基于多特征约束的露天采场道路点云提取

doi:10.12068/j.issn.1005-3026.2024.09.014

摘要/Abstract

摘要：

针对露天采场道路点云数据通过法向量、路缘石等点云特征难以准确提取的问题，提出了一种多特征约束的露天采场道路点云提取方法.以辽阳市千山石灰石矿露天采场激光点云为数据源，首先对原始数据进行降采样；然后基于单点RGB信息、邻域RGB信息、邻域高差、邻域粗糙度、反射强度5类点云特征，制作并划分了训练集和验证集，利用随机森林算法构建了道路点云提取模型并进行了优化，进一步引入欧式聚类算法改进了道路点云提取模型结果，最后评估了露天采场道路点云提取结果.结果表明，本文方法可以实时有效准确地提取露天采场道路点云数据.

关键词: 露天采场, 道路点云, 点云特征信息, 随机森林算法, 欧式聚类算法

Abstract:

Aiming at the problem that road point cloud data in open pit is difficult to be accurately extracted through point cloud features such as normal vector and kerb， a method of road point cloud extraction in open pit with multi?feature constraints was proposed. Taking the laser point cloud in the open pit of Qianshan limestone mine in Liaoyang City as the data source， the original data was downsampled firstly， and then the training set and verification set were made and divided based on the five kinds of point cloud features including single point RGB information， neighborhood RGB information， neighborhood height difference， neighborhood roughness， and reflection intensity. The road point cloud extraction model was constructed and optimized using the random forest algorithm. Furthermore， European clustering algorithm was introduced to improve the road point cloud extraction model. Finally， the road point cloud extraction results were evaluated in open pit. The results show that the proposed method can effectively and accurately extract the road point cloud data in open pit in real time.

Key words: open pit, road point cloud, point cloud characteristic information, random forest algorithm, European clustering algorithm

中图分类号:

P 231

毛亚纯, 杨哲玺, 曹旺, 齐迹. 基于多特征约束的露天采场道路点云提取[J]. 东北大学学报（自然科学版）, 2024, 45(9): 1326-1333.

Ya-chun MAO, Zhe-xi YANG, Wang CAO, QI Ji. Extraction of Road Point Cloud in Open Pit Based on Multi-feature Constraints[J]. Journal of Northeastern University(Natural Science), 2024, 45(9): 1326-1333.

图/表 13

图1 辽阳市千山石灰石矿（a）—影像数据；（b）—点云数据.

Fig.1 Qianshan limestone mine，Liaoyang City

图2 大疆M300无人机及禅思L1三维激光扫描仪

Fig.2 DJI M300 drone and Zenith L1 three?dimensional laser scanner

图3 露天采场道路点云提取流程图

Fig.3 Flow chart for point cloud extraction of open?pit mine roadways

图4 数据集划分（a）—训练集；（b）—验证集.

Fig.4 Data set partitioning

图5 两种类型点云特征值（a）—单点R值；（b）—单点G值；（c）—单点B值；（d）—邻域R均值；（e）—邻域G均值；（f）—邻域B均值；（g）—邻域高差；（h）—反射强度；（i）—邻域粗糙度.

Fig.5 Two types of point cloud characteristic values

图6 随机森林Bagging思想

Fig.6 Random forest Bagging idea

图7 参数优化（a）—决策树最大深度；（b）—最佳决策树数量；（c）—最大特征数；（d）—叶节点所需的最小样本数；（e）—分割内部节点所需的最小样本数.

Fig.7 Parameter optimization

表1 最优参数总体得分

Table 1 Overall score of optimal parameters

参数名	最优参数	Accuracy	F1	Oob
决策树划分标准	Gini	0.962?773	0.961?037	0.981?782
最佳决策树数量	37	0.963?026	0.961?533	0.985?605
决策树最大深度	10	0.964?306	0.962?922	0.980?794
分割内部节点所需的最小样本数	16	0.964?448	0.963?075	0.980?963
叶节点所需的最小样本数	2	0.963?453	0.961?997	0.981?033
最大特征数	5	0.963?879	0.962?459	0.982?851

图8 露天采场点云分类结果

Fig.8 Classification result of point cloud in open pit

图9 测试集分类后道路点云（a）—聚类前；（b）—聚类后.

Fig.9 Road point cloud after test set classification

图10 目视解译露天采场道路点云

Fig.10 Visual interpretation of road point cloud in open pit

表2 露天采场道路点云提取误差

Table 2 Extraction errors of road point cloud in open pit

类别	正确点个数	错误点个数	总数	误差/%
道路点	90?854	1?622	92?476	1.754
非道路点	1 902 353	6 793	1?909?146	0.356
总数	1 993 207	8 415	2?001?622	0.420

