Automatic Registration Method of Vehicle‐Borne Laser Point Cloud Combining Ground Points and Rods

Mengbing Xu; Xianlin Liu; Xueting Zhong; Panke Zhang; Siyun Chen

doi:10.3788/CJL220689

Chinese Journal of Lasers, Volume. 50, Issue 2, 0210002(2023)

Automatic Registration Method of Vehicle‐Borne Laser Point Cloud Combining Ground Points and Rods

Mengbing Xu^1,2、*, Xianlin Liu³, Xueting Zhong¹, Panke Zhang^1,2, and Siyun Chen^1,2

Author Affiliations

¹College of Resource Environment and Tourism, Capital Normal University, Beijing 100048, China

²Beijing GEO-Vision Tech. Co., Ltd., Beijing 100070, China

³Chinese Academy of Surveying & Mapping, Beijing 100830, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(14)

Fig. 1. Overall technical flowchart

Download full size

Fig. 2. Schematic diagram of gradient calculation

Download full size

Fig. 3. Overlapping area of round-trip ground point clouds of the same road

Download full size

Fig. 4. Flowchart of proposed algorithm

Download full size

Fig. 5. Schematic diagram of 2D grid projection of tree and street light

Download full size

Fig. 6. Ground point cloud filtering effect. (a) Original road point cloud; (b) gradient filtering result;

Download full size

Fig. 7. Ground point cloud range of long road section. (a) Overlapping area of round-trip point clouds;（b）local detail magnification

Download full size

Fig. 8. Elevation registration effect. (a) Point clouds of original road section; (b) target point sets and point sets to be registered;

Download full size

Fig. 9. Plane registration effect. (a) Point clouds of original road section; (b) target point sets and point sets to be registered;

Download full size

Fig. 10. Comparison of registration results using different methods. (a) Point-to-point ICP method; (b) CPD method; (c) RANSAC-ICP method; (d) GICP method; (e) improved ICP method proposed in this paper

Download full size

Fig. 11. Accuracy verification of experimental results. (a) Distribution of improved ICP and manual registration parameters;

Download full size

Table 1. Performance parameters of Riegl laser scanner and IMU

View table

Table 1. Performance parameters of Riegl laser scanner and IMU

Device	Parameter		Value
Laser scanner Riegl-VUX-1HA	Laser pulse repetition rate /kHz		up to 1000
	Laser beam divergence /mrad		0.5
	Scanning mechanism Accuracy /mm Field of view（selectable）		Rotating mirror 5 360°，full circle
	Scan speed（selectable）/（r·s^-1）		10~250
IMU	Gyro drift	Roll：0.01°/h；pitch：0.01°/h；yaw：0.01°/h/cos l （l representing latitude）
IMU	Maintain accuracy	Roll：1‰；pitch：1‰；yaw：3‰~5‰

Table 2. Main parameter settings in algorithm

View table

Table 2. Main parameter settings in algorithm

Parameter category	Parameter settings
Data preprocessing	Mileage split value 25 m，overlapping area ≥140 m²，
Elevation registration	voxel grid size 0.05 m，upper dynamic range 0.6 m GNSS duration interruption time 4 min，rotation angle threshold 0.04°，lower dynamic range 0.1 m，iterative decrement unit 0.02 m
Plane registration	Fixed range threshold 0.5 m，surface curvature threshold ≤0.03，cumulative times threshold 40
Thread and ending conditions	Number of parallel threads 12（dynamic），error threshold ≤2×10^-2 m，
Thread and ending conditions	front-to-back difference threshold 10^-7 m，maximum iteration 270

Table 3. Registration accuracy and efficiency analysis of different algorithms
View table
Table 3. Registration accuracy and efficiency analysis of different algorithms
Algorithm Registration accuracy /m Registration time /s
Point-to-point ICP 13.073 217.253
CPD 2.934 >300.000
RANSAC-ICP 0.036 236.940
GICP 0.157 28.652
Improved ICP 0.014 9.776

Tools

Get Citation

Copy Citation Text

Mengbing Xu, Xianlin Liu, Xueting Zhong, Panke Zhang, Siyun Chen. Automatic Registration Method of Vehicle‐Borne Laser Point Cloud Combining Ground Points and Rods[J]. Chinese Journal of Lasers, 2023, 50(2): 0210002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites