Research And Design Of UAV Autonomous Landing System Based On Image Recognition

Unmanned aerial vehicles(UAVs)have attracted general interest in their ability to carry out missions such as persistent surveillance,hotspot patrolling,urban traffic monitoring and searchrescue operations.These assignments require UAVs to have the abilities to detect targets,as well as long-term stability.The development of image processing,computer vision technology and small UAVs have promoted the development of visually capable UAVs.Therefore,it is necessary to research and design UAV automatic landing system based on image recognition.This paper presents a real-time,UAV landing algorithm based on landing mark recognition and pose estimation.On the one hand,the landing algorithm consists of algorithms used to acquire visual targets(UAV aprons),and then navigate to the target.The target recognition part utilizes the color,shape and local corner features of the landing target to achieve accurate detection.On the other hand,the signal between the UAV and the control system is transmitted by private mobile network.This paper combines image processing and global positioning system navigation for accurate target detection and identification.The UAV updates its landing target parameters almost in real time and uses these parameters to track the path of the landing point.Finally,the system was tested and the results indicate that image identification algorithm based on multi-feature fusion and SURF can identify target quickly and accurately.The mission of autonomously landing on a given target is complete and specific target is tracked accurately.