﻿ 扫描线段特征用于三维点云地面分割
 光电工程  2019, Vol. 46 Issue (7): 180268      DOI: 10.12086/oee.2019.180268

Ground segmentation from 3D point cloud using features of scanning line segments
Cheng Ziyang, Ren Guoquan, Zhang Yin
Department of Vehicle and Electrical Engineering, Army Engineering University, Shijiazhuang, Hebei 050003, China
Abstract: Aiming at the problem of accurately segmenting the ground in real-time from 3D LiDAR point cloud, a ground segmentation algorithm based on the features of scanning line segments is proposed. The algorithm first performs de-noising and pose correction on the 3D point cloud, then divides the scanning line according to the Euclidean distance and absolute height difference between adjacent points. Next, the characteristics of the adjacent line segments such as spacing, slope, and absolute height difference are analyzed. The maximum likelihood estimation is used to solve the feature threshold function, which improves the adaptability of threshold. Finally, comprehensively considering the undulating and inclined complex terrain, the scanning line segments are marked as segments of flat ground, segments of slope and segments of obstacle by formulating the new horizontal and vertical classification strategies. This algorithm has been successfully applied to the unmanned ground platform. The usage and comparative test show that the algorithm can detect the ground stably and efficiently in both urban and field scenarios.
Keywords: 3D LiDAR    ground segmentation    segment features    complex terrain    real-time

1 引言

2 研究现状

3 分割算法

 图 1 扫描线分布样式 Fig. 1 Distribution pattern of scanning lines
3.1 扫面线分段

1) 当相邻点的绝对高度差小于阈值${T_{\rm{h}}}$时，如果当前点和前一个点的距离R小于阈值${T_{\rm{r}}}$，则被判定属于同一扫描线段；如果距离大于阈值${T_{\rm{r}}}$，则将前一个点标记为线段的终止端点，当前点标记为新一条线段的起始端点。

2) 当相邻点的绝对高度差大于阈值${T_{\rm{h}}}$时，则将前一个点标记为线段的终止端点，当前点标记为新一条线段的起始端点。本文使用的Velodyne HDL 32线激光雷达的角分辨率为0.16°，因此相邻扫描点间的距离理论上为

 $R = \frac{{D \times 0.16{\rm{ \mathsf{ π} }}}}{{180}},$ (1)

3.2 特征提取

3.2.1 平坦地形

3.2.2 颠簸地形

 ${d_{{\rm{p}} - c}} = \frac{{h \sin\theta }}{{\sin (90^\circ - \theta + \beta )}},$ (2)

 ${d_{{\rm{p}} - f}} = \frac{{h \sin(\theta - 1.33)}}{{\sin (90^\circ - (\theta - 1.33) + \beta )}}\\ = \frac{{h \sin(\theta - 1.33)}}{{\sin (91.33^\circ - \theta + \beta )}},$ (3)

 ${T_{\rm{d}} } = \mu \times [\frac{{h \sin\theta }}{{\sin (90^\circ - \theta + \beta )}} - \frac{{h \sin(\theta - 1.33^\circ )}}{{\sin (91.33^\circ - \theta + \beta )}}],$ (4)

3.2.3 斜坡地形

 ${S_{\rm{f}}} = \frac{1}{{\frac{{\sum\nolimits_{k = 1}^K {{D_k}} }}{K} - \frac{{\sum\nolimits_{n = 1}^N {{D_n}} }}{N}}} \times \left( {\frac{{\sum\nolimits_{k = 1}^K {{H_k}} }}{K} - \frac{{\sum\nolimits_{n = 1}^N {{H_n}} }}{N}} \right),$ (5)

3.3 算法流程

 图 2 算法流程图 Fig. 2 Algorithm flowchart
4 实验与分析

 图 3 地面无人实验平台 Fig. 3 Ground unmanned experimental platform
4.1 对比试验

 图 4 试验场景 Fig. 4 Test scenarios

 图 5 城市结构化道路场景的点云分割结果。(a)人工标记的结果；(b)本文方法的分割结果；(c)文献[13]算法的分割结果；(d)文献[14]算法的分割结果 Fig. 5 Segmentation results in urban structured road scene. (a) Results for human remark; (b) Segmentation results of this paper; (c) Segmentation results in ref.[13]; (d) Segmentation results in ref.[14]

 图 6 野外起伏道路场景的点云分割结果。(a)人工标记的结果；(b)本文方法的分割结果；(c)文献[13]算法的分割结果；(d)文献[14]算法的分割结果 Fig. 6 Segmentation results in wild undulating road scene. (a) Results for human remark; (b) Segmentation results of this paper; (c) Segmentation results in ref.[13]; (d) Segmentation results in ref.[14]

 图 7 复杂起伏场景的点云分割结果。(a)人工标记的结果；(b)本文方法的分割结果；(c)文献[13]算法的分割结果；(d)文献[14]算法的分割结果 Fig. 7 Segmentation results in complex undulating scene. (a) Results for human remark; (b) Segmentation results of this paper; (c) Segmentation results in ref.[13]; (d) Segmentation results in ref.[14]

4.2 定量评估

 $T_{\mathrm{TPR}}=\frac{T_{\mathrm{p}}}{T_{\mathrm{p}}+F_{\mathrm{N}}},$ (6)
 $F_{\mathrm{FPR}}=\frac{F_{\mathrm{P}}}{F_{\mathrm{p}}+T_{\mathrm{N}}},$ (7)

 试验场景 分割算法 TPR/% FPR/% 城市道路 文献[13]方法 95.46 4.52 文献[14]方法 93.75 5.09 本文方法 94.52 4.32 野外道路 文献[13]方法 85.23 14.62 文献[14]方法 89.41 10.55 本文方法 91.92 7.96 复杂颠簸场景 文献[13]方法 77.73 21.47 文献[14]方法 72.92 27.51 本文方法 90.94 8.53

