融合角点检测的点云配准研究
Point Cloud Registration Method Based on Fusion of Corner Detection
宋庆军 1高义强 1姜海燕 1来庆昱 1王晓双1
作者信息
- 1. 山东科技大学智能装备学院,泰安 271000
- 折叠
摘要
为提高点云配准效率,提出一种基于点云曲率信息的Harris角点提取算法,通过引入曲率约束,对传统Harris角点提取做出改进;计算每个点的曲率值并与整幅点云的平均曲率作比较,保留曲率值大于平均曲率的点云,实现对配准价值较低的点的剔除,为后续点云配准减少计算量;构建所提取角点之间的特征匹配,结合采样一致性(sample consensus initial alignment,SAC-IA)粗配准和正态分布变换(normal distributions transform,NDT)精配准,寻找最优的变换矩阵,实现源点云和目标点云的重合.通过比较不同算法在公开数据集上进行实验验证,结果表明该方法对复杂点云、多拐点点云都有较好的表现,对初始位置相差较大的点云也有良好的适用性.
Abstract
In order to improve the registration efficiency of high point cloud,a Harris corner extraction al-gorithm based on the curvature information of point cloud is proposed.By introducing curvature con-straints,the traditional Harris corner extraction is improved.The curvature value of each point is calculated and compared with the average curvature of the whole point cloud.The point clouds whose curvature value is greater than the average curvature are retained to eliminate the points with low registration value and re-duce the calculation amount for subsequent point cloud registration.Construct feature matching between ex-tracted corner points,and combine sample consensus initial alignment(SAC-IA)and normal distributions transform(NDT)with fine registration.Find the optimal transformation matrix to achieve the coincidence of source point cloud and target point cloud.By comparing different algorithms on public data sets,the ex-perimental results show that this method has good performance on complex point clouds and multi-inflection point clouds,and also has good applicability to point clouds with large initial position difference.
关键词
机器人视觉/曲率约束/角点检测/矩阵变换/点云配准Key words
robot vision/curvature constraint/corner detection/matrix transformation/point cloud registration引用本文复制引用
基金项目
国家自然科学基金面上项目(52174145)
山东省自然科学基金面上项目(ZR2020MF101)
出版年
2024
NETL
NSTL
万方数据