|Online ISSN : 2349-8080
Issues : 12 per year
Publisher : Excellent Publishers
Email : email@example.com
A novel technology for rapid identification of Machilus Nees species using the visible-near infrared spectrum (300-1100 nm) is described in this study. The reflectivities of new leaves of seedlings from 9 species of the genus Machilus Nees were collected. Stepwise discriminant analysis was applied to the spectral information of the leaves, and 18 unique bands were selected from 126 bands total. After obtaining the spectral information for the unique bands, the Bayesian discriminant method was applied to establish the discriminant analysis model for Machilus Nees species. According to the discrimination model, combinations of 6, 12, and 18 unique bands were selected, and the discrimination accuracies of 180 training samples reached 76.111%, 83.889%, and 93.889%, respectively, while the accuracies of 90 testing samples were 77.778%, 84.444%, and 95.556%, respectively. These results validated the discrimination model for Machilus Nees species constructed from the spectral information of 18 selected unique bands. The application of visible-near infrared spectrum technology combined with discriminant analysis could provide a novel approach for the rapid and accurate identification of Machilus Nees species.