Abstract: According to the problem of target tracking for UAV, a target tracking and locating algorithm based on Camshift algorithm and binocular vision is proposed. The left and right images from binocular camera are used to calculate the central interesting point of the target with Camshift algorithm. After the central interesting point is reconstructed, the relative position and yaw angle between UAV and target under the body coordinate system are got. Data is optimized with the kalman filter. The estimated data is used as the flight control system loopback input data to achieve the autonomous tracking of UAV. The result of the experiment shows that the error of the algorithm is little. This algorithm has strong stability and accuracy.

Key words: Unmanned Aerial Vehicle（UAV）, binocular stereo vision, target tracking, locating

摘要： 针对小型无人飞行器跟踪目标的问题，提出了一种基于双目视觉和Camshift算法的无人飞行器目标跟踪以及定位算法。双目相机得到的左右图像通过Camshift算法处理可得到目标中心特征点，对目标中心特征点进行三维重建，得到机体坐标系下无人飞行器与目标间的相对位置和偏航角，应用卡尔曼滤波算法对测量值进行了优化，将所得估计值作为飞行控制系统的反馈输入值，实现了无人飞行器自主跟踪飞行。结果表明所提算法误差较小，具有较高的稳定性与精确性。

关键词: 无人飞行器, 双目视觉, 目标跟踪, 定位