基于区域生长的非结构巷道点云去噪方法

连忠文; 任助理; 郝英豪; 杨帆; 白刚; 方程; 袁瑞甫

doi:10.13272/j.issn.1671-251x.2024010037

基于区域生长的非结构巷道点云去噪方法

连忠文^{1, 2,},
任助理^{1, 3, 4, ,},
郝英豪²,
杨帆²,
白刚²,
方程²,
袁瑞甫^{1, 4}

1.
河南理工大学能源科学与工程学院，河南焦作　454003
2.
中天合创能源有限责任公司煤炭分公司葫芦素煤矿，内蒙古鄂尔多斯　017000
3.
河南理工大学河南省矿产资源绿色高效开采与综合利用重点实验室，河南焦作　454003
4.
煤炭安全生产与清洁高效利用省部共建协同创新中心，河南焦作　454003

基金项目: 国家自然科学基金青年基金项目（52204168）；河南省矿产资源绿色高效开采与综合利用实验室重点项目（KCF2201）。

详细信息

作者简介:
连忠文（1981—），男，山西朔州人，正高级工程师，主要研究方向为巷道变形智能监测与控制，E-mail：1515918491@qq.com

通讯作者:
任助理（1990—），男，河南周口人，副教授，主要研究方向为数字矿山与智能采矿，E-mail：zhuliren@hpu.edu.cn。

中图分类号: TD67
计量
- 文章访问数: 122
- HTML全文浏览量: 37
- PDF下载量: 19
出版历程
- 收稿日期: 2024-01-11
- 修回日期: 2024-03-14
- 网络出版日期: 2024-04-10
- 刊出日期: 2024-03-19

A point cloud denoising method for unstructured roadways based on regional growth

LIAN Zhongwen^{1, 2,},
REN Zhuli^{1, 3, 4, ,},
HAO Yinghao²,
YANG Fan²,
BAI Gang²,
FANG Cheng²,
YUAN Ruifu^{1, 4}

1.
School of Energy Science and Engineering, Henan Polytechnic University, Jiaozuo 454003, China
2.
Hulusu Coal Mine, Coal Branch of ZTHC Energy Limited Liability Company, Ordos 017000, China
3.
Henan Key Laboratory for Green and Efficient Mining & Comprehensive Utilization of Mineral Resources, Henan Polytechnic University, Jiaozuo 454003, China
4.
Collaborative Innovation Center of Coal Work Safety and Clean High Efficiency Utilization, Jiaozuo 454003, China

