Research On Precise Localization Algorithm For Ground Target Based On Vision-based UAV

In recent years,drones have received wide attention in various fields because of their unique advantages.The tasks performed by drones are mostly based on the target positioning technology.Therefore,it is of practical significance to study drones on ground target localization algorithms.Firstly,we study the target detection algorithm using visual methods to obtain the pixel coordinates of the center point of the image.To locate the target on the ground,we must first use the on-board sensors to measure the relative position information between the target and the drone.The system uses a small quadrotor drone equipped with a camera.It needs to collect and process images on the ground target.,to get the pixel coordinates of the target point in the image,so as to indirectly measure the target relative drone position information.Secondly,this dissertation studies the single target UAV ground target localization algorithm based on coordinate transformation.At present,the commonly used target location methods are based on the single station target location method based on coordinate transformation.In order to determine the main error sources in the target location process,the target location solution process of the algorithm is given,and the target location is established.The mathematical model of the error,and through the mathematical simulation to verify the impact of each error source on the accuracy of the positioning results for the subsequent improvement of the algorithm to improve the direction.Thirdly,according to the main error sources of the target location process,the method to reduce the influence of error sources from the algorithm is studied.In order to reduce the adverse effect of the main error sources on the accuracy of the target location process,the distance location algorithm and the single station vision location method of the spatial three-point intersection were studied,and the target location scheme for multiple targets of the same target point was given.,and through the simulation to verify the improved accuracy to improve the analysis of the advantages and disadvantages of the algorithm and the scope of application;and then for a small four-rotor drone platform without a pod,given a multi-point measurement using visual angle measurement Observing the target location method,establishing the system's state equation and observation equation,using unscented Kalman filter to estimate the target position,and verifying the feasibility of the positioning method through mathematical simulation.Finally,the experimental research on the vision-based multi-point angle observation target positioning method based on the four-rotor UAV independently developed by the laboratory is carried out.The experimental results verify the positioning effect of this positioning method,and error analysis of the experimental results is performed.The future direction of the positioning algorithm is summarized and prospected.