基于点云数据的树木骨架提取算法

打开文本图片集

DOI：10.13733/j. jcam.issn.2095-5553.2026.03.010

中图分类号：S758；TP391.4 文献标识码：A 文章编号：2095-5553（2026）03-0068-07

Tree skeleton extraction algorithm based on point cloud data

Ren Lisheng'²，Wang Lei1'²，Wang Fang1 ² （1. College of Information Science and Technology， Hebei Agricultural University， Baoding，O71oo1，China; 2. Hebei Key Laboratory of Agricultural Big Data，Baoding，O7lool，China）

Abstract：Treepoint cloud models areofvital importancefor aplicationsin tree conservation，growth monitoring， anddigital twinconstruction.Traditional point cloudskeletonextractionalgorithmsoftenencounterdificultiesin accuratelycapturing thebifurcation structuresand terminal branch extremitiesof tree point clouds.Toaddress these limitations，thispaperpresentsan improved pointcloud-based skeleton extractionalgorithm fortres.The proposed method first employs a modifiedLl—median skeleton extraction algorithm incorporating K—means clustering toidentify branchcenterpoints.Subsequently，convex hullalgorithmsareutilizedtoextracttheouterboundarypointsof the tree structure.Furthermore，a novel nearest-point search algorithm based on these boundarypoints is introduced to accuratelydetect bifurcation points.Thefinal skeletalpointcloudisobtainedby integrating theseextractedcenter points，bifurcation pointsandboundary points.The results demonstrate that theskeleton points extracted by this algorithmexhibitsuperiorcorrespondencewith theoriginal pointcloud，asevidencedbyquantitativeevaluations using Hausdorffdistanceandchamferdistance metrics.Comparedwith traditionalmethods，theskeleton points extracted bythis algorithm，after normalization processing，havea Hausdorffdistance and chamfer distance of less than O.5，and overallperformance issuperiortothecomparisonalgorithm.Additionall，theimplementationofcubic Beziercurves forconecting the skeletal pointseffectively preserves theoriginal topologicalcharacteristicsof the tree structure， thereby significantly improving both the accuracy and completeness of the skeleton extraction process.

Keywords：tree skeleton；point cloud data；convex hull algorithm；voxel

0 引言

激光雷达在获取树木的点云信息方面有着很大的优势[，在林业领域，三维激光雷达可以准确、迅速地获得林木表面3D信息。（剩余8505字）