Research On Multi-band Image Moving Target Tracking System Carried By UAV

With the rapid development of artificial intelligence and automation industry,UAV intelligent system has more and more applications.But the visible airborne camera is only suitable for collecting images in normal environment,it cannot be applied to changeful scene.So this paper combines visible images and infrared images to obtain good target detection result.Then it realizes motion target detection and tracking in multi-band images.It mainly studies the ground station task of UAV intelligent system,including UI,motion detection and target tracking.An intelligent tracking system based on UAV is designed.Its hardware part consists of unmanned aerial vehicle platform,infrared camera,visible camera and ground station.The image transmission module of UAV is used to transmit and feedback the data collected by the camera.Its software part uses VS2013 to design ground task UI and function of each module.According to the properties of pictures captured by a camera on the UAV,the paper studies three methods of motion detection in dynamic background,including improved frame difference,background subtraction and optical flow.It mainly studies that motion estimation and compensation is used to improve the influence of global background movement.The fusion results of OSTU threshold segmentations in infrared and visual images are used to improve the influence of local background movement.According to the classic CamShift single target tracking,this paper uses the LARK feature to global match for single tracking called GFMT.It can weaken background interference.According to the characteristics of changeable target and LARK is sensitive to small spatial structure,a tracking method based on local feature statistical matching called LFSMT is proposed.It can robustly track changeable target.From the experiments,the intelligent tracking system designed in this paper can accurately detect and track moving targets in dynamic scenes.