表3 露天采场道路点云判定误差

Table 3 Judgment errors of road point cloud in open pit

类别	道路判定误差/%
正确提取	93.043
错误提取	6.957

参考文献 20

1	Tong X H， Liu X F， Chen P，et al.Integration of UAV‑based photogrammetry and terrestrial laser scanning for the three‑dimensional mapping and monitoring of open‑pit mine areas［J］.Remote Sensing，2015，7（6）：6635-6662.
2	Ma L F， Li Y， Li J，et al.Mobile laser scanned point‑clouds for road object detection and extraction：a review［J］.Remote Sensing，2018，10（10）：1531.
3	Huang W X， Liang H W， Lin L L，et al.A fast point cloud ground segmentation approach based on coarse‑to‑fine Markov random field［J］.IEEE Transactions on Intelligent Transportation Systems，2022，23（7）：7841-7854.
4	Mi X X， Yang B S， Dong Z，et al.Automated 3D road boundary extraction and vectorization using MLS point clouds［J］.IEEE Transactions on Intelligent Transportation Systems，2022，23（6）：5287-5297.
5	Li K， Shao J J， Guo D.A multi‑feature search window method for road boundary detection based on LIDAR data［J］.Sensors，2019，19（7）：1551.
6	Lu X W， Ai Y F， Tian B.Real‑time mine road boundary detection and tracking for autonomous truck［J］.Sensors，2020，20（4）：1121.
7	Kumar P， Lewis P， McElhinney C P，et al.Snake energy analysis and result validation for a mobile laser scanning data‑based automated road edge extraction algorithm［J］.IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing，2017，10（2）：763-773.
8	Xiong H Y， Zhu T H， Liu Y，et al.Road‑model‑based road boundary extraction for high definition map via LIDAR［J］.IEEE Transactions on Intelligent Transportation Systems，2022，23（10）：18456-18465.
9	程效军，郭王，李泉，等.基于强度与颜色信息的地面LiDAR点云联合分类方法［J］.中国激光，2017，44（10）：273-280.
	Cheng Xiao‑jun， Guo Wang， Li Quan，et al.Joint classification method for terrestrial LiDAR point cloud based on intensity and color information［J］.Chinese Journal of Lasers，2017，44（10）：273-280.
10	Yadav M.A multi‑constraint combined method for road extraction from airborne laser scanning data［J］.Measurement，2021，186：110077.
11	Grilli E， Remondino F.Machine learning generalisation across different 3D architectural heritage［J］.ISPRS International Journal of Geo‑information，2020，9（6）：379.
12	Biçici S， Zeybek M.Effectiveness of training sample and features for random forest on road extraction from unmanned aerial vehicle‑based point cloud［J］.Transportation Research Record：Journal of the Transportation Research Board，2021，2675（12）：401-418.
13	Ferraz A， Mallet C， Chehata N.Large scale road network extraction in forested moutainous areas using airborne laser scanning data［C］//2014 IEEE Geoscience and Remote Sensing Symposium.Quebec City：IEEE，2014：4315-4318.
14	Ferraz A， Mallet C， Chehata N.Large‑scale road detection in forested mountainous areas using airborne topographic lidar data［J］.ISPRS Journal of Photogrammetry and Remote Sensing，2016，112：23-36.
15	Deng L Y， Yang M， Liang Z D，et al.Fusing geometrical and visual information via superpoints for the semantic segmentation of 3D road scenes［J］.Tsinghua Science and Technology，2020，25（4）：498-507.
16	Ma L F， Li Y， Li J，et al.BoundaryNet：extraction and completion of road boundaries with deep learning using mobile laser scanning point clouds and satellite imagery［J］.IEEE Transactions on Intelligent Transportation Systems，2022，23（6）：5638-5654.
17	张爱武，刘路路，张希珍.道路三维点云多特征卷积神经网络语义分割方法［J］.中国激光，2020，47（4）：269-277.
	Zhang Ai‑wu， Liu Lu‑lu， Zhang Xi‑zhen.Multi‑feature 3D road point cloud semantic segmentation method based on convolutional neural network［J］.Chinese Journal of Lasers，2020，47（4）：269-277.
18	李健，姚亮.融合多特征深度学习的地面激光点云语义分割［J］.测绘科学，2021，46（3）：133-139，162.
	Li Jian， Yao Liang.Ground laser point cloud semantic segmentation based on multi‑feature deep learning［J］.Science of Surveying and Mapping，2021，46（3）：133-139，162.
19	Breiman L.Random forests［J］.Machine Language，2001，45（1）：5-32.
20	毛亚纯，伏雨文，曹旺，等.基于无人机点云数据的露天采场矿车提取方法［J］.东北大学学报（自然科学版），2021，42（6）：842-848，863.
	Mao Ya‑chun， Fu Yu‑wen， Cao Wang，et al.Extraction method of open pit mine car based on UAV point cloud data［J］.Journal of Northeastern University （Natural Science），2021，42（6）：842-848，863.

[1]	丁敬国，郭锦华. 基于主成分分析协同随机森林算法的热连轧带钢宽度预测[J]. 东北大学学报（自然科学版）, 2021, 42(9): 1268-1275.
[2]	毛亚纯，伏雨文，曹旺，赵占国. 基于无人机点云数据的露天采场矿车提取方法[J]. 东北大学学报（自然科学版）, 2021, 42(6): 842-849.