The Relationship Of Hyperspectral Characteristic And Pigment Content On Different Tobacco Varieties

Researched thevariation of pigments and canopy spectral and their relationship of 14 tobacco varieties (4 major types) by field experiment, the detail results showed as that:1 The Pigment content of diferent tobacoo varieties in a series growth stages was bottom leaves maturity>topping period>vigorous period>ventral leaf maturity>rosette stage.2 The vanopy spectrumReflectivity of different tobacco varietiesShowed differences.The canopy spectrumReflectivity of the same speciesShowed differencesin different periods.in the rang of visible,the size of canopy spectrumReflectivity was bottom leaves maturity>topping period>vigorous period>central leaf maturity>rosette stage.in the rang of green light(510nm-560nm),for the different tobacco varieties,the size of canopy spectrumReflectivity was rustica tobacco>flue-cured Tobacco>sun-cured tobacco>wild Tobacco.The canopy spectrumreflectivity of Flue-cured Tobacco was G80>yun 87>yun 85>K394>HZ07-5-1>K326,ofsun-cured tobacco was xiaohuboxiang>dafu>aizi> hefeng Sun-cured tobacco. The canopy spectrumReflectivity of K326 and K394 was lower than their CMS lines.3 The original spectral reflectance and pigment content of leaveswas a significant correlation. Different species showed the same rule. K326, for example, in the whole wavelength range, their were 769,797,782 and 818 wave bands with that chlorophyll a, chlorophyll b, chlorophyll a+b, carotenoid content was significant correlation between levels of respectively, and of whichwere a significant negative correlation between bands accounted for more than 90%.4 Most tobacco varieties was a significant negative correlation with High spectral parametersD、Db、Rb、Rg、SDr、SDtb,and a significant Positivecorrelation with SDy.Filter edout the most relevant parameters of all varieties, and established pigment content regression model by these high spectral parameters.By SPSS analysis, the estimates of pigment content of all varieties were made to test for differences with their true value, which proved the feasibility of the monitoring model.