Abstract: This paper presents a new method for an estimation of the axis of rotation based on line element geometry and kinematic equation.The method first projects the points of 3-D space to line element space and approximates the kinematic equation on it then gets the axis by the parameter witch can fit the equation.And this paper also presents the klo-RANSAC technique，to extract rotational surface from the noise and the back ground data，which improves the performance and robustness greatly.Experiments show that the proposed method can recognize and extract the rotational surface from 3-D cloud data and estimate the axis of them with great efficiency.

Key words: cloud points, line element geometry, rotational, axis of rotation

摘要： 提出了一种基于线元几何与运动方程的旋转体点云数据特征分离与旋转轴提取算法。该算法将三维空间中的点云数据投影到线元空间中，通过拟合运动方程并分析运动特征参数而计算出旋转轴位置。算法使用基于特征约束的klo-RANSAC（k-local-RANSAC）算法将特征点从自由曲面背景以及噪声中提取和分离，进一步提高了算法的收敛速度与健壮性。实验证明，本算法无需精确估算曲面法矢即能快速地从海量点云数据中迅速地识别与分离具有旋转特征的数据点并提取其旋转轴特征。

关键词: 点云, 线元几何, 旋转面, 旋转轴