Development Of A Multispectral Imager For Early Prevention And Control Of Large Areas Of Pine Wilt Disease Trees By Unmanned Aerial Vehicles

Pine wilt disease has the characteristics of rapid infestation and wide spread,which is extremely harmful to forests.Treatment of the disease with drugs or biological control methods is not effective.Cutting down diseased pine trees and transporting them out for destruction to interrupt the source of pest transmission is the most effective method of pine wilt disease control.The traditional node of pine wood nematode control is in autumn,after the main vector of pine wood nematode –the monochamus alternatus feathering and migration,although this control method can control pine wood nematode epidemic wood,but has missed the best control time.The early prevention and control monitoring of pine woods should be carried out before the flight of the monochamus alternatus to improve the prevention and control effect and stop the spread of pine wilt disease from the source.The key to this control method is to monitor the pine forest and pinpoint the infected pine trees.However,there are three main problems in the traditional methods for monitoring pine wilt disease trees: Lack of a dedicated monitoring channel center band and bandwidth,low optical resolution of the sensor,and high influence by light changes in the monitoring environment.Therefore,there is an urgent need to develop a multispectral imager dedicated to the early control of pine wilt disease trees with reasonable optical resolution and accurate spectral reflectance acquisition during long flight time,which can efficiently and accurately serve the early control monitoring of pine wilt disease trees in large areas.In this study,we used an unmanned aerial vehicle(UAV)hyperspectral imager to obtain the vegetation canopy characteristic characteristic spectrum of pine wilt disease trees for early prevention and control,and used pine wilt disease trees characteristics and spectral analysis techniques,spectral dimensionality reduction techniques and consistency analysis,combined with the operational specifications of the UAV pine wilt disease trees monitoring,to select a reasonable channel center band and its bandwidth and optical resolution parameters for the UAV pine wilt disease trees multispectral imager for early prevention and control.Then,using photoelectric detection technology and intelligent instrument development technology,we developed a multispectral imager for early prevention and control of pine wilt disease trees in UAV and the supporting software.The research content as well as the results of this paper include three working phases:(1)The UAV pine wilt disease trees early prevention and control large area monitoring multispectral imager parameters selection,its research includes three parts: First,the correlation analysis of pine wilt disease trees early prevention and control monitoring features and canopy characteristic spectrum,for pine wilt disease trees in the early prevention and control period,the characteristic characteristic spectrum of healthy pine trees,diseased pine trees,initial diseased pine trees and other major disturbance forest grass,through the disease trees Based on the spectral analysis technique,545 nm,580nm,670 nm and 750 nm were selected as the central bands of the characteristic spectral channels for early prevention and control monitoring of pine wilt disease trees;in the second part,the bandwidths of the four channels were determined as 30 nm,60nm,14 nm and 30 nm by using the spectral downscaling technique;in the last part,the bandwidths of the four channels were determined as 30 nm,60nm,14 nm and 30 nm by using the spectral downscaling technique.In the last part,based on the spatial resolution required by the industry standard for pine wilt disease trees monitoring,and comparing the operational efficiency and quality under different resolutions,different camera chip sizes,and different flight heights,a reasonable 600 m operational height and an industrial camera with 1200 W pixel optical resolution were selected.(2)Hardware and software development and integration of the multispectral imager: This part of the equipment development is based on the determined parameters of the multispectral imager for early prevention and control of pine wilt disease trees in large areas,which includes hardware development and software design and integration.The hardware development includes the configuration of the multispectral imager channel centre band and bandwidth,camera and lens selection,instrument shell design,Industrial Personal Computer installation and adaptation,illuminance meter development,imager synchronous trigger module integration,electromagnetic shielding design and heat dissipation design.The software component includes the multispectral imager operating software,which relies on the Open CV static library,and the post-processing software for the identification and location of pine wilt disease trees.The post-processing software uses synchronous correction models and GPS correction techniques to pre-process the original multichannel images,and special software to align and stitch together the multispectral images,combining spectral operations,expansion and erosion operations and connected-domain center-of-mass algorithms to identify,locate and export the pine wilt disease trees.Finally,the imager is adapted and integrated with a long flight time unmanned airborne platform using a combination of hardware and software.(3)Calibration and field validation of the dedicated multispectral imager equipment: This part provides calibration validation and practical validation of the developed illuminance meter module and the multispectral imager for early prevention and control monitoring of pine wilt disease trees.Simultaneous calibration validation: The developed illuminance meter module was calibrated with a high-precision ST-85 illuminance meter with R2 as high as 0.9979;Spectral characteristics calibration validation: The developed instrument was calibrated with an ASD Field Spec 4 feature hyperspectrometer with reflectance R2 of four channels and hyperspectrometer as high as 0.979,0.9503,0.9156 and 0.9822.Field validation:Through the field monitoring of 10,000 mu of pine forest,the accuracy of the instrument can reach 94.6% for monitoring diseased pine trees,with a positioning error of 0.739m~5.245 m,and 92.8% for monitoring initial diseased pine trees,with a positioning error of 1.12m~6.998 m.The instrument solves the difficulties of early monitoring of existing pine wilt disease trees,and its monitoring recognition accuracy and positioning precision reach the good and excellent levels of UAV pine wilt disease trees prevention and control monitoring standards respectively,and can be applied to UAV pine wilt disease trees early prevention and control large area monitoring.