动态扫描场景下GM-APD激光雷达点云高精度配准方法研究

钟国舜; 刘秋佐; 李萌; 彭涛; 孙剑峰; 刘建伟

doi:10.37188/CO.2025-0073

动态扫描场景下GM-APD激光雷达点云高精度配准方法研究

doi: 10.37188/CO.2025-0073

cstr: 32171.14.CO.2025-0073

1.
中国电子科技集团公司第十一研究所, 北京, 100015
2.
哈尔滨工业大学激光空间信息全国重点实验室, 黑龙江哈尔滨 150001

基金项目: 国家自然科学基金（No. 62105240，No. 62075159）；国家重点研发计划（No. 2019YFB2203002）

详细信息

作者简介:
钟国舜（1983—），男，湖南常德人，研究生学历，主要从事光电子技术方向。E-mail：james830411@163.com

中图分类号: TN958.98;
计量
- 文章访问数: 23
- HTML全文浏览量: 9
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2025-04-25
- 录用日期: 2025-07-03
- 网络出版日期: 2025-07-22

Research on high-precision registration methods for GM-APD LiDAR point clouds in dynamic scanning scenarios

1.
The 11th Research Institute of China Electronics Technology Group Corporation, Beijing, 100015, China
2.
National Key Laboratory of Laser Spatial Information, Harbin Institute of Technology, Harbin, Heilongjiang 150001, China

Funds: Supported by

More Information

Corresponding author: james830411@163.com

摘要

摘要:
本研究针对盖革雪崩光电二极管（Geiger-mode avalanche photodiode，GM-APD）激光雷达在动态扫描场景下相邻帧点云重叠率低、易强制配准非匹配点对的问题，提出了一种基于双向匹配机制和多分辨率邻域扩展的改进ICP算法，以提高点云配准精度和鲁棒性。首先，通过基于K-D tree的双向匹配机制提取相邻帧点云的重叠区域，利用重叠区域信息建立初始配准模型，解决了低重叠率场景下配准精度下降的问题。其次，采用多分辨率邻域扩展技术，结合局部曲率相似性加权求解变换矩阵，避免了动态配准中强制对齐非匹配点对的现象。最后，通过级联补偿机制实现全局点云的精确配准。实验结果表明，在2 km和400 m扫描成像中，平均距离误差分别为0.21 m和0.10 m。该方案有效解决了动态扫描场景下的点云配准难题，为三维重构提供了高精度数据支持，具有重要应用价值。
- GM-APD /
- 点云配准 /
- 位姿校正 /
- 双向匹配 /
- 多分辨率邻域扩展
Abstract:
This paper addresses the challenges of low overlap and mismatched point pairs in Geiger-mode avalanche photodiode (GM-APD) LiDAR point clouds under dynamic scanning conditions. To improve registration accuracy and robustness, an enhanced Iterative Closest Point (ICP) algorithm is proposed, integrating a bidirectional matching scheme and multi-resolution neighborhood expansion. First, a K-D tree-based bidirectional search identifies overlapping regions between consecutive frames, enabling accurate initial alignment. Then, a high-resolution neighborhood expansion approach, weighted by local curvature similarity, is applied to refine the transformation matrix and suppress mismatched correspondences. Finally, a cascaded compensation mechanism ensures global consistency across frames. Experiments demonstrate that our method achieves average distance errors of 0.21 m (2 km scene) and 0.10 m (400 m scene), effectively improving registration precision in dynamic scenarios and offering valuable support for 3D reconstruction.
- GM-APD /
- point cloud registration /
- pose correction /
- bidirectional matching /
- multi-resolution neighborhood expansion

HTML全文

图 1 重叠区域提取实现流程

Figure 1. Workflow of overlapping region extraction

下载: 全尺寸图片幻灯片

图 2 动态扫描非重合性偏移

Figure 2. Offset between frames in dynamic scanning

下载: 全尺寸图片幻灯片

图 3 高分辨率邻域扩展

Figure 3. High-resolution neighborhood expansion

下载: 全尺寸图片幻灯片

图 4 仿真城市场景

Figure 4. Urban scene simulation

下载: 全尺寸图片幻灯片

图 5 仿真场景理想局部配准结果

Figure 5. Ideal local registration result in simulation scene

下载: 全尺寸图片幻灯片

图 6 仿真场景局部配准结果对比。(a)传统ICP；(b)改进ICP；(c)重叠点云提取结合传统ICP；(d)重叠点云提取结合改进ICP

Figure 6. Comparison of local registration results in simulation scenarios.(a) Conventional ICP; (b) Improved ICP; (c) Conventional ICP with Overlapping Point Cloud Extraction; (d) Improved ICP with Overlapping Point Cloud Extraction

