双目立体视觉的无人机位姿估计算法及验证

针对无人飞行器在未知复杂环境下的导航问题，提出了一种基于双目立体视觉的无人飞行器位置和姿态估计算法。用双目摄像机采集立体图像序列，对图像进行立体校正后使用Harris算法提取特征角点，用NCC算法获取匹配特征点，导出摄像机坐标系下的特征点坐标，得到三维立体特征信息，使用RANSAC算法与L-M迭代算法得到无人飞行器姿态和位置估计值。实验结果表明，基于双目立体视觉的位姿估计算法能适应未知环境变化，计算结果与实际位姿量相比误差小，能满足无人飞行器导航要求，可为无人飞行器的导航实现提供一套新途径。

Pose estimation algorithm and verification based on binocular stereo vision for unmanned aerial vehicle

To the navigation of Unmanned Aerial Vehicle(UAV) in the complex unknown environment, an algorithm of position and attitude estimation based on the binocular stereo vision is described in this paper. Stereo vision pairs are taken by the binocular camera. The feature points are detected with the Harris algorism after the stereo vision pairs are rectified and the feature points are matched with NCC algorithm. Then the 3D stereo information of the feature points can be calculated in the camera coordinate system and the position and attitude of UAV are estimated with RANSAC algorithm and L-M iteration algorithm. The result of the experiment shows that the position and attitude estimation algorithm based on binocular stereo vision has strong adaptability of the unknown environment. Compared to the real position and attitude, there is a small error. The algorithm can meet the requirements of UAV navigation. A new way can be provided for the UAV navigation.