针对在单目视觉目标位姿测量过程中，特征点提取出现离群点的情况，提出一种基于去除离群点策略的位姿测量方法(ORPE)。建立了以特征点误差极大极小为原则的最优化目标函数，通过确定特征点最大观测误差值边界，判定并去除离群点，由此可消除离群点误差对位姿测量的影响。仿真实验使用ORPE对1 m×1 m×1 m的立方体目标进行位姿测量，验证了算法的正确性；使用ORPE测量Boeing飞机模型的位姿，平均姿态角误差2.07°,平均位移误差1.6%。通过和最小二乘测姿法(LSPE)结果对比分析可得ORPE法误差小于LSPE法误差。表明ORPE能有效去除离群点，同时提高位姿测量精度。

Based on the circumstance that the outliers appear sometimes when the pose of target is measured by mono-vision method, a new method is presented for estimating the pose (position & attitude) based on outliers-removal (ORPE). The optimization function is established based on the principle of max-min error. Then the outliers can be detected and removed through eliminating frontier of maximum error. Furthermore the impact of outliers’ error on the accuracy of pose estimation can be eliminated. The pose of 1 m×1 m×1 m cube target is measured by ORPE, which proves the validity of the algorithm. The pose of Boeing plane model is measured by ORPE. As a result, the average attitude error is 2.07°, and the average position error is 1.6%. Comparatively, the average error of ORPE is less than the one of the lest-square pose estimation (LSPE) method. Consequently, it’s demonstrated that ORPE can do outliers-removal effectively, and improve the accuracy of pose estimation simultaneously.