无人机安全着陆是无人机研究的重点和难点，为了提高无人机在公路等简易机场着陆的安全性，提出了基于少量任意分布特征点估计无人机姿态和位置的计算机视觉算法。解算无人机着陆过程中相对跑道的位置和姿态是文章核心；首先利用N点算法解算出特征点在摄像机坐标系上的坐标，然后利用正交化算法求出摄像机坐标系和跑道坐标系之间的旋转矩阵和平移向量；针对特征点的成像过程容易受噪声影响的情况，引入了最小中值法，减小噪声的影响，提高算法的鲁棒性，并解算出姿态和位置；通过卡尔曼滤波方法，进一步提高位置和姿态的精度。仿真结果表明：所提算法满足无人机自主着陆的精度要求。

The safe landing of Unmanned Aerial Vehicles(UAV) is the key for the study on UAVs.In order to improve the safety of UAV landing on the flight strip，a vision assisted method based on a few feature points distributed on the ground arbitrarily was presented for estimating the attitude and position of it.Firstly,the coordinates of the feature points in the camera coordinate system was acquired with the Npoint algorithm.Then,the rotation matrix and the translation vector between the camera coordinate system and the runway coordinate system were acquired by using orthogonalization method.Considering that the imaging process of feature points may be influenced by noise,the leastmediansquares algorithm was introduced to diminish the noise influence and improve the robustness.Kalman filter was used for improving the precision of position and attitude.The simulation indicated that the presented algorithm can meet the precision demand of UAV landing.