基于邻域曲率改进的迭代最近点激光雷达目标点云配准

Lidar Target Point Cloud Alignment Based on Improved Neighborhood Curvature with Iteration Closest Point Algorithm

为解决激光雷达目标点云配准技术中精确配准步骤中所存在的匹配速度慢和匹配误差大的问题,提出了一种基于邻域曲率改进的迭代最近点(ICP)精准化匹配算法。初始配准采用传统的主成分贴合法,给精确配准找到一个较好的初始位置,精配准采用基于领域曲率改进的ICP算法。以斯坦福兔子和场景点云作为实验研究对象,配准结果和数值分析共同表明,基于邻域曲率改进的ICP算法在点云配准中的可行性,且与其他算法相比,所提算法的配准速度更快、匹配精度更高,为三维数据重建和目标识别技术提供一种更高效的新方法。

To solve the problems of slow matching speed and large matching error in the precise alignment step of lidar target point cloud alignment technology,an iteration closest point(ICP) precision matching algorithm based on neighborhood curvature improvement is proposed.The registration provides a good initial position;the neighborhood curvature is introduced into the traditional ICP algorithm to achieve the fine registration.Perform registration and numerical analysis experiments on the Stanford Bunny and the scene point cloud.The experimental results demonstrate that the improved ICP algorithm based on the neighborhood curvature can efficiently perform the point cloud alignment,and compared with other algorithms,the alignment speed of the proposed algorithm is better than the alignment matching accuracy,which provides an efficient method to improve the 3D reconstruction and target recognition technology.