Paper Info
Title
Extracting Wood Point Cloud of Individual Trees Based on Geometric Features
Abstract
The wood structure is an important parameter that represents the geometrical and topological characteristics of trees. Accurate extraction of the wood component of a tree is of great importance in visualizing trees. Light detection and ranging (LiDAR) has been applied to obtain the 3-D structural properties of vegetation. However, it is difficult to separate the wood and leaf components from point clouds data in situations where wood and leaves are mixed and overlapping. This letter proposes an effective method for extracting wood point cloud of individual trees based on different geometric features of leaves and wood by combining classification and segmentation methods. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Magnolia grandiflora</italic> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Cinnamomum camphor</italic> trees were scanned using a high-resolution terrestrial LiDAR. The complexity of the structure of the canopy was reduced using a slicing method. The wood and/or leaf components were classified using the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> -means and random sampling consistency (RANSAC) algorithm. The cylindrical segmentation method based on the RANSAC algorithm was used for the precise extraction of the wood component in wood and leaf mixed point clouds. The trunk under the canopy was extracted completely. The average recall and precision in extracting canopy wood point clouds of the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Magnolia grandiflora</italic> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Cinnamomum camphor</italic> achieved 94.60%, 92.02% and 93.62%, 91.46%, respectively. The results indicate that the proposed method has the potential for accurately extracting the wood point cloud from terrestrial LiDAR data.
Year
DOI
Venue
2019
10.1109/lgrs.2019.2896613
IEEE Geoscience and Remote Sensing Letters
Keywords
Field
DocType
Three-dimensional displays,Vegetation,Feature extraction,Laser radar,Clustering algorithms,Shape,Distance measurement
RANSAC,Remote sensing,Precision and recall,Feature extraction,Lidar,Sampling (statistics),Point cloud,Cluster analysis,Mathematics,Canopy
Journal
Volume
Issue
ISSN
16
8
1545-598X
Citations 
PageRank 
References 
0
0.34
0
Authors
4
Name
Order
Citations
PageRank
Zhonghua Su101.69
Shihua Li202.37
Hanhu Liu300.34
Yuhan Liu4419.47