Using point cloud data for tree organ classification and real leaf surface construction

文献类型： 外文期刊

第一作者： Yun, Ting

作者： Yun, Ting;Xue, Lianfeng;Yun, Ting;Chen, Bangqian;Li, Weizheng;Sun, Yuan;Yun, Ting;Xue, Lianfeng

作者机构：

关键词： Terrestrial Laser Scanning (TLS);Point Cloud Data (PCD);Tree organ classification;Leaf surface reconstruction

期刊名称：BULGARIAN CHEMICAL COMMUNICATIONS （ 影响因子：0.242； 五年影响因子：0.322 ）

ISSN： 0324-1130

年卷期： 2017 年 49 卷 1 期

页码：

收录情况： SCI

摘要： Terrestrial Laser Scanning (TLS) enables easy and fast Point Cloud Data (PCD) acquisition from objects, and it has been widely used in complex scene survey. However, trees have seriously irregular and complex morphology, and scanning process always be influenced by external environment variation and have occlusion effect, so quantifying the 3- D morphology structure and assessing parameters of forest stands by TLS is challenging. In order to solve these problems, we applied computer technique to improve Terrestrial Laser Scanning (TLS) performance in forestry measurement. Here, new PCD feature vectors, including shape, orientation, normal vector distribution and normal vectors of tangent plane, was proposed, and Supervised Locally Linear Embedding (SLLE) algorithm and Gaussian Mixture Model (GMM) were adopted for the feature dimensionality reduction and PCD classification as well. Hence, the algorithm efficiency was improved and various tree organs could be automatically identified. Moreover, a leaf modeling method using polynomial fitting method and Moving Least Squares (MLS) were presented to depict real foliage silhouette and eliminate ghost points, yielding accurate reconstruction of complex foliage surface. As detailed experimental comparison stated, the recognition rate remained higher than 87.51 % while our classification method was applied to different tree PCD, and accurate 3D morphological reconstruction of leaf models have similar leaf area versus manually LI-3000C measurement results. Thus, our method show promise in further exploration of utilizing TLS as an effective tool for forestry parameter retrieval.

分类号：

  • 相关文献

[1]A Novel Approach for Retrieving Tree Leaf Area from Ground-Based LiDAR. Yun, Ting,An, Feng,Yun, Ting,Xue, Lianfeng,Li, Weizheng,Sun, Yuan,Cao, Lin. 2016

作者其他论文 更多>>

微信小程序