点云的刚体运动参数估计方法的比较

点云配准是使不同视角下的点云数据实现正确拼接,解算三维空间刚体平移和旋转参数是点云配准中的核心问题。文中针对刚体运动参数求解问题,分析现有的奇异值分解法(SVD)、正交分解法(OD)、单位四元数法(UQD)、对偶四元数法(DQD)4种不同的估计方法,通过模拟实验及实例展开分析,探讨各自的优缺点及适用性,结合实例结果表明,对偶四元数求解刚体运动参数的总体性能最优,在实际应用中可以优先使用对偶四元数进行刚体变换参数求解。

Comparison of parameter estimation methods for rigid motion of point cloud

The point cloud registration is to make the point cloud data from different visual angles to be correctly spliced.Calculating the three-dimensional rigid body translation and rotation parameters is the core problem in point cloud registration.To solve the problem of rigid motion parameters,the four estimation methods of singular value decomposition method(SVD),orthogonal decomposition method (OD),unit quaternion(UQD)and dual quaternion method(DQD)are analyzed in this paper.Through simulation experiments and examples,their ow n advantages and disadvantages and applicability are discussed.T he result show s that the overall performance of the dual quaternion is optimized.T he parameters of the rigid body transformation can be solved by dual quaternion in the practice application.