In order to solve the beacon point detection and matching problem in unmanned aerial vehicle (UAV) probe and drogue refueling,a near-infrared beacon detection system was designed,and a robust beacon point matching algorithm was proposed.The near-infrared beacon detection system was constituted with eight 850 nm wavelength near-infrared Light-Emitting Diode (LED) lights arranged uniformly in outer ring of the refueling drogue,and 850 nm band-pass filter integrated with Logitech C310 camera.To solve the mismatch problem caused by interference points,a beacon point matching algorithm based on geometric feature constraint was presented,which can accomplish the far interference point exclusion,feature point sorting,near interference point exclusion and feature point marking.The algorithm performance was validated through experiments in various complex environments.The results show that the beacon detection system has the advantages of high detection accuracy and reliability,and the beacon point matching algorithm can realize exact matching of beacon points robustly in various complex environments.Experimental results of relative position calculation show that the calculation is of high accuracy.