Date of Publication :15th July 2024

Abstract: Near-infrared (NIR) hyperspectral imaging method had been utilized for the characterization of pesticides on Butterhead lettuce leaves. For this study, butterhead lettuce leaves were contaminated with three different concentrations of pesticide. Imidacloprid was used as the contaminated pesticide with a concentration of 1:1000, 5:1000, and 10:1000. The hyperspectral data of each contaminated leaf was collected and recorded by the OCI-F series of BaySpec hyperspectral camera. The region of interest (ROIs) of the leaves were selected by CubeCreator from Bayspec. The wavelength range of the camera is from 400 to 955 nm, with a resolution of 9 nm. The Standard Normal Variate (SNV) algorithm was applied to scale the NIR hyperspectral data. Based on the different shapes of the NIR spectrum between 530 to 590 nm, the Root Mean Square Error (RMSE) was used to detect and characterize the pesticide residue. The Confusion matrix was implemented to calculate the accuracy of the algorithm. The proposed algorithm to characterize the pesticide residue on Butterhead lettuce leaves produced an accuracy of approximately 90.25%.

Reference :