摘要

摘要: 目前针对地下巷道点云去噪研究未完全满足巷道点云的特殊去噪需求，尤其是在狭长、密闭且复杂的地下巷道环境中，未能充分应对管壁附属物、粉尘和人为噪声等因素造成的挑战。通过分析井下非结构场景和传感器误差，考虑行人、移动设备和管网带来的噪声，提出一种基于区域生长的非结构巷道点云去噪方法。利用三维激光扫描技术获得井下巷道场景的3D点云信息，并分析其中由于井下非结构场景和传感器误差造成的异常点，以及行人、移动设备和风/水管网形成的噪声特点；利用k维树（kd-tree）构建点云的拓扑关系，选取适当的种子节点和生长准则，设定合适的曲率和角度阈值，通过区域生长算法实现巷道点云的有效分割，去除未加入分割区域的离群点云；根据噪声特点，基于巷道点云区域分割结果进一步去噪优化。试验结果表明：对于巷道中存在行人、设备等特征的情况，建议将区域生长算法的角度阈值设定为10°左右，曲率阈值设定为3左右；在实际应用中，应平衡数据量的减少与去噪效果，以确保数据处理的有效性，同时提高数据质量；采用基于区域生长的非结构巷道点云去噪方法进行去噪时，点云数量减少幅度介于SOR滤波器和低通滤波器之间，能有效移除行人、设备等噪声。
- 非结构巷道 /
- 三维激光扫描 /
- 点云去噪 /
- 点云区域分割 /
- 区域生长 /
- kd-tree
Abstract: Currently, research on point cloud denoising in underground roadways has not fully met the special denoising needs of roadway point clouds. Especially in narrow, enclosed, and complex underground roadway environments, the research has not fully addressed the challenges caused by pipe wall attachments, dust, and human noise. By analyzing the unstructured scenes and sensor errors underground, considering the noise caused by personnel, mobile devices, and pipeline networks, a point cloud denoising method for unstructured roadways based on region growth is proposed. The method uses 3D laser scanning technology to obtain 3D point cloud information of underground roadway scenes, and analyzes the abnormal points caused by unstructured underground scenes and sensor errors, as well as the noise features formed by personnel, mobile devices, and air and water pipelines. The method uses k-dimensional trees (kd-tree) to construct the topological relationship of point clouds, selects appropriate seed nodes and growth criteria, and sets appropriate curvature and angle thresholds. The method implements effective segmentation of roadway point clouds through region growth algorithms, and removes outlier point clouds that have not been added to the segmentation area. Based on the features of noise, further denoising optimization is carried out based on the segmentation results of the roadway point cloud region. The experimental results indicate that for situations where there are features such as personnel and equipment in the roadway, it is recommended to set the angle threshold of the region growth algorithm to around 10° and the curvature threshold to around 3. In practical applications, it is necessary to balance the reduction of data volume with the denoising effect to ensure the effectiveness of data processing and improve data quality. When using a region growth based unstructured roadway point cloud denoising method for denoising, the reduction in point cloud quantity is between SOR filter and low-pass filter, which can effectively remove noise such as personnel and equipment.
- unstructured roadway /
- 3D laser scanning /
- point cloud denoising /
- point cloud region segmentation /
- regional growth /
- kd-tree

HTML全文

图 1 巷道点云组成

Figure 1. Roadway point cloud composition

下载: 全尺寸图片幻灯片

图 2 巷道点云去噪流程

Figure 2. Denoising process of roadway point cloud

下载: 全尺寸图片幻灯片

图 3 kd-tree原理

Figure 3. Principle of kd-tree

下载: 全尺寸图片幻灯片

图 4 区域生长算法流程

Figure 4. Regional growth algorithm flow

下载: 全尺寸图片幻灯片

图 5 不同曲率阈值下巷道点云区域分割效果

Figure 5. Effect of roadway point cloud region segmentation under different curvature thresholds

下载: 全尺寸图片幻灯片

图 6 三维激光扫描系统

Figure 6. Three-dimensional laser scanning system

下载: 全尺寸图片幻灯片

图 7 井下巷道场景三维激光扫描图

Figure 7. 3D laser scanning image of underground roadway scene

下载: 全尺寸图片幻灯片

图 8 主运大巷21406工作面段三维点云

Figure 8. 3D point cloud in 21406 working face section of the main haulage tunnel

下载: 全尺寸图片幻灯片

图 9 21407工作面回风巷三维点云

Figure 9. 3D point cloud in return airway of 21407 working face

下载: 全尺寸图片幻灯片

图 10 含噪点云样本

Figure 10. Point cloud sample with noise

下载: 全尺寸图片幻灯片

图 11 巷道点云区域分割结果

Figure 11. Results of roadway point cloud region segmentation

下载: 全尺寸图片幻灯片

图 12 巷道点云去噪优化结果

Figure 12. Optimization results of roadway point cloud denoising

下载: 全尺寸图片幻灯片

图 13 不同阈值下巷道点云区域分割效果

Figure 13. Effect of roadway point cloud region segmentation under different thresholds

下载: 全尺寸图片幻灯片

图 14 不同方法的巷道点云去噪效果

Figure 14. Denoising effect of different methods for roadway point cloud

下载: 全尺寸图片幻灯片

表 1 去噪过程点云数量变化

Table 1 The number of point clouds changes during the denoising process

样本序号	点云数量/个			点云总体减少比例/%
样本序号	原始点云	分割点云	去噪后点云	点云总体减少比例/%
1	148 690	123 424	123 424	16.99
2	123 338	102 271	101 014	18.10
3	145 379	129 775	98 114	32.51
4	210 808	184 270	148 858	29.30

