基于改进ICP算法的三维点云刚体配准方法

Rigid registration method of 3D point cloud basedon improved ICP algorithm

针对含有噪声和外点的三维点云刚体配准问题,由于迭代最近点(iterative closest point,ICP)算法的配准精度较低,为此,该文提出了一种基于改进ICP算法的三维点云刚体配准方法。考虑到伪Huber损失函数对噪声和外点不敏感、鲁棒性强,首先,建立了基于伪Huber损失函数的三维点云刚体配准模型。其次,利用RGB-D点云数据中颜色信息辅助建立点云对应关系,以提高改进ICP算法中对应点匹配的准确性。最后,结合奇异值分解(singular value decomposition,SVD)和Levenberg-Marquardt(LM)的优化算法对三维点云刚体配准模型进行优化求解。实验结果表明,该文所提三维点云刚体配准方法的配准精度高,能够有效抑制噪声和外点对配准精度的影响。

Aiming at the problem of rigid registration of three-dimensional(3D)point cloud with noise and outliers,due to the low registration accuracy of the iterative closest point(ICP)algorithm,a rigid registration method of 3D point cloud based on improved ICP algorithm is proposed in this paper.Firstly,considering that the pseudo Huber loss function is insensitive to noise and outliers,and has strong robustness,a 3D point cloud rigid registration model based on pseudo Huber loss function is established.Secondly,in order to improve the matching accuracy of the corresponding points in the improved ICP algorithm,color information of RGB-D point cloud data is used to assist in establishing the corresponding relationship between point clouds.Finally,singular value decomposition(SVD)and Levenberg-Marquardt(LM)optimization algorithms are combined to optimize the 3D point cloud rigid registration model.Experimental results show that the proposed rigid registration method of 3D point cloud can ensure high registration accuracy and effectively suppress the influence of noise and outliers on the registration accuracy as well.