UAV Object Recognition And Tracking

In recent years,due to its simple structure,the high degree of freedom,high maneuverability,and stable flight capability,multi-rotor UAVs have become more and more popular in disaster relief field and civil field.In view of this,this paper studies the design of a visual tracking system based on multi-rotor UAV,including target recognition and location algorithm in video stream and position control algorithm between UAV and target.Finally,the UAV and ground station are built to verify the design.The main contents of this paper are as follows:By studying the basic theory and common algorithms of image denoising,this paper proposes to use image restoration algorithm to deblur the collected video frames in the early stage of image processing.In order to solve the defocus blurring caused by the failure of UAV camera focusing or the motion blurring caused by the relative motion of UAV and target.The PSNR value is used to evaluate the experimental results.Automatic target search.Image deblurring keeps the edge information of the target well and provides a low loss image for target search.The template matching method is used to locate the target in the image token by UAV.The correlation coefficient matching method can correctly locate the target position.This method has achieved good results.Improve the KCF algorithm.The long-term tracking effect of target anti-occlusion is achieved by adding the target template.The template is used for similarity calculation with the target located,so as to judge the state of the target.Target state is divided into: full occlusion,semi-occlusion andnon-occlusion.The anti-occlusion experiment results show that the improved algorithm can accomplish the given tasks and perform well.Object following.Learn about bebop_autonomy UAV autopilot software platform and how to control UAV attitude by sending command messages.The most important aspect of attitude control is the distance between target and UAV.In this paper,a reference tracking distance of 5 m is set.The rotation degree of UAV around yaw axis is controlled by the angle between the yaw axis of UAV and the center of target marking frame.The angle between pitch axis of UAV and the center of target marking frame controls whether the UAV starts tracking or not.Finally,the UAV target recognition and tracking system are tested.The test results show that the proposed scheme can accomplish certain tasks,including video transmission and signal transmission in Wi-Fi signal transmission process;image restoration and target identification and tracking in image processing;automatic searching target in tracking phase,turn on the improved KCF algorithm and keeping the distance between target and UAV.The system designed in this paper has a good performance in tracking effect,can achieve "anti-occlusion" long-term tracking,and can take account of both data processing speed and real-time.