论文部分内容阅读
提出了一种多视角林地地面激光点云自动化配准方法。首先,计算地面激光点云在不同高程的切片数据,并对切片数据进行自适应距离聚类和圆柱拟合,保留满足圆柱拟合条件的分割段,利用保留下来的分割段提取树干,计算树干跟地面的交点,组成初始的特征三角形;然后,根据相似性测度计算相邻两站的同名三角形对;最后,利用基于最小生成树思想的多站匹配策略实现多视角地面激光点云的自动化配准。实验结果表明,该方法成功地提高了点云自动化配准的精度和效率。
A multi-view forest ground laser point cloud automatic registration method is proposed. Firstly, slice data of ground laser point cloud at different elevations are calculated, and the segment data are adaptively clustered by distance and cylinder fitting, the segment which satisfies the cylinder fitting condition is reserved, and the trunk is extracted by the reserved segment, and the trunk is calculated And the intersection of the ground with the ground to form the initial characteristic triangle. Then, the triangles with the same name of two adjacent stations are calculated according to the similarity measure. Finally, the multi-station matching strategy based on the minimum spanning tree is used to automate the multi-view ground laser point cloud quasi. Experimental results show that this method successfully improves the accuracy and efficiency of point cloud automated registration.