下载: 全尺寸图片幻灯片

图 7 模拟地图与全局点云主观对比

Figure 7. Subjective comparison between simulated map and global point cloud

下载: 全尺寸图片幻灯片

图 8 模拟地图与全局精配准点云几何相似度

Figure 8. Geometric similarity between simulated map and globally fine-registered point cloud

下载: 全尺寸图片幻灯片

图 9 大视场扫描成像实拍场景

Figure 9. Real-world scene captured with large field-of-view scanning

下载: 全尺寸图片幻灯片

图 10 真实场景局部配准结果对比。(a)传统ICP；(b)改进ICP；(c)重叠点云提取结合传统ICP；(d)重叠点云提取结合改进ICP

Figure 10. Comparison of local registration results in real-world scenes.(a) Conventional ICP; (b) Improved ICP; (c) Conventional ICP with overlapping point cloud extraction; (d) Improved ICP with overlapping point cloud extraction

下载: 全尺寸图片幻灯片

图 11 真实场景点云与卫星地图对比

Figure 11. Point cloud-to-satellite map comparison in real-world scenes

下载: 全尺寸图片幻灯片

图 12 真实场景全局精配准点云几何相似度

Figure 12. Geometric similarity evaluation of globally optimized point clouds in real-world scenes

下载: 全尺寸图片幻灯片

表 1 仿真场景局部精配准客观评价

Table 1. Local fine registration objective evaluation of simulation scene

	传统ICP	改进ICP	提取重叠区域+ 传统ICP	提取重叠区域+ 改进ICP
平均值	9.03 m	3.67 m	0.26 m	0.21 m
最大值	9.36 m	5.97 m	4.52 m	0.80 m
最小值	8.20 m	2.69 m	0.04 m	0.04 m
标准差	0.34 m	0.58 m	0.77 m	0.16 m
耗时	0.094 s	0.172 s	0.110 s	0.186 s

下载: 导出CSV

表 2 仿真场景全局精配准绝对空间一致性分析

Table 2. Absolute spatial consistency analysis for global fine registration in simulated scenarios

	平均值	最大值	最小值	标准差
距离误差	0.47 m	0.80 m	0.22 m	0.50 m

下载: 导出CSV

表 3 真实场景局部配准精度

Table 3. Local registration accuracy in real-world scenes

	传统ICP	改进ICP	提取重叠区域+ 传统ICP	提取重叠区域+ 改进ICP
平均值	7.23 m	1.96 m	0.28 m	0.10 m
最大值	7.66 m	2.94 m	0.78 m	0.13 m
最小值	6.79 m	1.29 m	0.09 m	0.05 m
标准差	0.33 m	0.69 m	0.29 m	0.02 m
耗时	0.078 s	0.125 s	0.094 s	0.140 s

下载: 导出CSV

参考文献(20)

