Abstract: Based on the on-ground simulation system of spacecraft rendezvous and docking，a related pose measurement algorithm is proposed to ensure the measurement precise.It is robust to feature extracting，and of greater calculating efficiency.Firstly，extract the features with adaptive threshold method，then by inverse searching，search the first frame in the global scope，while the following with Kalman filter predicting and estimating method.The method is useful to the efficiency and anti-disturbing ability. Using this information，calculate the related pose between the chasing spacecraft and the target with P3P algorithm，and transfer the result to the controlling system to send out correct instruction to accomplish the process.The measurement displays that the searching scope error precise is two pixels，the calculating precise is 0.001 m in length and 0.001° in andle，and the speed with Kalman filter is improved by 63.2%.

Key words: spacecraft rendezvous and docking, feature extracting, Kalman filter, P3P algorithm

摘要： 基于模拟航天器交会对接的地面实验系统，提出一种相对位姿测量算法，在保证测量精度的基础上，对特征提取具有稳健性，同时计算效率更高。首先使用自适应阈值方法对特征进行判断和提取，然后采用倒序搜索的算法，对第一帧在全局范围内进行搜索，后续的图像利用Kalman滤波器进行预测估计搜索，极大地提高了搜索效率和抗干扰能力，再由提取到的特征点信息利用P3P等算法进行追踪航天器和目标航天器的相对位姿计算，将此结果传递给控制系统以使控制系统发出相应的指令。实验结果显示，可缩小搜索区间范围至2像素，计算精度为0.001 m和0.001°，同时计算速度提高63.2%。

关键词: 交会对接, 特征提取, Kalman滤波器, 3P算法