摘要
文章在阐述不同视角下在对一对三维点云数据集两两配准的基础之上,针对ICP精确匹配算法须使初始点云收敛否则无法获取准确匹配结果的问题,提出了基于FPFH特征描述子的特征点云粗匹配;调整两片点云的初始位置,为ICP算法提供了良好的初始位置进一步提高点云的匹配精度;并且在此基础上通过大量实验得到点云对应点对之间的最大距离与拟合系数的函数关系,得到粗匹配最优值,进而得到最佳配准效果。实验证明通过粗匹配最高能将匹配拟合系数提高60.3%。
The article describes a different perspective on the basis of three-dimensional point cloud pairwise data registration,for accurate matching algorithms ICP need to make an initial point cloud convergence,otherwise unable to obtain accurate matching results,this artice proposes a feature based on FPFH Descriptors coarse matching feature method.Adjust the initial position of two point clouds for ICP algorithm provides a good initial position to further improve the matching precision point cloud.And on this basis to get through a lot of experiments as a function of the largest point distance and the best fitness score between correspondence pairs to obtain the optimal value match,it is proved that coarse matching can make the best fitness score up to 60.3％.
出处
《计算机测量与控制》
2016年第7期232-236,共5页
Computer Measurement &Control
基金
公安部重点实验室项目(2014GABJC02)
关键词
两两配准
FPFH特征
粗匹配
registration
FPFH descriptors
coarse matching