[1]	LESLIE M. On-chip LiDAR technology advances for cars, cell phones[J]. Engineering, 2022, 18: 3-5. doi: 10.1016/j.eng.2022.09.003
[2]	GUO M, SUN M X, ZHOU T F, et al. Novel trajectory optimization algorithm of vehicle-borne LiDAR mobile measurement system[J]. Sensors and Materials, 2020, 32(11): 3935-3953. doi: 10.18494/SAM.2020.3052
[3]	桑洋, 纪新春, 魏东岩, 等. 基于测速测距激光雷达的飞行器地形匹配导航方法[J]. 中国惯性技术学报,2024,32(1):8-15. SANG Y, JI X CH, WEI D Y, et al. Terrain matching navigation method for air vehicle based on FMCW LiDAR[J]. Journal of Chinese Inertial Technology, 2024, 32(1): 8-15.
[4]	孙国祥, 黄银锋, 汪小旵, 等. 基于LIO-SAM建图和激光视觉融合定位的温室自主行走系统[J]. 农业工程学报,2024,40(3):227-239. doi: 10.11975/j.issn.1002-6819.202311146 SUN G X, HUANG Y F, WANG X CH, et al. Autonomous navigation system in a greenhouse using LIO-SAM mapping and laser vision fusion localization[J]. Transactions of the Chinese Society of Agricultural Engineering, 2024, 40(3): 227-239. doi: 10.11975/j.issn.1002-6819.202311146
[5]	郭明, 齐慧慧, 郭可才, 等. 融合北斗/GNSS定位和5G通讯的地基激光雷达测量系统[J]. 光学精密工程,2023,31(4):450-458. doi: 10.37188/OPE.20233104.0450 GUO M, QI H H, GUO K C, et al. Ground-based LiDAR measurement system integrating BeiDou/GNSS positioning and 5G communication[J]. Optics and Precision Engineering, 2023, 31(4): 450-458. doi: 10.37188/OPE.20233104.0450
[6]	GUO M, ZHOU Y Q, ZHAO J H, et al. Urban Geospatial Information Acquisition Mobile Mapping System based on close-range photogrammetry and IGS site calibration[J]. Geo-spatial Information Science, 2021, 24(4): 558-579. doi: 10.1080/10095020.2021.1924084
[7]	王召泽, 韩若愚, 赵智伟, 等. 手持三维激光扫描仪在城市更新立面测量中的应用[J]. 测绘通报,2025(1):133-137. WANG ZH Z, HAN R Y, ZHAO ZH W, et al. The application of handheld 3D laser scanner in urban renewal facade measurement[J]. Bulletin of Surveying and Mapping, 2025(1): 133-137.
[8]	潘斌, 蔡志刚, 刘微微, 等. 机载激光雷达点云的城市建筑物直线特征提取方法[J]. 测绘工程,2022,31(5):16-23. PAN B, CAI ZH G, LIU W W, et al. Straight line feature extraction method of city airborne LiDAR point clouds[J]. Engineering of Surveying and Mapping, 2022, 31(5): 16-23.
[9]	MARINO R M, DAVIS W R. Jigsaw: a foliage-penetrating 3D imaging laser radar system[J]. Lincoln Laboratory Journal, 2005, 15(1): 23-36.
[10]	MARINO R M, STEPHENS T, HATCH R E, et al. A compact 3D imaging laser radar system using Geiger-mode APD arrays: system and measurements[J]. Proceedings of SPIE, 2003, 5086: 1-15. doi: 10.1117/12.501581
[11]	MARINO R M, DAVIS W R, RICH G C, et al. High-resolution 3D imaging laser radar flight test experiments[J]. Proceedings of SPIE, 2005, 9591: 138-151.
[12]	ALBOTA M A, AULL B F, FOUCHE D G, et al. Three-dimensional imaging laser radars with Geiger-mode avalanche photodiode arrays[J]. Lincoln Laboratory Journal, 2002, 13(2): 351-370.
[13]	ALBOTA M, GURJAR R, MANGOGNIA A, et al. The airborne optical systems testbed (AOSTB)[R]. Lexington: MIT Lincoln Laboratory Lexington United States, 2017. (查阅网上资料, 未找到本条文献信息, 请确认).
[14]	KNOWLTON R. Airborne ladar imaging research testbed[R]. Lexington: MIT Lincoln Laboratory, 2011: 1.
[15]	PAWLIKOWSKA A M, HALIMI A, LAMB R A, et al. Single-photon three-dimensional imaging at up to 10 kilometers range[J]. Optics Express, 2017, 25(10): 11919-11931. doi: 10.1364/OE.25.011919
[16]	ZHANG Y F, HE Y, YANG F, et al. Three-dimensional imaging lidar system based on high speed pseudorandom modulation and photon counting[J]. Chinese Optics Letters, 2016, 14(11): 111101-111105. doi: 10.3788/COL201614.111101
[17]	ZHANG Z J, XU Y N, WU L, et al. Photon counting range-intensity image strategy in low-light level environments[J]. Optics Letters, 2014, 39(8): 2467-2470. doi: 10.1364/OL.39.002467
[18]	陈勇强, 贺岩, 罗远, 等. 基于盖革APD阵列脉冲式三维成像激光雷达系统[J]. 中国激光,2023,50(2):0210001. doi: 10.3788/CJL220683 CHEN Y Q, HE Y, LUO Y, et al. Pulsed three-dimensional imaging lidar system based on Geiger-mode APD array[J]. Chinese Journal of Lasers, 2023, 50(2): 0210001. doi: 10.3788/CJL220683
[19]	韩超, 刘志勇, 刘德伟, 等. 单光子激光雷达中时间走时时差校准算法研究[J]. 中国光学,2023,16(2):271-280. HAN CH, LIU ZH Y, LIU D W, et al. Time walk correction algorithm for single-photon LiDAR systems[J]. Chinese Optics, 2023, 16(2): 271-280. (查阅网上资料, 未找到本条文献信息, 请确认).
[20]	高翔, 马玉波, 王晨, 等. 高精度激光雷达点云配准算法综述[J]. 中国光学,2022,15(3):495-509. GAO X, MA Y B, WANG CH, et al. Review of high-precision point cloud registration algorithms for LiDAR[J]. Chinese Optics, 2022, 15(3): 495-509. (查阅网上资料, 未找到本条文献信息, 请确认).