使用局部几何特征与空洞邻域的点云语义分割

项学泳; 李广云; 王力; 宗文鹏; 吕志鹏; 向奉卓

doi:10.13203/j.whugis20200567

使用局部几何特征与空洞邻域的点云语义分割

doi: 10.13203/j.whugis20200567

1 信息工程大学地理空间信息学院, 河南郑州, 450000;
2 西安测绘研究所, 陕西西安, 710000;
3 61287 部队, 四川成都, 610036

基金项目:

国家自然科学基金（42071454），国家重点研发计划资助（2017YFF0206000）。

详细信息

作者简介:
项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

Semantic segmentation of point clouds using local geometric features and dilated neighborhoods

1 Institute of Geospatial Information, Information Engineering University, Zhengzhou, 450000, China;
2 Xi'an Institute of Surveying and Mapping, Xi'an, 710000, China;
3 61287 Troops, Chengdu, 610036, China

Funds:

The National Natural Science Foundation of China (42071454)

摘要: 点云具有数据量大、无拓扑结构等特点，现有的深度学习语义分割模型难以充分挖掘大范围邻域内点云中所隐藏的几何特征。为此，提出一种基于空洞邻域并结合角度等几何特征作为模型输入的点云语义分割模型。首先，在局部邻域构建过程中，将图像处理的空洞卷积操作扩展至点云，建立空洞领域结构，以扩大感受野。其次，在特征提取过程中，采用中心点与邻域点之间相对坐标、距离、角度等基本几何特征作为模型输入，最大程度挖掘邻域内的几何特征。最后，基于本文提出的邻域结构与特征提取算法构建了点云语义分割模型。采用Semantic3D数据集进行实验验证，结果表明：本文提出的模型分割效果优于对比的点云语义分割算法，空洞邻域与局部几何输入特征能够有效改善点云语义模型的性能。
- 点云 /
- 语义分割 /
- 空洞邻域 /
- 几何特征
Abstract: Objectives Point cloud has no topological structure, current deep learning semantic segmentation algorithm is difficult to capture geometric features implied in irregular points. In addition, the point cloud is in three-dimensional space with a large amount of data size. If we blindly expand the captive field size during extract neighborhood information, it will increase the number of model parameters, which will make model training difficult. Methods To this end, we propose a point cloud semantic segmentation model based on the dilated convolution and combining elementary geometric features such as angle as the model input. First of all, during feature extraction, basic geometric features such as the relative coordinates, distance and angle between the centroid and the neighboring points are used as the model input to mine the geometric information. Secondly, in the process of building local neighborhoods, we expand the image dilated convolution operator to point cloud processing, the point cloud dilated operator can expand the receptive field size with no increasing the number of parameters of the model. Then the dilated convolution operator, multi-geometric features encoding modules and U-Net architecture are combined to form a complete point cloud semantic segmentation model. Results Semantic3D is applied to verify the proposed algorithms. The results show that compared with the traditional neighborhood structure, the OA of dilated neighborhood structure is increased by 1.4%. Compared with the model that only uses coordinates as input, multi-geometric features encoding module is increased by 10.7%. The final model based on the two proposed algorithms get mIoU and OA are 91.2% and 68.2%, respectively. Conclusions The dilated neighborhood structure can effectively extract point cloud information in a larger range without increasing the number of model parameters. multi-geometric features encoding module can maximize the capture of shape information in the neighborhood.
- point cloud /
- semantic segmentation /
- dilated neighborhood /
- geometric features