下载: 导出CSV

表 2 不同方法去噪后点云数量对比

Table 2 Comparison of the number of point clouds after denoising by different methods

样本序号	原始点云数量/个	去噪后点云数量/个
样本序号	原始点云数量/个	SOR 滤波器	低通滤波器	本文方法
1	148 690	142 549	99 004	123 424
2	123 338	118 979	74 039	101 014
3	145 379	136 073	90 882	98 114
4	210 808	195 975	143 882	148 858

下载: 导出CSV

参考文献(27)

[1]	王国法,杜毅博. 智慧煤矿与智能化开采技术的发展方向[J]. 煤炭科学技术,2019,47(1):1-10. WANG Guofa,DU Yibo. Development direction of intelligent coal mine and intelligent mining technology[J]. Coal Science and Technology,2019,47(1):1-10.
[2]	王国法,刘峰,孟祥军,等. 煤矿智能化(初级阶段)研究与实践[J]. 煤炭科学技术,2019,47(8):1-36. WANG Guofa,LIU Feng,MENG Xiangjun,et al. Research and practice on intelligent coal mine construction(primary stage)[J]. Coal Science and Technology,2019,47(8):1-36.
[3]	余润泽,单伽锃,李培振,等. 基于三维点云的工程结构平面分割及表面检测[J]. 中南大学学报(自然科学版),2023,54(6):2233-2243. DOI: 10.11817/j.issn.1672-7207.2023.06.014 YU Runze,SHAN Jiazeng,LI Peizhen,et al. Point cloud-based plane segmentation and surface inspection of engineering structures[J]. Journal of Central South University(Science and Technology),2023,54(6):2233-2243. DOI: 10.11817/j.issn.1672-7207.2023.06.014
[4]	KUMAR S S,PRATAP B B,SIMIT R. A review of laser scanning for geological and geotechnical applications in underground mining[J]. International Journal of Mining Science and Technology,2023,33(2):133-154. DOI: 10.1016/j.ijmst.2022.09.022
[5]	戴文祥,陈雷,闫鹏飞,等. 基于三维激光扫描的煤矿巷道形变监测方法[J]. 工矿自动化,2023,49(10):61-67,95. DAI Wenxiang,CHEN Lei,YAN Pengfei,et al. A deformation monitoring method for coal mine roadway based on 3D laser scanning[J]. Journal of Mine Automation,2023,49(10):61-67,95.
[6]	RUSU R B,COUSINS S. 3D is here:point cloud library (PCL)[C]. IEEE International Conference on Robotics and Automation,Shanghai,2011:6324-6327.
[7]	KOLLURI R. Provably good moving least squares[C]. The 16th Annual ACM-SIAM Symposium on Discrete Algorithms,Vancouver,2008:1008-1017.
[8]	LIPMAN Y,COHEN-OR D,LEVIN D,et al. Parameterization-free projection for geometry reconstruction[J]. ACM Transactions on Graphics(TOG),2007,26(3). DOI: 10.1145/1276377.1276405.
[9]	HUANG Hui,WU Shihao,GONG Minglun,et al. Edge-aware point set resampling[J]. ACM Transactions on Graphics,2013,32(1):1-12.
[10]	HUANG Hui,LI Dan,ZHANG Hao,et al. Consolidation of unorganized point clouds for surface reconstruction[J]. ACM Transactions on Graphics,2009,28(5):1-7.
[11]	MATTEI E,CASTRODAD A. Point cloud denoising via moving RPCA[J]. Computer Graphics Forum:Journal of the European Association for Computer Graphics,2017,36(8):123-137.
[12]	SUN Yujing,SCHAEFER S,WANG Wenping. Denoising point sets via L0 minimization[J]. Computer Aided Geometric Design,2015,35:2-15.
[13]	ROVERI R,ÖZTIRELI A C,PANDELE I,et al. Pointpronets:consolidation of point clouds with convolutional neural networks[J]. Computer Graphics Forum,2018,37(2):87-99. DOI: 10.1111/cgf.13344
