With the rapid development of unmanned aerial vehicles (UAV), the indoor positioning of UAV is still a difficult problem. It is of high cost to provide location information in advance, and does not meet the economic benefits. An improved LK optical flow algorithm was proposed for four rotor unmanned aerial vehicle (UAV), in which Shi-Tomasi corner detection algorithm and bilinear interpolation method were used to improve the positioning accuracy and stability of UAV. The algorithm was implemented on the computer. The results show that the improved LK optical flow method has high precision and good stability. UAV can carry on fixed point flight under the stable illumination condition. Fixed point accuracy is a circle with radius 30 cm.