基于点云数据的树木三维重建方法改进

发布时间：2018-02-05 02:34

本文关键词： 树木三维重建激光点云主方向分割骨架点　出处：《农业机械学报》2017年02期 　论文类型：期刊论文

【摘要】：激光点云数据以其详尽、高精度的三维信息,在森林参数估算、精确重建植物形态结构三维模型方面具有特殊优势。为进一步提高三维模型精度,综合集成多种算法,在改进现有PC2Tree软件基础上,基于点云数据进行树木三维重建。首先根据树木局部点云的主方向相似度和局部点云轴向分布密度分离枝干与树叶;其次采取水平集和最小二乘法提取枝干部分的骨架点,通过下采样方法提取冠层部分的特征点;最后根据骨架点和特征点拓扑结构重构树木三维模型。以樟树为例,分析枝叶分割精度,自动分割与手动分割结果相近;以无叶的鸡蛋花树为例,分析重建模型精度,模型主枝长度相对误差范围集中在0~8.0%,半径相对误差范围集中在0~10%;枝条重建过程避免了噪声点的干扰,对噪声点具有一定的不敏感性;重建三维模型与原始点云吻合度高,基本解决了冠层内部枝干遮挡严重带来的三维建模困难的问题;依据模型提取树高、冠幅、胸径、体积等参数,增加了重建模型的应用范围。
[Abstract]:Laser point cloud data has a special advantage in estimating forest parameters and reconstructing 3D model of plant morphological structure with its detailed and high precision 3D information. In order to further improve the accuracy of 3D model. Integrated with various algorithms, on the basis of improving the existing PC2Tree software. Tree 3D reconstruction based on point cloud data. Firstly, branches and leaves are separated according to the similarity of the main direction of the local point cloud and the axial distribution density of the local point cloud. Secondly, level set and least square method are used to extract skeleton points of branch and stem, and characteristic points of canopy are extracted by downsampling method. Finally, the tree 3D model is reconstructed according to the topological structure of skeleton points and feature points. Taking camphor tree as an example, the segmentation accuracy of branches and leaves is analyzed. The result of automatic segmentation is similar to that of manual segmentation. The accuracy of the model was analyzed by taking the leafless egg-flower tree as an example. The relative error range of the main branch length of the model was concentrated at 0 ~ 8.0 and the relative error range of radius was 0 ~ 10. The branch reconstruction process avoids the interference of the noise points and is insensitive to the noise points. The reconstructed 3D model has a high consistency with the original point cloud, which basically solves the problem of 3D modeling which is caused by the serious occlusion of branches and branches in the canopy. The parameters of tree height, crown width, DBH and volume were extracted according to the model, and the application range of the reconstructed model was increased.
【作者单位】：福州大学空间数据挖掘与信息共享教育部重点实验室;福州大学地理空间信息技术国家地方联合工程研究中心;
【分类号】：TN958.58;;TP391.9
【正文快照】： 引言虚拟植物已被认为是一种潜在的有利于理解、表达和模拟植物形态结构生长发育的有效工具[1]。基于植物形态结构的精确模拟使虚拟植物实现了面向多层次、多方面的应用,如植物生长对资源的竞争、光辐射模拟研究[2]等。三维激光扫描技术在非破坏性的前提下能够自动、快速、准

【相似文献】