Hyper Spectral Characteristics And Diagnostic Models Of Diseases Tobacco

Hyper spectral remote sensing inject new energy into the application of remote sensing on agriculture through it’s more bands and abundant data. It is very important that monitori ng the crop growth status and the diseases, insect pests’damage to them. To study the feasi bility of monitoring the diseases and insect pests in tobacco using hyperspectral remote se nsing, leaf spectrum of tobacco infected with diseases and insect pests at different severity levels was measured by using ASD hand-held spectroradiometers. The main research conte nts include:(1investigation on disease index and classification of diseasesDuring the growth period of tobacco, dividing tobacco diseases into four classifications based on the percent of lesion area:the first class (0%), the second class (0-10%), the thir d class (10-20%); the forth class (>20%).(2)Study on high-spectral characteristics of tobacco leaves infected with diseasesTo find out the special high-spectral characteristics of tobacco leaves infected with dise ases, the spectral curve of tobacco infected with diseases and healthy, the first derivative sp ectrum curve were studied.(3)Study on high-spectral characteristics of tobacco canopy infected with diseasesTo find out the special high-spectral characteristics of tobacco canopy infected with dis eases, the spectral curve of tobacco infected with diseases and healthy, the first derivative s pectrum curve were studied.(4)Recognition and model foundation of tobacco diseases based on high-spectralThe correlation between spectrum curves and diseases index was obvious, several mod els were found due to this correlation, and it is helpful to monitor the growth of tobacco.The results suggested:The leaf spectral reflectance decreased due to the damage of dis eases and insect pests; the spectrum of tobacco leaf infected diseases moved to the directio n of short wave, that is the red-edge slope decreased, the location of red-edge moved to the short wave. Meanwhile, the canopy spectral reflectance infected in diseases moved to the direction of long wave. In this study, the wavelength bands between420nm and506nm,712nm～730nm;631nm and638nm as well as696rm and733nm were selected out as sen sitive bands region to the severity levels. Accoding to above fact, some spectrum identify a nd inversion SL models on tobacco canopy and single leaf of disease all reached the best si gnificantly level. The results suggested that spectrum characteristics on tobacco canopy an d leaf infe cted with diseases were obvious. Those model in which spectrum identify and in version SL of modes of diseases on tobacco canopy and single leaf, the first derivative spec trum at627nm、812nm and707nm:FD627nm/FD812nm y=2.7846x-0.5616R2=0.8326;FD707nm y=1.9887x-0.5372R2=0.7536; could invert accurately the tobacco canopy and single leaf infected with diseases, and they may be used to forecasting the position of tobacco infected with diseases,quantitatively.This research is the basis to monitor the diseases and insect pests in tobacco, and it has a practical significance for applying remote sensing monitoring and determining the appro priate control time.