In order to make the rotor UAV land on the ground platform quickly and accurately, a pose estimation method based on visual mark detection was proposed. Firstly, based on the geometric features of the standard helipad, a five-step landmark extraction algorithm was used to obtain the visual landmarks from the images captured by the airborne camera. In order to satisfy the rapid and real-time requirements of UAV autonomous landing process, a distance-based three-point corner detection algorithm was proposed, and 12 H-shaped corners were obtained. Then, by classifying and numbering the corners, the corners of the visual mark in the current image were matched with the corresponding corners in the reference image, and the homography matrix containing the pose information was calculated. Finally, the attitude angle of UAV was obtained by decomposing homography matrix with direct linear transformation (DLT), and the position of UAV relative to visual mark was calculated according to the similar triangle formed by camera imaging. The real-time performance and accuracy of the proposed methodology were proved by simulating the pose of UAV in different flight states on an experiment platform. The outcomes showed that the average running time of the given algorithm was 307.2ms, the maximum root mean square error (RMSE) of position estimation was 0.0062m, and the maximum RMSE of attitude angle estimation was 0.313°, which can satisfy the requirements of accuracy and real-time in the autonomous landing process.