College of Mechanical and Electrical Engineering,Northeast Forestry University,Harbin 150040,China
Show less
文章历史+
Received
Accepted
Published
2025-07-23
Issue Date
2026-05-13
PDF (4093K)
摘要
单木骨架提取是树木三维建模的关键步骤,对于精准管理林业和森林资源具有重要意义。背包式激光雷达(backpack LiDAR scanning,BLS)作为一种新兴的移动测量技术,具有灵活性和便携性优势,但其点云数据存在不均匀分布和噪声干扰等问题,影响骨架提取的精度。针对这些问题,以广西壮族自治区国有高峰林场的杉木(Cunninghamia lanceolata)为研究对象,基于背包式激光雷达扫描数据,提出一种基于关键路径探测的分层递进骨架提取方法。该方法结合几何约束与层级分析方法,实现枝干主轴的精准定位,并利用中垂线交点计算构建连续且拓扑完整的单木骨架。采用地基激光雷达(terrestrial laser scanning,TLS)数据作为验证数据,通过体素滤波和局部高程归一化等预处理技术优化背包式激光雷达数据质量。结果表明,在枝干分级评估中,该方法表现出较高的性能。F1分数在0.771~0.788,精确度范围为93.33%~100%,召回率范围为66.67%~90.63%。此外,对BLS数据分枝角度的估测结果与TLS数据分枝角度对比显示,决定系数R2 (coefficient of determination)达到0.84,均方根误差(root mean square error,RMSE)为7.22°。研究结果为单木三维建模提供高精度的技术框架,为林业资源管理、生态模拟等奠定数据基础。
Abstract
Single-tree skeleton extraction is a critical step in 3D tree modeling, holding significant importance for precision forestry and forest resource management. Backpack LiDAR scanning (BLS), as an emerging mobile measurement technology, offers advantages in flexibility and portability. However, its point cloud data suffers from uneven distribution and noise interference, which affect the accuracy of skeleton extraction. To address these issues, this study focuses on Cunninghamia lanceolata in the State-owned Gaofeng Forest Farm of Guangxi Zhuang Autonomous Region and proposes a hierarchical progressive skeleton extraction method based on key path detection using BLS data. This approach integrates geometric constraints and hierarchical analysis to achieve precise localization of branch axes, while employing perpendicular bisector intersection calculations to construct a continuous and topologically complete single-tree skeleton. Terrestrial laser scanning (TLS) data was used as validation data, and preprocessing techniques such as voxel filtering and local elevation normalization are applied to enhance BLS data quality. The results indicate that the proposed method exhibited high performance in branch classification. F1-scores ranged from 0.771 to 0.788, with precision ranging from 93.33% to 100%, and recall ranging from 66.67% to 90.63%. Furthermore, the comparative analysis of branch angle estimations based on BLS and TLS data yields a coefficient of determination (R²) of 0.84 and a root mean square error (RMSE) of 7.22°. This study provides a high-precision technical framework for single-tree 3D modeling, laying a data foundation for forest resource management and ecological simulation.
YANGY Z.Tree point cloud data processing and 3D modeling based on terrestrial LiDAR[D].Harbin:Northeast Forestry University,2020.
[3]
蔡越.基于地面LiDAR的单株毛竹地上生物量测算方法研究[D].杭州:浙江农林大学,2018.
[4]
CAIY.Study on the method of estimating aboveground biomass of Phyllostachys eudlis based on terrestrial LiDAR[D].Hangzhou:Zhejiang A&F University,2018.
[5]
BUIZERM, HUMPHREYSD, DE JONGW.Climate change and deforestation:The evolution of an intersecting policy domain[J].Environmental Science & Policy,2014,35:1-11.
LUJ, LIUX Z, MENGW L,et al.Methodology of individual tree 3D reconstruction based on terrestrial laser scanning point cloud data[J].Journal of Nanjing Forestry University (Natural Sciences Edition),2021,45(6):193-199.
[8]
孟园.基于地面三维激光扫描技术的树冠参数的估测研究[D].哈尔滨:东北林业大学,2017.
[9]
MENGY.Estimation of tree crown parameters based on terrestrial 3D laser scanning technology[D].Harbin:Northeast Forestry University,2017.
ZHAOY, YUX X, XINZ B,et al.Application and outlook of terrestrial 3D laser scanning technology in forestry[J].World Forestry Research,2010,23(4):41-45.
[12]
HUANGY, YUB L, ZHOUJ H,et al.Toward automatic estimation of urban green volume using airborne LiDAR data and high resolution remote sensing images[J].Frontiers of Earth Science,2013,7(1):43-54.
[13]
KOCHB, HEYDERU, WEINACKERH.Detection of individual tree crowns in airborne LiDAR data[J].Photogrammetric Engineering & Remote Sensing,2006,72(4):357-363.
LIUF, TANC, ZHANGG,et al.Estimation of forest parameter and biomass for individual pine trees using airborne LiDAR[J].Transactions of the Chinese Society for Agricultural Machinery,2013,44(9):219-224,242.
[16]
ZHAOG, SHIY, WANGM,et al.Rapid reconstruction of tree skeleton based on voxel space[C]//2015 International Conference on Electrical,Electronics and Mechatronics.Paris,France:Atlantis Press,2015.
[17]
JINX, KIMJ.A 3D skeletonization algorithm for 3D mesh models using a partial parallel 3D thinning algorithm and 3D skeleton correcting algorithm[J].Applied Sciences,2017,7(2):139.
[18]
CAOJ J, TAGLIASACCHIA, OLSONM,et al.Point cloud skeletons via Laplacian based contraction[C]//2010 Shape Modeling International Conference.Aix-en-Provence,France,2010.
[19]
ZHOUJ L, LIUJ, ZHANGM.Curve skeleton extraction via k-nearest-neighbors based contraction[J].International Journal of Applied Mathematics and Computer Science,2020,30(1):123-132.
[20]
JIANGA L, LIUJ, ZHOUJ L,et al.Skeleton extraction from point clouds of trees with complex branches via graph contraction[J].The Visual Computer,2021,37(8):2235-2251.
ZHOUG Y, XUY, CHENN,et al.Method for constructing tree skeleton based on point cloud data[J].Geomatics & Spatial Information Technology,2015,38(9):174-176.
LIR H, CHENY N, GANX Z,et al.Tree-skeleton generation method by thinning voxels of point cloud[J].Laser & Optoelectronics Progress,2019,56(19):245-254.
[27]
潘周.基于三维点云的单株阔叶树可视化模拟研究[D].南京:南京林业大学,2020.
[28]
PANZ.The research on single broad-leaved tree visualization simulation based on three-dimensional point cloud[D].Nanjing:Nanjing Forestry University,2020.
SHIY.Key technologies of apple tree canopy illumination distribution and growth process digitization based on point cloud[D].Yangling:Northwest A & F University,2019.
[31]
尤磊.基于点云数据的树干干形测量[D].北京:中国林业科学研究院,2016.
[32]
YOUL.Stem form measurement based on point cloud data[D].Beijing:Chinese Academy of Forestry,2016.
[33]
XUH, GOSSETTN, CHENB Q.Knowledge and heuristic-based modeling of laser-scanned trees[J].ACM Transactions on Graphics,2007,26(4):19.
[34]
HEG Z, YANGJ, BEHNKES.Research on geometric features and point cloud properties for tree skeleton extraction[J].Personal and Ubiquitous Computing,2018,22(5):903-910.