[1]	Hackel T, Wegner J D, Schindler K. Fast Semantic Segmentation of 3D Point Clouds With Strongly Varying Density[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2016, 3(3):177-184
[2]	Castillo E, Liang J, Zhao H. Point Cloud Segmentation and Denoising Via Constrained Nonlinear Least Squares Normal Estimates[M]. Innovations for Shape Analysis. Springer, Berlin, Heidelberg, 2013
[3]	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(7):88-100
[4]	Andres M, Ignacio V, Jens B, et al. RangeNet++:Fast and Accurate Lidar Semantic Segmentation[C]. IEEE International Conference on Intelligent Robots and Systems, Macau, China, 2019
[5]	Maturana, Daniel, Sebastian Scherer. Voxnet:A 3d Convolutional Neural Network for Real-time Object Recognition[C]. IEEE International Conference on Intelligent Robots and Systems, Hamburg, Germany, 2015
[6]	Qi C, Su H, Mo K, et al. PointNet:Deep Learning on Point Sets for 3d Classification and Segmentation[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 2017
[7]	Qi C, Yi L, Su H, et al. PointNet++:Deep Hierarchical Feature Learning on Point Sets in a Metric Space[C]. Advances in Neural Information Processing Systems, Long Beach, CA, USA, 2017
[8]	Yu, Fisher, and Vladlen Koltun. Multi-scale context aggregation by dilated convolutions[C]. International Conference on Learning Representations, San Juan, Puerto Rico, 2016
[9]	Jiang M, Wu Y, Zhao T, et al. Pointsift:a Sift-like Network Module for 3D Point Cloud Semantic Segmentation[J]. Arxiv Preprint, ArXiv:1807.00652, 2018
[10]	Li Y, Bu R, Sun M, et al. Pointcnn:Convolution on X-Transformed Points[C]. Advances in Neural Information Processing Systems, Montréal, Canada, 2018
[11]	Liang Z, Yang M, Li H, et al. 3D Instance Embedding Learning With a Structure-Aware Loss Function for Point Cloud Segmentation[J]. IEEE Robotics and Automation Letters, 2020, 5(3):4915-4922
[12]	Ronneberger O, Fischer P, Brox T. U-Net:Convolutional Networks for Biomedical Image Segmentation[C]. International Conference on Medical image computing and computer-assisted intervention, Munich, Germany, 2015
[13]	Zhao H, Jiang L, Fu C. PointWeb:Enhancing local neighborhood features for point cloud processing[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, CA, USA, 2019
[14]	Thomas H, Qi C, Deschaud J, et al. Kpconv:Flexible and Deformable Convolution for Point Clouds[C]. Proceedings of the IEEE International Conference on Computer Vision, Seoul, South Korea, 2019
[15]	Hackel T, Savinov N, Ladicky L, et al. Semantic3D Net:a New Large-scale Point Cloud Classification Benchmark[C]. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2017, (IV-1)W1:91-98
[16]	Wang F, Zhuang Y, Gu H, et al. OctreeNet. A Novel Sparse 3-D Convolutional Neural Network for Real-Time 3-D Outdoor Scene Analysis[J]. IEEE Transactions on Automation Science and Engineering, 2019, 17(2):735-747
[17]	Contreras J, Joachim D. Edge-Convolution Point Net for Semantic Segmentation of Large-Scale Point Clouds[C]. IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 2019
[18]	Thomas H, Goulette F, Deschaud J, et al. Semantic Classification of 3D Point Clouds with Multiscale Spherical Neighborhoods[C]. International Conference on 3D Vision, Verona, Italy, 2018
[19]	Roynard X, Deschaud J, Goulett F. Classification of Point Cloud Scenes with Multiscale Voxel Deep Network[J]. ArXiv Preprint, ArXiv:1804.03583, 2018

