Study On Monitoring Of Cotton Aphid Damage Level Based On UAV Hyperspectral Images

Cotton aphid is one of the main pests in cotton production in China,and its monitoring of the occurrence and severity is of great significance.The investigation of the occurrence of cotton aphid by the plant protection department is still based on field sampling.Traditional methods have problems such as point-to-face,time-consuming and labor-intensive,destructive sampling,and poor timeliness.In view of the limitations of traditional methods,it is of great significance to explore new modern plant protection survey methods.This paper aims to use the UAV imaging hyperspectral to monitor the cotton aphid damage level at the canopy level,and extract the seven Vegetation indexs according to the characteristics of hyperspectral resolution,using single factor and partial least squares(PLSR)regression method to construct the estimation model of the cotton canopy leaf area index(LAI)from different damage levels correspond and the estimation model of Aphid stress index.According to the characteristics of the cotton leaf curling degree as the index,in order to improve the estimation model of the aphid stress index,the cotton canopy LAI is increased as an independent variable.Finally,the spatial distribution map of the different grades of aphid infestation is made by the optimal model.The results show that:1)there are significant differences in spectral reflectance of cotton canopy corresponding to different damage levels;2)significant negative correlation between damage level and cotton canopy LAI;3)using PLSR regression algorithm combined with multiple vegetation indices the accuracy of the LAIII-PLSR model is high,R2 is 0.91.4)The accuracy of the II-PLSR model constructed by PLSR regression analysis combined with multiple spectral indices is improved.The accuracy of the II-LAI-PLSR model is improved by 6.6%,R2=0.95.With the best predictive effect,it can be used as a new technical means to monitor the damage level of cotton aphid.The visual spatial distribution map of the cotton aphid damage after the model is filled can make the relevant personnel more intuitively understand the occurrence of cotton aphids in the field.