For the problem of positioning hovering under weak or missing GPS signals, this paper combines the improved ORB algorithm with the characteristic optical flow algorithm to estimate the optical flow of the aircraft for the shortcomings of the assumption that the traditional sparse optical flow method is difficult to satisfy. The fusion of the optical flow information data with the inertial navigation data of the aircraft eliminate the erroneous estimation of the speed of the aircraft due to the jitter of the aircraft itself and improve the attitude stability of the indoor aircraft. The feasibility of the method presented in this paper is verified by a self-built experimental platform.