Research And Development Of UAV Forest Images Transmission APP And Pest Detection System

In the process of forestry resources investigation,due to the wide range of forest distribution,the diverse geographical environment of forest land,and the difficulty of real-time surveying tasks,it has brought great difficulties to forest worker to carry out forest resource investigation and pest detection and prevention.Unmanned aerial vehicle(UAV)has been widely used in various application fields in recent years due to its characteristics of light size,simple operation and strong automationDJI Mobile SDK,an open source software development tool of DJI,is combined with in this thesis to develop a system that can control the UAV to continuously take images of forest land and detect insect pest areas.Firstly,an APP is developed to support DJI Phantom 3 Professional drones and DJI Mavic 2 Pro drones.By using this APP,users can manually control the UAV flying automatically according to the pre-planned route.During the flight of UAV,users can choose to take single photos manually or take photos continuously at a fixed time.The photos will be automatically downloaded to the mobile phone album and uploaded to the forest server.Secondly,the UAV images of pine forests infected by Dendroctonus valens are chosen as the experiment data,and the linear spectral clustering(LSC)superpixel algorithm is used to realize the detection of the affected area,which can mark the detection results in UAV images.Finally,the images of pine forest infested by Dendroctonus valens are used as the data set.The detection model is trained by using the deep learning network based on YOLOv3,in order to detect the affected area in the pine forest imagesTesting results indicate that the UAV forest images transmission APP and pest detection system runs smoothly.The user can successfully connect the UAV through the APP to achieve image acquisition in the forest area.The image is successfully sent back to the forest server.The server can obtain the detection results by calling the forest pest area detecting program based on LSC superpixel or pest area detection program based on YOLOv3 model.The detected pest areas have certain reference value after manual comparison,which has made theoretical guidance for forestry pest control work.