 图 8 不同算法的分割耗时。(a)城市结构化道路场景；(b)野外颠簸道路场景；(c)复杂起伏场景 Fig. 8 Elapsed time of different algorithms. (a) Urban structured road scene; (b) Wild bumpy road scene; (c) Complex undulating scene
4.3 综合分析

5 结论

 [1] Gong Y P, Jin T, Tong S G. Segmentation method for point cloud data[J]. Journal of Engineering Graphics, 2006, 27(4): 8-13. 龚友平, 金涛, 童水光. 点云数据区域分割方法[J]. 工程图学学报, 2006, 27(4): 8-13 [Crossref] [2] Wu L S, Yu T, Chen H W. Reduction algorithm of point cloud segmentation based on adaptive elliptical distance[J]. Computer Applications and Software, 2016, 33(2): 42-45. 吴禄慎, 俞涛, 陈华伟. 基于自适应椭圆距离的点云分区精简算法[J]. 计算机应用与软件, 2016, 33(2): 42-45 [Crossref] [3] Huangfu Z M, Yan L, Zhang S. A new method for estimation of normal vector and curvature based on scattered point cloud[J]. Journal of Computational Information Systems, 2012, 8(19): 7937-7945. [4] Klasing K, Wollherr D, Buss M. A clustering method for efficient segmentation of 3D laser data[C]//Proceedings of 2008 IEEE International Conference on Robotics and Automation, 2008: 4043-4048. [5] Rabbani T, Van den Heuvel F A, Vosselman M G. Segmentation of point clouds using smoothness constraints, international archives of photogrammetry[J]. Remote Sensing and Spatial Information Sciences, 2006, 36(5): 248-253. [6] Vo A V, Truong-Hong L, Laefer D F, et al. Octree-based region growing for point cloud segmentation[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 104: 88-100. [Crossref] [7] Guo B Q, Yu Z J, Zhang N, et al. 3D point cloud segmentation, classification and recognition algorithm of railway scene[J]. Chinese Journal of Scientific Instrument, 2017, 38(9): 2103-2111. 郭保青, 余祖俊, 张楠, 等. 铁路场景三维点云分割与分类识别算法[J]. 仪器仪表学报, 2017, 38(9): 2103-2111 [Crossref] [8] Hoffman R, Jain A K. Segmentation and classification of range images[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1987, PAMI-9(5): 608-620. [Crossref] [9] Zhang X M, Wan W G, Xiao L, et al. Mean shift clustering segmentation and RANSAC simplification of color point cloud[C]//Proceedings of 2014 International Conference on Audio, Language and Image Processing, 2014: 837-841. [10] Kisner H, Thomas U. Segmentation of 3D point clouds using a new spectral clustering algorithm without a-priori knowledge[C]//Proceedings of the International Conference on Computer Vision Theory and Applications, 2018: 315-322. [11] Kuçak R A, Özdemir E, Erol S. The segmentation of point clouds with k-means and ANN (artifical neural network)[J]. ISPRS-International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2017, 38(XLⅡ-1/W1): 595-598. [12] Wang X H, Wu L S, Chen H W, et al. Region segmentation of point cloud data based on improved particle swarm optimization fuzzy clustering[J]. Optics and Precision Engineering, 2017, 25(4): 1095-1105. 王晓辉, 吴禄慎, 陈华伟, 等. 应用改进的粒子群优化模糊聚类实现点云数据的区域分割[J]. 光学精密工程, 2017, 25(4): 1095-1105 [Crossref] [13] Rummelhard L, Paigwar A, Nègre A, et al. Ground estimation and point cloud segmentation using SpatioTemporal conditional random field[C]//Proceedings of 2017 IEEE Intelligent Vehicles Symposium, 2017: 1105-1110. [14] Fan L, Liu J H, Huang Q Q. Point cloud segmentation algorithm based on feature fusion used for understory environments[J]. Journal of Beijing Forestry University, 2016, 38(5): 133-138. 樊丽, 刘晋浩, 黄青青. 基于特征融合的林下环境点云分割[J]. 北京林业大学学报, 2016, 38(5): 133-138 [Crossref] [15] Moosmann F, Pink O, Stiller C. Segmentation of 3D Lidar data in non-flat urban environments using a local convexity criterion[C]//Proceedings of 2009 IEEE Intelligent Vehicles Symposium, 2009: 215-220. [16] Hernandez J, Marcotegui B. Point cloud segmentation towards urban ground modeling[C]//Proceedings of 2009 Joint Urban Remote Sensing Event, 2009: 1-5. [17] Himmelsbach M, Hundelshausen F V, Wuensche H J. Fast segmentation of 3D point clouds for ground vehicles[C]//Proceedings of 2010 IEEE Intelligent Vehicles Symposium, 2010: 560-565. [18] Zhu Z, Liu J L. Real-time Markov random field based ground segmentation of 3D Lidar data[J]. Journal of Zhejiang University (Engineering Science), 2015, 49(3): 464-469. 朱株, 刘济林. 基于马尔科夫随机场的三维激光雷达路面实时分割[J]. 浙江大学学报(工学版), 2015, 49(3): 464-469 [Crossref] [19] Zhao K, Xu Y C, Wang R D, et al. A preprocessing method of 3D point clouds registration in urban environments[J]. Opto-Electronic Engineering, 2018, 45(12): 180266. 赵凯, 徐友春, 王任栋. 一种城市环境三维点云配准的预处理方法[J]. 光电工程, 2018, 45(12): 180266 [Crossref]