[14]	ZHANG Dongbo,LU Xuequan,QIN Hong,et al. Pointfilter:point cloud filtering via encoder-decoder modeling[J]. IEEE Transactions on Visualization and Computer Graphics,2020,27(3):2015-2027.
[15]	RAKOTOSAONA M J,LA BARBERA V,GUERRERO P,et al. Pointcleannet:learning to denoise and remove outliers from dense point clouds[J]. Computer Graphics Forum,2020,39(1):185-203. DOI: 10.1111/cgf.13753
[16]	CASAJUS P H,RITSCHEL T,ROPINSKI T. Total denoising:unsupervised learning of 3D point cloud cleaning[C]. IEEE/CVF International Conference on Computer Vision,Seoul,2019:52-60.
[17]	JONES T R. Feature preserving smoothing of 3D surface scans[D]. Cambridge:Massachusetts Institute of Technology,2003.
[18]	FLEISHMAN S,DRORI I,COHEN-OR D. Bilateral mesh denoising[J]. ACM Transactions on Graphics,2003,22(3):950-953. DOI: 10.1145/882262.882368
[19]	VOLLMER J,MENCL R,MUELLER H. Improved laplacian smoothing of noisy surface meshes[J]. Computer Graphics Forum,1999,18(3):131-138. DOI: 10.1111/1467-8659.00334
[20]	DESBRUN M,MEYER M,SCHRODER P,et al. Implicit fairing of irregular meshes using diffusion and curvature flow[C]. The 26th Annual Conference on Computer Graphics and Interactive Techniques,1999:317-324. DOI: 10.1145/311535.311576.
[21]	COMANICIU D,RAMESH V,MEER P. The variable bandwidth mean shift and data-driven scale selection[C]. The 8th IEEE International Conference on Computer Vision,Vancouver,2001:438-445.
[22]	GEORGESCU B,SHIMSHONI I,MEER P. Mean shift based clustering in high dimensions:a texture classification example[C]. The 9th IEEE International Conference on Computer Vision,Nice,2003:456-463.
[23]	COMANICIU D,MEER P. Mean shift analysis and applications[C]. The 7th IEEE International Conference on Computer Vision,Kerkyra,1999:1197-1203.
[24]	XU Xiangyang,YANG Hao,NEUMANN I. Time-efficient filtering method for three-dimensional point clouds data of tunnel structures[J]. Advances in Mechanical Engineering,2018,10(5). DOI: 10.1177/16878140187731.
[25]	余鹏磊,于海洋,谢秋平,等. 基于RANSAC算法的地铁隧道点云数据滤波[J]. 测绘工程,2014,23(8):28-31. DOI: 10.3969/j.issn.1006-7949.2014.08.007 YU Penglei,YU Haiyang,XIE Qiuping,et al. Filtering for subway tunnel point cloud data based on RANSAC algorithm[J]. Engineering of Surveying and Mapping,2014,23(8):28-31. DOI: 10.3969/j.issn.1006-7949.2014.08.007
[26]	ZHU Ningning,JIAA Y,LUO Lun. Tunnel point cloud filtering method based on elliptic cylindrical model[J]. The International Archives of the Photogrammetry,Remote Sensing and Spatial Information Sciences,2016,41:735-740.
[27]	陈建华,马宝,王蒙. 基于二次特征提取的煤矿巷道表面点云数据精简方法[J]. 工矿自动化,2023,49(12):114-120. CHEN Jianhua,MA Bao,WANG Meng. A method for simplifying surface point cloud data of coal mine roadways based on secondary feature extraction[J]. Journal of Mine Automation,2023,49(12):114-120.