Online Variety Discrimination of Rice Seeds Using Multispectral Imaging and Chemometric Methods

Abstract

Variety identification plays an important role in ensuring the quality and quantity of yield in rice production. The feasibility of a rapid and nondestructive determination of varieties of rice seeds was examined by using a multispectral imaging system combined with chemometric data analysis. Мethods of the partial least squares discriminant analysis (PLSDA), principal component analysis-back propagation neural network (PCA-BPNN), and least squares-support vector machines (LS-SVM) were applied to classify varieties of rice seeds. The results demonstrate that clear differences among varieties of rice seeds could be easily visualized using the multispectral imaging technique and an excellent classification could be achieved combining data of the spectral and morphological features. The classification accuracy was up to 94% in a validation set with the LS-SVM model, which was better than the PLSDA (62%) and PCA-BPNN (84%) models.