Research On Image Processing Technology Of UAV Autonomous And Precise Landing

Recently,Multi-rotor UAV has gradually become a hot spot in the field of civil/commercial and research because of its small size,light weight,high flexibility,concealment and other advantages.And it is essential to realize the independent and precise landing function of UAV for its general application in inspection,mapping,monitoring,express delivery and other industries and completing tasks efficiently and completely.However,UAV can just obtain the rough location information of the target through radar detection and GPS positioning technology,but can not achieve accurate landing.So it is necessary to study about image processing technology based on the method of detection,recognition and tracking of landing target.Then we can figure out the relative position between UAV and landing target,and finally UAV can achieve precise landing independently.This article studies the UAV landing algorithm based on target detection,recognition and tracking technology.The content mainly includes image preprocessing,detection and recognition of landing target,coding of landing signs,tracking of moving targets in the view of the platform,and self-correction of tracking results.The main work and related contributions of this article are:(1)This paper studies two kinds of detection and recognition algorithms,which are based on Aruco mark or classifier.Two different types of landing marks are tested and analyzed to show using Aruco mark as landing mark can improve the accuracy and flexibility of UAV in the process of autonomous landing.At the same time,the advantages and disadvantages of using two detection and recognition algorithms in the process of landing are analyzed.(2)In order to achieve autonomous and precise landing,UAV needs to locate the landing target accurately and in real time.This paper studies the traditional target tracking algorithm,including TLD algorithm combined with target detection and tracking,KCF algorithm based on kernel correlation filter,and carries out a simple algorithm combined with the characteristics of the above algorithm,.(3)In this paper,a target tracking algorithm based on TLD framework and improved KCF is proposed to weaken the influence of illumination,movement,occlusion,size change and other factors on the tracking results in the landing process.The algorithm combines KCF and TLD algorithm,and improves the KCF algorithm by achieving feature fusion and scale adaptive,The accuracy and robustness of the tracking module in the landing process are improved as much as possible.(4)In view of the difficulties in the application of the algorithm in the actual landing process,the self correction of the tracking results is added to improve the accuracy of the tracking algorithm based on the TLD framework and the improved KCF target tracking algorithm.At the same time,the final algorithm is tested and compared with the original TLD algorithm,KCF algorithm and two fusion algorithms,It can be seen that compared with the overall algorithm proposed in this paper,the final algorithm has higher real-time performance and accuracy,and has stronger adaptability and robustness in the actual application process.So the autonomous landing accuracy of UAV can reach centimeter level.