A Method Coupling NDT and VGICP for Registering UAV-LiDAR and LiDAR-SLAM Point Clouds in Plantation Forest Plots DOI Open Access
Fan Wang, Jiawei Wang, Yun Wu

et al.

Forests, Journal Year: 2024, Volume and Issue: 15(12), P. 2186 - 2186

Published: Dec. 12, 2024

The combination of UAV-LiDAR and LiDAR-SLAM (Simultaneous Localization Mapping) technology can overcome the scanning limitations different platforms obtain comprehensive 3D structural information forest stands. To address challenges traditional registration algorithms, such as high initial value requirements susceptibility to local optima, in this paper, we propose a high-precision, robust, NDT-VGICP method that integrates voxel features register point clouds at stand scale. First, are voxelized, their normal vectors distribution models computed, then transformation matrix is quickly estimated based on pair characteristics achieve preliminary alignment. Second, high-dimensional feature weighting introduced, iterative closest (ICP) algorithm used optimize distance between matching pairs, adjusting reduce errors iteratively. Finally, converges when conditions met, yielding an optimal achieving precise cloud registration. results show performs well Chinese fir stands age groups (average RMSE—horizontal: 4.27 cm; vertical: 3.86 cm) achieves accuracy single-tree crown vertex detection tree height estimation F-score: 0.90; R2 for estimation: 0.88). This study demonstrates effectively fuse collaboratively apply multi-platform LiDAR data, providing methodological reference accurately quantifying individual parameters efficiently monitoring structures.

Language: Английский

Quantifying Aboveground Carbon Stock at Species Level Using TLS LiDAR and UAV Photogrammetry for Urban Trees DOI
Rezaul Roni,

Shah Nurul Hasnat Sadi,

Abdullah Al‐Mamun

et al.

IntechOpen eBooks, Journal Year: 2025, Volume and Issue: unknown

Published: April 8, 2025

Urbanization is increasing the depletion of natural carbon sinks and intensification urban heat islands, creating vegetation critical for sequestration climate regulation. In this study, a fusion approach was applied that combined Terrestrial Laser Scanning (TLS) Light Detection Ranging (LiDAR) with high-resolution Unmanned Aerial Vehicle (UAV) imagery to estimate aboveground stock individual trees along Manik Mia Avenue, Dhaka, Bangladesh. UAV imageries dense point cloud data from TLS LiDAR were collected georeferenced using Real-Time Kinematic (RTK) GPS. After screening contouring models filter vegetation, it possible segment trees, measure tree height diameter at breast (DBH), calculate through species-specific allometric equations. The results indicate strong correlation between field-measured cloud-derived (r2 = 0.94, RMSE 0.49) DBH 0.88). While estimation achieved high 0.80), species aerial roots posed challenges in measurement, resulting low 0.26) when assessed separately. Limitations include insufficient scanning angles TLS, variability density, constraints non-invasive techniques. Future research could integrate multispectral geometric shape fitting address enhance precision, contributing management Sustainable Development Goals (SDGs) 11 15, which are related sustainable cities forest management.

Language: Английский

Citations

0
SVC-DAD: An novel local shape descriptor for cross-source point cloud registration DOI
Jian Li,

Huibin Li,

Guohe Han

et al.

Measurement, Journal Year: 2025, Volume and Issue: unknown, P. 117981 - 117981

Published: May 1, 2025

Language: Английский

Citations

0
Semantic Segmentation-Driven Integration of Point Clouds from Mobile Scanning Platforms in Urban Environments DOI Creative Commons
Joanna Koszyk, Aleksandra Jasińska, Karolina Pargieła

et al.

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(18), P. 3434 - 3434

Published: Sept. 16, 2024

Precise and complete 3D representations of architectural structures or industrial sites are essential for various applications, including structural monitoring cadastre. However, acquiring these datasets can be time-consuming, particularly large objects. Mobile scanning systems offer a solution such cases. In the case complex scenes, multiple required to obtain point clouds that merged into comprehensive representation object. Merging individual obtained from different sensors at times difficult due discrepancies caused by moving objects changes in scene over time, as seasonal variations vegetation. this study, we present integration two mobile platforms within built-up area. We utilized combination quadruped robot an unmanned aerial vehicle (UAV). The PointNet++ network was employed conduct semantic segmentation task, enabling detection non-ground experimental tests used Toronto dataset DALES training. Based on performance, model trained chosen further research. proposed algorithm involved both clouds, dividing them square subregions, performing subregion selection checking emptiness when subregions contained points. Parameters local density, centroids, coverage, Euclidean distance were evaluated. Point cloud merging augmentation enhanced with clustering resulted exclusion points associated movable clouds. comparative analysis method simple performed based file size, number points, mean roughness, noise estimation. provided adequate results improvement quality indicators.

Language: Английский

Citations

0
A Method Coupling NDT and VGICP for Registering UAV-LiDAR and LiDAR-SLAM Point Clouds in Plantation Forest Plots DOI Open Access
Fan Wang, Jiawei Wang, Yun Wu

et al.

Forests, Journal Year: 2024, Volume and Issue: 15(12), P. 2186 - 2186

Published: Dec. 12, 2024

The combination of UAV-LiDAR and LiDAR-SLAM (Simultaneous Localization Mapping) technology can overcome the scanning limitations different platforms obtain comprehensive 3D structural information forest stands. To address challenges traditional registration algorithms, such as high initial value requirements susceptibility to local optima, in this paper, we propose a high-precision, robust, NDT-VGICP method that integrates voxel features register point clouds at stand scale. First, are voxelized, their normal vectors distribution models computed, then transformation matrix is quickly estimated based on pair characteristics achieve preliminary alignment. Second, high-dimensional feature weighting introduced, iterative closest (ICP) algorithm used optimize distance between matching pairs, adjusting reduce errors iteratively. Finally, converges when conditions met, yielding an optimal achieving precise cloud registration. results show performs well Chinese fir stands age groups (average RMSE—horizontal: 4.27 cm; vertical: 3.86 cm) achieves accuracy single-tree crown vertex detection tree height estimation F-score: 0.90; R2 for estimation: 0.88). This study demonstrates effectively fuse collaboratively apply multi-platform LiDAR data, providing methodological reference accurately quantifying individual parameters efficiently monitoring structures.

Language: Английский

Citations

0