Estimation of Nicotine Content in Tobacco Leaves Based on Hyperspectral Imaging

We used spectroradiometry to rapidly and nondestructively predict the nicotine content in fluecured tobacco leaves and obtained hyperspectral images of tobacco leaves during the squaring development stage at field conditions. We tested the nicotine content in the lab and analyzed correlations between 28 hyperspectral characteristic parameters and the nicotine content of tobacco leaves to establish inversion models. The result showed that (SDr−Sy)/(SDr+Sy), Rg/ Ro, and (Rg−Ro)/(Rg+Ro) provided good correlations for the nicotine content; an inversion model based on (SDr−Sy)/(SDr+Sy) in relation to the nicotine content, Y=−4.2628+5.8974X(SDr−Sy)/(SDr+Sy) −1.3260X(SDr−Sy)/(SDr+Sy)2, and a multiple regression model based on Rg/Ro, (SDr− Sy)/(SDr+Sy) in relation to the nicotine content, Y=− 0.4753+1.6982 X(SDr−Sy)/(SDr+Sy) −0.1616XRg/Ro, could be used to predict the nicotine content of flue-cured tobacco leaves during the squaring stage.