS-SVM based on the calibration set of 40 samples, better than that of PLS-DA (81%). The generalization ability of the LS-SVM model was evaluated based on calibration sets with different numbers of samples. A correct answer rate of 100% was obtained when the number of samples for the calibration was 80. Based on the resulting coefficients and loading weights from PLS-DA, sensitive wavelength regions were screened, and five sensitive wavelengths (412, 421, 469, 681, and 717 nm) were proposed. LS-SVM identification models using these five wavelengths obtained a 98% correct answer rate based on a calibration set of 40 samples. The result shows that visible and near-infrared reflectance spectroscopy is a fast and effective technique to identify the varieties of Chinese cabbage seeds.