Research And Design Of UAV Straw Burning Monitoring System Based On Image Processing

In rural areas,a large number of straw are burned between autumn and winter annually,which leads to increasingly serious environmental problems.Great importance should be attached to monitoring and management of straw burning.So that it is particularly significant to research and develop fast and accurate technology and platform to detect straw burning.Because of strong adaptability,anti-interference ability and practical advantages,image processing technology is widely used in fire warning system.As a flexible platform with high speed movement,small unmanned aerial vehicle(UAV)is able to complete many difficult tasks by using various auxiliary equipment.Its tremendous value had been reflected.Traditional straw burning monitoring is mainly by artificial means,which may produce high cost,small real-time and blind angle.Straw burning detection system based on image processing technology and UAV compared with traditional ways has fast maneuvering,small limit in space and can be easily controlled,which has great significance for straw burning prevention and control.Unmanned aerial vehicle(UAV)straw burning monitoring system based on image processing research image characteristics by digital image processing,computer vision and information fusion technology.The overall framework of proposed system was divided into information collection module,information processing module and information application module.In the information collection module,CCD cameraon mini UAV collects video images of the monitored region,andconvertsthem into digital images which computers can handle.In the information processing module,the main job is to recognize burning and determine its risk level.First of all,digital image processing algorithms such as graying image,Gaussian filterde-noising,image enhancement were used to do image preprocessing work.In part of extracting foreground area,experiments showed that traditional threshold segmentation technology cannot effectively obtain accurate flame area in complex background environment.An algorithm based on improved Visual Background Extractor(ViBe)and pixel color weighted was put forward to solve this problem.The fast smoke extracting algorithm was based on color feature.In recognition of burning,multiple criterion of information fusion technology was applied.A burning danger hierarchy model was established to provide more intuitive measure to burning risk.In the last,three sets of tests were designed to verify performance of these algorithms.In the information application module,Visual Studio 2010,MFC and OpenCV were used to realize the PC interface.Video image,flame and smoke characteristics,risk and alarm information can be displayed and saved.With the advantage of image processing and UAV,this system offers platform to monitor ground from high view,which can effectively avoid disadvantages of traditional straw burning monitoring technologies.Tests showed that this system has high recognition accuracy,good real-time performance and more intuitive measure to burning risk,which had a potential practical value.