基于自适应分辨率的点对特征点云配准算法  

Point-to-feature Point Cloud Registration Algorithm Based on Adaptive Resolution

作　　者：方博 岳晓峰[1] 袁晓磊 马国元 FANG Bo;YUE Xiaofeng;YUAN Xiaolei;MA Guoyuan(School of Mechanical and Electrical Engineering,Changchun University of Technology,Changchun 130012,China)

机构地区：[1]长春工业大学机电工程学院,吉林长春130012

出　　处：《机械制造与自动化》2022年第3期113-116,共4页Machine Building & Automation

基　　金：吉林省发展和改革委员会项目(2020C018-3)。

摘　　要：针对基于原始点对特征的三维目标识别算法中存在的内存浪费及过多冗余点导致的配准效率不高问题,提出一种改进的自适应分辨率点云配准算法。通过使用自适应体素网格滤波器对原始点云进行下采样,有效地减少了目标模型哈希表中存在的大量冗余点对,节省了计算机的内存开销并提高了点云配准算法的配准速度。使用迭代最近点方法进一步对聚类结果进行优化处理。实验结果表明:所用算法与传统ICP算法相比,在配准误差相同的情况下配准效率提升显著。Aimed at the memory waste and the low registration efficiency caused by excessive redundant points in 3D target recognition algorithm based on the original point pair feature,an improved point cloud registration algorithm with adaptive resolution is proposed.An adaptive voxel grid filter is applied to down-sample the original point cloud,which effectively reduces a large number of redundant point pairs in the target model hash table,saves computer memory overhead and improves the point cloud registration algorithm.Iteration closest point is used to further optimize the clustering results.The experimental results show that the proposed algorithm improves registration efficiency significantly under the same registration error in comparison with the traditional ICP algorithm.

关 键 词：自适应分辨率 稠密点云 体素网格滤波器 哈希表 迭代最近点算法 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]