Estimation of Tobacco Biochemical Parameters Based on Non-Imaging Hyperspectral
In order to quickly and nondestructively detect the chlorine content of tobacco leaves,master the health status of tobacco plants,and control the quality of tobacco leaves,Yunyan No.97,which was mainly planted in the tobacco planting base of Da Mu Gu Township in Luliang County,Yunnan Province,was selected as the research object by collecting the hyperspectral data of tobacco leaves and determining its chlorine content.The correlation analysis and multiple stepwise regression or linear fitting method were used to analyze the relationship between chlorine content and the original spectral reflectance,a series of transformation forms of reflectance and hyper-spectral characteristic variables.The results showed that the characteristic wavelength of hyperspectral reflectance of chlorine content in tobacco leaves was 420 nm.The first derivative characteristic wavelength of the spectrum is 939 nm.The spectral logarithmic characteristic wavelength is 419～422 nm;The spectral logarithmic first deriva-tive characteristic wavelength is 372 nm;The normalized spectral characteristic wavelength is 1005nm;The nor-malized first-order differential characteristic wavelength is 1694nm.The characteristic variable λr(red edge posi-tion)had the best correlation with chlorine content in tobacco leaves.The spectral normalization stepwise regres-sion equation was the best for estimating chlorine content in tobacco leaves(R2=0.632,P＜0.01).
NETL
NSTL
万方数据
