摘要
针对树木与树枝分叉部分体积计算困难,三维扫描建模易产生黏连现象等问题,提出一种Alpha-Shapes分段改进算法。通过三维激光扫描技术获取三维模拟树枝的点云数据,利用Alpha-Shapes算法计算出三维模拟树枝的点云边缘轮廓并进行三维重建,利用分段算法对三维重建数据进行体积分割计算,通过点云最高层补偿以此保证分段算法的准确度。最后用Alpha-Shapes算法计算各个分段体积并进行累加。实验使用标准塑料直管段、三通管段和四通管段对树干以及树枝的分叉情况进行模拟并通过扫描真实树枝进行验证,通过三维激光扫描技术获取点云数据,实验充分考虑树枝分叉情况、检测精度、黏连问题,结合不同模拟树枝分叉特点,对比选择最合适的半径α值与分段高度β值。实验表明该改进算法的体积计算误差值在4%-5%之间,较改进前降低5%。
It is difficult to calculate the volume of branch and branch,and the three-dimensional scanning model is easy to produce adhesion problems.An improved Alpha-Shapes segmentation algorithm was proposed.The 3D laser scanning technology was used to obtain the point cloud data of the 3D simulation branch,and the Alpha-Shapes algorithm was used to calculate the point cloud edge contour of the 3D simulation branch and perform 3D reconstruction.The segmentation algorithm is used to calculate the volume division of the 3D reconstruction data through the point cloud.The highest level of compensation ensures the accuracy of the segmentation algorithm.Finally,the Alpha-Shapes algorithm is used to calculate the volume of each segment and accumulate it.The experiment uses the standard plastic straight pipe section,the three-way pipe section and the four-way pipe section to simulate the branching of the trunk and the branches and verify by scanning the real branches.The point cloud data is obtained by the three-dimensional laser scanning technology,and the experiment fully considers the branch bifurcation.The detection accuracy and adhesion problem,combined with the characteristics of different simulated branching forks,compare and select the most suitable radius α value and segment height β value.Experiments show that the volumetric calculation error of the improved algorithm is between 4%and 5%,which is about 5%lower than before the improvement.the improvement.
作者
张鹤
李东升
陈爱军
ZHANG He;LI Dongsheng;CHEN Aijun(College of Metrology&Measurement Engineering,China Jiliang University,Hangzhou 310018,China)
出处
《中国测试》
CAS
北大核心
2021年第3期49-58,共10页
China Measurement & Test
基金
国家自然科学基金(61471330)。
