RAPID DISCRIMINATION OF HIGH-QUALITY WATERMELON SEEDS BY MULTISPECTRAL IMAGING COMBINED WITH CHEMOMETRIC METHODS

摘要

This study focuses on the feasibility of nondestructive discrimination of high-quality watermelon seeds with a multispectral imaging system combined with chemometrics. Principal component analysis (PCA), least squares-support vector machines (LS-SVM), back propagation neural network (BPNN), and random forest (RF) were applied to determine the seed quality. The results demonstrate that both the spectral and the morphological features are essential for discrimination of the quality of watermelon seeds. Clear differences between high-quality watermelon seeds and other watermelon seeds including dead seeds and low-vigor seeds were visualized, and an excellent classification (with accuracies of 92% in the LS-SVM model for Julong and 91% in the RF model for Xiali, respectively) was achieved. These results indicate that multispectral imaging could be used for rapid and efficient nondestructive quality control of watermelon seeds.