Research On Fast Detection And Stability Tracking Method Of Rotor UAV To Ground Moving Target

Dynamic target detection and tracking technology in computer vision is a hot topic in current research.With the maturity of drone technology,its unique mobility and flexibility are welcomed by scholars at home and abroad.Applying computer vision technology to drones is a new challenge.Based on the background of the civil aircraft inspection project and the seventh generation of the international air robot competition,this paper studies the vision-based moving target detection and image servo-based four-rotor UAV tracking method.Firstly,according to the task requirements and rules,design the hardware scheme of the four-rotor UAV dynamic target tracking system,determine the hardware selection of the video acquisition and processing module and the UAV flight controller;optimize the design target tracking overall system software solution.Determine the processing flow of the image acquisition and processing algorithm and the communication method with the flight control module.Secondly,aiming at the target detection failure caused by environmental factors such as ground dynamic target state change and UAV jitter,the method of Gaussian mixture model fusion shape feature is proposed to identify the ground target and solve the pose information of the target relative drone..At the same time,according to the landmark information of the competition venue,the Hoff line detection algorithm combined with HSV color space is designed to detect the yaw angle of the drone,correct the motion direction of the ground target,and detect the unmanned by combining the boundary detection algorithm of HSV color space and contour feature.Whether the machine leaves the boundary of the site.Thirdly,the motion of the quadrotor UAV system is analyzed,and an image-based visual servo tracking controller is designed.The deviation between the center position of the dynamic target pixel and the center position of the image pixel is taken as the input of the controller,and the control data is calculated to complete the design of the PID tracking controller based on the image sequence.Finally,aiming at the requirements of the four-rotor UAV’s dynamic target fast detection and stable tracking task,a four-rotor UAV visual tracking software system was built to complete the target detection,boundary detection and target tracking experiments.The experimental results show that the system can successfully complete the fast recognition and stable tracking ability of dynamic targets,and verify the stability and effectiveness of the system.