| 基于K-近邻搜索的点云初始配准 |
| |
| 引用本文： | 邢正全,邓喀中,薛继群.基于K-近邻搜索的点云初始配准[J].测绘科学,2013,38(2). |
| |
| 作者姓名： | 邢正全 邓喀中 薛继群 |
| |
| 作者单位： | 中国矿业大学环境与测绘学院/江苏省资源环境信息工程重点实验室,江苏徐州,221116 |
| |
| 基金项目： | 国家自然科学基金项目资助，高等学校博士学科点专项科研基金，中央高校基本科研业务费专项资金资助 |
| |
| 摘 要： | 为了使用最近点迭代算法(ICP)实现点云的精确配准,需要点云有良好的初始姿态,这可以通过点云的粗配准实现。本文结合K-近邻搜索和法向量估计,通过组建不变角度作为匹配特征,求解旋转矩阵和平移向量实现粗配准,方法由Matlab7.1编程实现。具体的实验结果表明,利用该方法能得到理想的粗配准效果,可以进一步应用ICP算法实现精确配准,该方法是有效的。
|
| 关 键 词： | 点云配准 K-近邻搜索 法向量估计 最小二乘拟合 |
Initial registration for point cloud based on K-nearest neighbor search |
| |
| Abstract: | A well initial position is needed when Iterative Closest Point algorithm(ICP) is used to point cloud precise registration,which can be realized through initial registration.This paper combined K-nearest neighbor search and normal vector estimation,constructed invariant angle as matching feature,the rotation matrix and translation vector were obtained,and the initial registration was achieved finall by programming with Matlab7.1.The specific experiment result showed that ideal registration result could be obtained for applying ICP algorithm further in precise registration,thus this method is effective. |
| |
| Keywords: | point cloud registration K-nearest neighbor search normal vector estimation least square fitting |
| 本文献已被 万方数据 等数据库收录！ |
|
