Study In Hyperspectral Remote Sensing Monitoring The Main Disease Of Dalbergia Odorifera Leaf

This paper discussed the methods using hyperspectral remote sensing to monitor disease of Dalbergia odorifera under main disease of leaves,through building the inversion model of the disease index of Dalbergia odorifera under main disease of leaves,this results can be used as a reference for the rapid diagnosis of the Dalbergia odorifera disease degree under main disease of leaves stress,and lay the theoretical basis for the rapid,timely and large-scale monitoring of aerospace,aviation remote sensing in the future.Using non-imaging hyperspectral instruments made in SVC company from USA,taking the correlation analysis of the disease index and the original spectral reflectance,first derivative spectra,hyperspectral characteristic parameters of the Dalbergia odorifera under the main diseases of leaves stress,screening sensitive bands and training BP neural networks.The inversion model was evaluated synthetically by the correlation coefficient(r),F test value,Fitting R~2,Predicting R~2,Root Mean Square Error(RMSE).The results showed that:(1)Study on the hyperspectral inversion model of disease index under the main disease of leaves stress:Founding that the fitting R~2 and the prediction R~2 of the inversion of Dalbergia odorifera disease index under black scurf stress all pass the most significance difference,root mean square error was between 6.161 2～9.330 6.Among them,parabola model of variable FD722.8 have the highest accuracy,the fitting R~2 was 0.981,the prediction R~2 was 0.991,the root mean square error was 6.332 5.The fitting R~2 and the prediction R~2 of the inversion of Dalbergia odorifera disease index under colletotrichum stress all pass the most significance difference,root mean square error was between 8.251 6～9.918 0.Among them,unary cubic function model of variable FD718.8 have the highest accuracy,the fitting R~2 was 0.981,the prediction R~2 was 0.819,the root mean square error was 9.918 0.Showed that,the parabola model of variable FD722.8 and the unary cubic function model of variable FD718.8 was the best model.(2)Dalbergia odorifera under main disease of leaves stress disease index based on ground hyperspectral technology:Founding that the sensitive wave band and after PCA treatment of sensitive wave band under black scurf stress were used as input variables,the coefficient of determination(R~2)were 0.912 6 and 0.841 3,and the root mean square error were 6.723 6 and 12.872 7.The sensitive wave band and after PCA treatment of sensitive wave band under colletotrichum stress were used as input variables,the coefficient of determination(R~2)were 0.872 8 and 0.693 9,and the root mean square error were 6.933 0 and 19.488 4.Showed that,the extraction directly sensitive band and after PCA treatment of sensitive band were used as input variable was a kind of effective method,and the prediction results was very good,sensitive band directly was more accurate.