未标定相机的高精度位姿估计方法

An Accurate Algorithm for Uncalibrated Camera Pose Estimation

摘要:

针对未标定相机的位姿估计问题，提出了一种焦距和位姿同时迭代的高精度位姿估计算法。现有的未标定相机的位姿估计算法是焦距和相机位姿单独求解，焦距估计精度较差。提出的算法首先通过现有算法得到相机焦距和位姿的初始参数；然后在正交迭代的基础上推导了焦距和位姿最小化函数，将焦距和位姿同时作为初始值进行迭代计算；最后得到高精度的焦距和位姿参数。仿真实验表明提出的算法在点数为10，噪声标准差为2的情况下，角度相对误差小于1%，平移相对误差小于4%，焦距相对误差小于3%；真实实验表明提出的算法与棋盘标定方法的精度相当。与现有算法相比，能够对未标定相机进行高精度的焦距和位姿估计。

Abstract:

Objective: For solving uncalibrated camera pose problem, a novel accurate pose estimation algorithm is proposed based on simultaneous iteration of focal length and pose. Method: First, initial parameters are obtained by the existing algorithms; then, focal length and the position and orientation parameters are as iterative values all together; finally, accurate focal length and pose estimation are obtained. Result: Simulative data and real images are processed by different algorithm. Experimental results show that the proposed algorithm is capable of estimating uncalibrated camera high-precision focal length and pose. Conclusion: An accurate algorithm for uncalibrated camera pose estimation is proposed. Experimental results with simulative data and real images indicate that the precision of proposed algorithm is better than existing methods .