空间站脉冲编码被动式交会对接合作目标三维跟踪

针对空间站交会对接(RVD)过程中合作目标在视频中周期性发光的特点，提出基于累积帧差法的合作目标检测方法及相应的跟踪窗口确定方法，建立合作目标状态的识别机制。在跟踪过程中，针对均值漂移算法容易受到图像背景干扰产生较大跟踪偏差的问题，在光流计算的基础上，提出一种基于线搜索的二维跟踪算法。通过分析敏感器采集的视频，验证了文中提出的算法能够准确地检测和跟踪图像序列中的合作目标，进而结合像机位姿估计算法，连续地估计像机位姿在三维空间中的变化。

Cooperative Target 3D Tracking for Space Station Pulse Coded Passive Rendezvous and Docking

In this paper, a accumulated frame difference based cooperative target detection method and a tracking window establishment method are proposed, meanwhile a framework for cooperative target recognition has been put forward according to the fact that the cooperative target is sparkling periodically in the passive Rendezvous and Docking (RVD) process between two space crafts. Since the traditional Meanshift method can be easily disturbed by image backgrounds in a target tracking process, a line search based two dimensional target tracking method is proposed based on the result of the optical flow calculation. Through analysis of the videos collected by the RVD sensor, it is proved that the proposed methods can be used to accurately detect and track the cooperative target in a video sequence, so that the change of the camera pose in three dimensional space can be continuously estimated by combining the tracking result with a camera pose estimation algorithm.