基于地面激光雷达的单木枝干建模方法

The 3D reconstruction of tree branches from point cloud based on terrestrial laser scanner

基于激光点云数据的高精度活立木建模技术在林业无损测量中受到越来越多的重视,由于树木形态多样且拓扑结构复杂,因此传统的建模方法在针对活立木的建模上效果并不理想。笔者提出了一种针对地面激光雷达点云数据的活立木枝干三维建模方法。首先将采集到的树木点云进行枝叶分离,从原始活立木数据中分割出树叶点云;然后对枝干点云构建基于Laplace算子的加权矩阵,迭代收缩地进行骨架提取,得到枝干的骨架并保留;其次结合改进的随机Hough变换计算树枝的半径;最后运用广义圆柱体围绕枝干骨架绘制树枝几何结构,最终完成三维重建。该方法能直接对活立木枝干进行建模,无需手工交互,对立木无任何影响。通过在校园内采集到的4株活立木进行试验验证,结果表明该算法重建结果精度较高,并能很好地保持植株的细节特征。

The standing trees modeling methods based on the point cloud data with high accuracy in the nondestructive measurements of forestry received more and more attentions. Due to the various shapes and complex topological structures of different trees, the conventional methods in trees modeling always achieve unsatisfactory performance. In this paper, a novel method of 3D reconstruction of tree branches from terrestrial laser radar point cloud data was first proposed. The new method included the following steps：First, branches and leaves were divided apart from point clouds of trees, and the tree leaves were fragmented. After that, the skeletons of branches were extracted by constructing a weighted matrix based on Laplace, and an iterative procedure was accept to solve the Laplace equation. Then, the diameters of branches were calculated by an improved random-Hough arithmetic. Finally, the tree model was established by generalized cylinder, and the three-dimensional reconstruction of tree branches＇ skeletal structures can be generated. The approach is validated to carry on modeling for standing forest stock directly with no manual intervention. We test our proposed method on 4 different kinds of trees on campus. The results showed that the method allows higher precisions of reconstruction with detailed features.