Classification of Glycyrrhiza Seeds by Near Infrared Hyperspectral Imaging Technology

The near infrared hyperspectral imaging technology was used to classify glycyrrhiza seeds, which laid a theoretical foundation for the development of the equipment for the selection of glycyrrhiza seeds. Having Taken 475 glycyrrhiza seeds from four different origins, three varieties and two planting patterns as samples, 288 bands of hyperspectral images were collected in batches in the range of 948-2512 nm. Full bands were used to establish the PLS-DA discriminant model. The SVM discriminant model based on characteristic bands was established by PCA. Experiments showed that the two discriminant models had their own merits in the three aspects of origins, varieties and planting patterns, and the SVM model accuracies of training set and test set were more than 93%. We can draw a conclusion from the experiments that near infrared hyperspectral imaging combined with model discriminant analysis could be used to identify the origins, varieties and planting patterns of glycyrrhiza seeds.