Research On Vision-based Drogue Tracking Methods

Autonomous air-to-air refuelling(AAAR)can relieve stress for the pilot of manned aircraft and extend working hours for unmanned aerial vehicle(UAV).The drogue tracking system is the key enabling technology to realize the autonomous aerial refueling.Firstly,in this paper,a visual tracking system is built,the basic function of the drogue tracking including detection,tracking and locating is implemented.According to the analysis of the characteristics and difficulties in the scene of tracking,the correlation filter-based tracking algorithms and their improved algorithms are applied to the drogue tracking problem,and the performance of the algorithms are evaluated.Specific research contents are as follows:1.Research of the drogue tracking system.The overall drogue tracking system is divided into detection,tracking,positioning and other modules.Based on the investigation of detection and tracking algorithms,some effective algorithms are selected to achieve the function of a drogue tracking system.After adding the module of re-detection and phase transition methods,the basic function of a drogue tracking system is implemented.2.Study on the tracking algorithms.In order to improve performance of the drogue tracking system,the scale adaptation,image pyramids and other methods are added to the original ICLK algorithms.Then,the correlation filter-based tracking algorithms are analyzed.By adding the color histogram model,an improved algorithm is proposed based on the ECO algorithm.The algorithm can reduce model drift and improve the robustness on dealing with complex scenes.3.Software design and function realization of the drogue tracking system.First,the user interface,detection tracking and positioning system modules are designed and developed.Then,the intrinsic parameters of the camera are calibrated to realize the complete function of the software system.4.Experimental research.Through making datasets of drogue in different scenarios and designing the evaluating methods,the performance of improved algorithm is tested and evaluated.These analyses can provide an important reference for the design of the actual autonomous aerial refueling tracking system.