Chinese Journal of Lasers, Volume. 52, Issue 17, 1710004(2025)
Automatic Marker‐Free Registration for Terrestrial Laser Scanning Point Clouds in Forest Scenes Based on Tree Branching Structures
Figures & Tables(12)
Fig. 1. Flowchart of marker-free automatic registration method based on tree branching structures
Fig. 2. Process and principle of trunk detection. (a) Bottom point cloud; (b) clustering results of DBSCAN; (c) detection of normal vertical deviation and main direction vertical alignment; (d) final clustering results; (e) tree support stake; (f) target sphere
Fig. 3. Region growing segmentation principle based on graph constraints. (a) Point cloud to be segmented; (b) seed point selection; (c) adjacency graph construction process; (d) segmentation result
Fig. 4. Key point detection based on tree branching structure. (a) Distribution of key points; (b) clustering results of layers with detected key points
Fig. 6. Registration results. (a) Coarse registration results; (b) fine registration results; (c) close-up view
Fig. 7. Comparison results of different methods. (a) Super4PCS method; (b) proposed method
Fig. 8. Registration results of key points identified by different strategies. (a) SIFT; (b) Harris 3D; (c) ISS; (d) proposed method
Table 1. Parameter of the RIEGL VZ-400 scanner
View tableTable 1. Parameter of the RIEGL VZ-400 scanner
Parameter Specification Measurement range Minimum： 1.5 m
Maximum： 600 m （natural targets ≥90% reflectivity）
Maximum： 280 m （natural targets ≥20% reflectivity）
Laser beam divergence 0.3 mrad （corresponds to 30 mm increase in beam diameter per 100 m distance）
Vertical： 100° （+60°/-40°）
Scan angle range
Measurement accuracy
Measurement precision
Scan speed
Horizontal： 360°
5 mm
3 mm
Horizontal： 0°/sec to 60°/sec
Table 2. Registration accuracy results of the proposed method and the Super4PCS method
View tableTable 2. Registration accuracy results of the proposed method and the Super4PCS method
Scene Proposed method Super4PCS Coarse registration Fine
registration
RMSE MAE RMSE MAE RMSE MAE Average 3.1 2.8 2.4 2.1 3.0 2.8 01 3.0 2.7 2.5 2.2 2.9 2.6 02 2.9 2.6 2.3 2.0 2.9 2.6 03 3.0 2.7 2.2 1.9 2.9 2.6 04 3.3 3.1 2.6 2.3 3.4 3.1 05 3.2 2.9 2.6 2.3 3.1 2.9
Table 3. Comparison of overlap between the Super4PCS method and the proposed method
View tableTable 3. Comparison of overlap between the Super4PCS method and the proposed method
Scene Super4PCS Proposed method Coarse registration Fine registration （ θ=5 cm） Fine registration （ θ=2 cm） Average 0.610 0.980 1.000 0.998 01 0.577 0.975 1.000 0.999 02 0.602 0.980 1.000 0.998 03 0.582 0.982 1.000 0.998 04 0.511 0.982 1.000 0.999 05 0.742 0.979 1.000 0.998
Table 4. Registration overlap and runtime of key points identified by different strategies
View tableTable 4. Registration overlap and runtime of key points identified by different strategies
Scene SIFT+4PCS Harris 3D+4PCS ISS+4PCS Proposed method Overlap Runtime/s Overlap Runtime/s Overlap Runtime/s Overlap Runtime/s Average 0.722 117.4 0.685 19.2 0.670 76.64 0.980 9.74 01 0.551 79.2 0.578 6.5 0.734 56.9 0.982 5.7 02 0.865 156.4 0.507 5.5 0.513 7.6 0.982 5.8 03 0.946 128.7 0.708 5.6 0.837 60.4 0.979 5.5 04 0.590 37.1 0.878 8.5 0.707 12.8 0.975 6.4 05 0.658 185.7 0.754 69.7 0.559 260.5 0.980 25.3
Tools
Get Citation
Copy Citation Text
Yuhang Gao, Pei Wang, Changdong Shi, Jing Ren, Hanlong Li, Mingtai Zhang, Lingyun Zhang, Wenxin Chen, Jingdong Sun. Automatic Marker‐Free Registration for Terrestrial Laser Scanning Point Clouds in Forest Scenes Based on Tree Branching Structures[J]. Chinese Journal of Lasers, 2025, 52(17): 1710004
Paper Information
Category: remote sensing and sensor
Received: Mar. 3, 2025
Accepted: May. 6, 2025
Published Online: Sep. 4, 2025
The Author Email: Pei Wang (wangpei@bjfu.edu.cn)
CSTR:32183.14.CJL250579
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20