UAV Image Target Tracking And Positioning

Unmanned aerial vehicle(UAV)is a kind of aircraft which is not required to be driven manually,has the advantages of rapid movement,high flexibility,and is suitable for the task of monitoring,surveying and material transportation under complex scenes.UAV has been in the military,industrial and civilian use has a wide range of uses.In this paper,the UAV image target tracking and positioning for the research project to rotor unmanned aerial vehicle as the research platform,focusing on research based on UAV image tracking and positioning algorithm,and UAV image target tracking and positioning system Design and implementation.Specific studies include:(1)Design of UAV target tracking and positioning system framework.Based on the principle of high reliability,high mobility,high precision and low energy consumption,this paper designs the system framework from the hardware and software systems.Which can be effectively applied to the task of real UAV image target tracking and positioning.(2)An improved tracking algorithm based on TLD(Tracking-Learning-Detection)is proposed.In this paper,two methods are proposed to improve the feature selection of TLDs and the problem of too many candidate scan frames in detection.The two modules are based on FAST feature points and Kalman filter based target detection module.Experiments show that the improved TLD algorithm has better tracking performance.(3)An improved tracking algorithm TFD(Tracking-Fusion-Detection)based on CMT(Consensus-based Tracking and Matching of Keypoints for Object Tracking)and STC(Spatio-Temporal Context)is proposed.The TFD algorithm makes full use of the robust tracking of the STC algorithm and the precise detection based on the feature point matching in the CMT algorithm.The new fusion module is responsible for the merging of the results and updating of the tracking and detection module model.Experimental results show that the TFD tracking performance is more robust and accurate.(4)Design of UAV target location program.This paper analyzes and defines UAV model,tracking target motion model,and camera calibration model from UAV,tracking target and airborne camera.To achieve through the image of the target pixel coordinates to the real environment of the geographical coordinates of the positioning process.