Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data

Forest tree information is crucial for monitoring forest resources and developing forestry management strategies. Airborne laser scanning is an efficient and rapid method for acquiring 3-D point cloud data of forests, making up for the shortcomings of optical remote sensing. Especially for unmanned...

Full description

Saved in:
Bibliographic Details
Main Authors: Shihua Li, Shunda Zhao, Zhilin Tian, Hao Tang, Zhonghua Su
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10907937/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850151411726155776
author Shihua Li
Shunda Zhao
Zhilin Tian
Hao Tang
Zhonghua Su
author_facet Shihua Li
Shunda Zhao
Zhilin Tian
Hao Tang
Zhonghua Su
author_sort Shihua Li
collection DOAJ
description Forest tree information is crucial for monitoring forest resources and developing forestry management strategies. Airborne laser scanning is an efficient and rapid method for acquiring 3-D point cloud data of forests, making up for the shortcomings of optical remote sensing. Especially for unmanned aircraft vehicle laser scanning, the high-density point cloud provides the possibility of a detailed depiction of forest structures. Extracting individual trees from point cloud data is a prerequisite for fine forestry research. Currently, existing methods fail to fully utilize the height information, density information, and vertical structure details of tree crowns within point cloud data, resulting in complex algorithmic processes and reduced reliability. This study proposes a novel method that combines the canopy height model (CHM) and point cloud data for segmenting individual trees and employs height and density information for morphological detection of tree canopies. First, based on CHM-based segmentation, the point cloud density feature is introduced to identify the wrong segmented trees. Next, the height and density information are combined to guide the selection of vertical profiles. Finally, a 3-D morphological detection is conducted on the profiles to extract the potential trees. We validate the accuracy of the method on German forests. The results indicated that the average <italic>F</italic>1-scores of six study plots were 0.95, which improved by 5&#x0025; after the introduction of density information. The primary source of errors was the irregularity of tree morphology. Our method produced reasonable results across different parameter settings, demonstrating its insensitivity to parameter configurations.
format Article
id doaj-art-272292cf60a64d8ebff6b759606a54a3
institution OA Journals
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-272292cf60a64d8ebff6b759606a54a32025-08-20T02:26:15ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01188897890910.1109/JSTARS.2025.354665110907937Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR DataShihua Li0https://orcid.org/0000-0003-4807-5012Shunda Zhao1Zhilin Tian2https://orcid.org/0000-0002-2901-0574Hao Tang3https://orcid.org/0009-0002-4948-8900Zhonghua Su4https://orcid.org/0009-0003-4553-4874School of Resources and Environment, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Resources and Environment, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Resources and Environment, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Resources and Environment, University of Electronic Science and Technology of China, Chengdu, ChinaSchool of Computer and Software Engineering, Xihua University, Chengdu, ChinaForest tree information is crucial for monitoring forest resources and developing forestry management strategies. Airborne laser scanning is an efficient and rapid method for acquiring 3-D point cloud data of forests, making up for the shortcomings of optical remote sensing. Especially for unmanned aircraft vehicle laser scanning, the high-density point cloud provides the possibility of a detailed depiction of forest structures. Extracting individual trees from point cloud data is a prerequisite for fine forestry research. Currently, existing methods fail to fully utilize the height information, density information, and vertical structure details of tree crowns within point cloud data, resulting in complex algorithmic processes and reduced reliability. This study proposes a novel method that combines the canopy height model (CHM) and point cloud data for segmenting individual trees and employs height and density information for morphological detection of tree canopies. First, based on CHM-based segmentation, the point cloud density feature is introduced to identify the wrong segmented trees. Next, the height and density information are combined to guide the selection of vertical profiles. Finally, a 3-D morphological detection is conducted on the profiles to extract the potential trees. We validate the accuracy of the method on German forests. The results indicated that the average <italic>F</italic>1-scores of six study plots were 0.95, which improved by 5&#x0025; after the introduction of density information. The primary source of errors was the irregularity of tree morphology. Our method produced reasonable results across different parameter settings, demonstrating its insensitivity to parameter configurations.https://ieeexplore.ieee.org/document/10907937/Densityindividual tree segmentationmorphologyregion growingunmanned aerial vehicle (UAV) light detection and ranging (LiDAR)
spellingShingle Shihua Li
Shunda Zhao
Zhilin Tian
Hao Tang
Zhonghua Su
Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Density
individual tree segmentation
morphology
region growing
unmanned aerial vehicle (UAV) light detection and ranging (LiDAR)
title Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data
title_full Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data
title_fullStr Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data
title_full_unstemmed Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data
title_short Individual Tree Segmentation Based on Region-Growing and Density-Guided Canopy 3-D Morphology Detection Using UAV LiDAR Data
title_sort individual tree segmentation based on region growing and density guided canopy 3 d morphology detection using uav lidar data
topic Density
individual tree segmentation
morphology
region growing
unmanned aerial vehicle (UAV) light detection and ranging (LiDAR)
url https://ieeexplore.ieee.org/document/10907937/
work_keys_str_mv AT shihuali individualtreesegmentationbasedonregiongrowinganddensityguidedcanopy3dmorphologydetectionusinguavlidardata
AT shundazhao individualtreesegmentationbasedonregiongrowinganddensityguidedcanopy3dmorphologydetectionusinguavlidardata
AT zhilintian individualtreesegmentationbasedonregiongrowinganddensityguidedcanopy3dmorphologydetectionusinguavlidardata
AT haotang individualtreesegmentationbasedonregiongrowinganddensityguidedcanopy3dmorphologydetectionusinguavlidardata
AT zhonghuasu individualtreesegmentationbasedonregiongrowinganddensityguidedcanopy3dmorphologydetectionusinguavlidardata