[1]	杨钰琪, 陈驰, 杨必胜, 胡平波, 崔扬. 基于UAV影像密集匹配点云多层次分割的建筑物层高变化检测 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20190030
[2]	孙文潇, 王健, 梁周雁, 马伟丽, 陈喆. 法线特征约束的激光点云精确配准 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20180315
[3]	张瑞菊, 周欣, 赵江洪, 曹闵. 一种古建筑点云数据的语义分割算法 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20180428
[4]	蒋腾平, 杨必胜, 周雨舟, 朱润松, 胡宗田, 董震. 道路点云场景双层卷积语义分割 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20200081
[5]	张蕊, 李广云, 王力, 李明磊, 周阳林. 车载LiDAR点云混合索引新方法 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20160441
[6]	危双丰, 刘明蕾, 赵江洪, 黄帅. 利用点云检测室内导航元素的方法综述 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20180144
[7]	杨军, 林岩龙, 张瑞峰, 王小鹏. 一种大规模点云k邻域快速搜索算法 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20140191
[8]	卢昊, 庞勇, 徐光彩, 李增元. 机载激光雷达全波形数据与系统点云差异的定量分析 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20130443
[9]	应申, 毛政元, 李霖, 许光. 利用3D Voronoi图的兔子点云聚类分割 . 武汉大学学报 ● 信息科学版,
[10]	王永波, 杨化超, 刘燕华, 牛晓楠. 线状特征约束下基于四元数描述的LiDAR点云配准方法 . 武汉大学学报 ● 信息科学版,
[11]	万雪, 张祖勋, 柯涛. 一种利用零交叉点理论的改进SIFT特征提取算法 . 武汉大学学报 ● 信息科学版,
[12]	康志忠, 王薇薇, 李珍. 多源数据融合的三维点云特征面分割和拟合一体化方法 . 武汉大学学报 ● 信息科学版,
[13]	孙毅中, 姚驰, 陈少勤, 徐文祥. 顾及几何特征的地理要素空间位置唯一性标识方法 . 武汉大学学报 ● 信息科学版,
[14]	孙杰, 马洪超, 钟良. 利用LiDAR点云的真正射影像遮蔽检测 . 武汉大学学报 ● 信息科学版,
[15]	郑莉, 张剑清, 罗跃军. 多视结构光点云的自动无缝拼接 . 武汉大学学报 ● 信息科学版,
[16]	巫兆聪, 覃茂运, 张潇. 一种顾及几何特征的云模型遥感影像分割方法 . 武汉大学学报 ● 信息科学版,
[17]	万幼川, 徐景中, 赖旭东, 张圣望. 基于多分辨率方向预测的LIDAR点云滤波方法 . 武汉大学学报 ● 信息科学版,
[18]	黄先锋, 陶闯, 江万寿, 龚健雅. 机载激光雷达点云数据的实时渲染 . 武汉大学学报 ● 信息科学版,
[19]	史文中. 三维地理信息系统中几何特征的误差模型 . 武汉大学学报 ● 信息科学版,
[20]	邵振峰, 孙悦鸣, 席江波, 李岩. 智能优化学习的高空间分辨率遥感影像语义分割 . 武汉大学学报 ● 信息科学版, doi: 10.13203/j.whugis20200640

点击查看大图

计量

文章访问数: 53
HTML全文浏览量: 10
PDF下载量: 7
被引次数: 0

姓名
邮箱
手机号码
标题
留言内容
验证码

留言板

使用局部几何特征与空洞邻域的点云语义分割

doi: 10.13203/j.whugis20200567

作者简介:
项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

Semantic segmentation of point clouds using local geometric features and dilated neighborhoods

计量

使用局部几何特征与空洞邻域的点云语义分割

doi: 10.13203/j.whugis20200567

1 信息工程大学地理空间信息学院, 河南郑州, 450000;

2 西安测绘研究所, 陕西西安, 710000;

3 61287 部队, 四川成都, 610036

作者简介:
项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

English Abstract

Semantic segmentation of point clouds using local geometric features and dilated neighborhoods

1 Institute of Geospatial Information, Information Engineering University, Zhengzhou, 450000, China;

2 Xi'an Institute of Surveying and Mapping, Xi'an, 710000, China;

3 61287 Troops, Chengdu, 610036, China

全文HTML

目录

留言板

使用局部几何特征与空洞邻域的点云语义分割

doi: 10.13203/j.whugis20200567

作者简介: 项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

Semantic segmentation of point clouds using local geometric features and dilated neighborhoods

计量

出版历程

使用局部几何特征与空洞邻域的点云语义分割

doi: 10.13203/j.whugis20200567

1 信息工程大学地理空间信息学院, 河南 郑州, 450000; 2 西安测绘研究所, 陕西 西安, 710000; 3 61287 部队, 四川 成都, 610036

作者简介: 项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

English Abstract

Semantic segmentation of point clouds using local geometric features and dilated neighborhoods

1 Institute of Geospatial Information, Information Engineering University, Zhengzhou, 450000, China; 2 Xi'an Institute of Surveying and Mapping, Xi'an, 710000, China; 3 61287 Troops, Chengdu, 610036, China

全文HTML

目录

作者简介:
项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

1 信息工程大学地理空间信息学院, 河南郑州, 450000;

2 西安测绘研究所, 陕西西安, 710000;

3 61287 部队, 四川成都, 610036

作者简介:
项学泳,博士生,主要从事三维场景环境感知研究,ahhsxxy@163.com。

1 Institute of Geospatial Information, Information Engineering University, Zhengzhou, 450000, China;

2 Xi'an Institute of Surveying and Mapping, Xi'an, 710000, China;

3 61287 Troops, Chengdu, 610036, China