Research on Model Estimation of Chlorophyll Content in Fresh Tobacco Leaves Based on Hyperspectral Technology
A hyperspectral imager is used to collect hyperspectral images of fresh tobacco leaves under natural light condi-tions.After six preprocessing methods such as multiple scattering correction,normalization,and SG convolution smoothing,the original spectral data are preprocessed,and then the continuous projection method(SPA)is used to extract characteristic sensitive wavelengths,it takes the spectral reflectance corresponding to the selected sensitive wavelengths as input variables,and uses the BP neural network algorithm to predict and model the content of chlorophyll a and chlorophyll b in fresh tobacco leaves.The results show that the chlorophyll a content prediction model(SNV-SPA-BP)and the chlorophyll b content prediction model(SG2-SPA-BP)have the best results,and the correlation coefficients of the prediction sets are 0.897 and 0.936,respectively,which can be combined with hyperspectral imaging technology.The chlorophyll content of outdoor fresh tobacco leaves is quickly and non-destructively predicted.
国家科技期刊平台
NSTL
万方数据
