CAJ | 학술논문

树木骨架是树木可视化及建模的基础，目前的研究中多数利用三维网络模型或体素化模型，缺少直接在点云中提取骨架的研究；特别是在稀疏点云条件下，数据精度不高，树叶与枝干之间存在遮挡，骨架提取变得异常困难。为此，本文提出了一种提取算法，主要包括如下步骤：①通过模糊 c-均值聚类，对离散点云进行分割；②通过 B 样条曲线拟合聚类团质心，形成枝干自由型结构；③利用法矢方向变化率阈值，确定树木节点的位置。将该方法分别应用到地面及机载 LiDAR 数据的树木骨架提取中，得到了较好的效果。
수목골가시수목가시화급건모적기출，목전적연구중다수이용삼유망락모형혹체소화모형，결소직접재점운중제취골가적연구；특별시재희소점운조건하，수거정도불고，수협여지간지간존재차당，골가제취변득이상곤난。위차，본문제출료일충제취산법，주요포괄여하보취：①통과모호 c-균치취류，대리산점운진행분할；②통과 B 양조곡선의합취류단질심，형성지간자유형결구；③이용법시방향변화솔역치，학정수목절점적위치。장해방법분별응용도지면급궤재 LiDAR 수거적수목골가제취중，득도료교호적효과。
Tree skeleton is a basis about virtual trees and modeling.In the current study,most researchers use three dimensional network and subvoxel models,which is lack of the study on point cloud skeleton extraction directly.Especially with sparse point cloud,data accuracy is not high;there is a dense covering between leaves and branches,and it is very difficult for skeleton extraction.Therefore,this paper presented a kind of extraction algorithm,which mainly included the following steps:(1)the division of discrete point cloud based on fuzzy c-means clustering;(2)the centroids of clustering groups were fitted into free-form structures of branches by B-spline curves;(3)the position of branch nodes were detected by the threshold for direction varying ratio of normal vectors.Methods were applied to tree skeleton of ground and airborne LiDAR data,which demonstrated better performance.