广西师范大学学报（自然科学版） ›› 2023, Vol. 41 ›› Issue (5): 37-48.doi: 10.16088/j.issn.1001-6600.2023021901

• 研究论文 • 上一篇    下一篇

结合残差动态图卷积与特征强化的点云分类

梁正友1,2*, 蔡俊民1, 孙宇1, 陈磊1   

  1. 1.广西大学 计算机与电子信息学院, 广西 南宁 530004;
    2.广西多媒体通信与网络技术重点实验室(广西大学), 广西 南宁 530004
  • 收稿日期:2023-02-19 修回日期:2023-04-02 发布日期:2023-10-09
  • 通讯作者: 梁正友(1968—),男,广西崇左人,广西大学教授,博士。E-mail: zhyliang@gxu.edu.cn
  • 基金资助:
    国家自然科学基金(62171145)

Point Cloud Classification Based on Residual Dynamic Graph Convolution and Feature Enhancement

LIANG Zhengyou1,2*, CAI Junmin1, SUN Yu1, CHEN Lei1   

  1. 1. School of Computer and Electronics Information, Guangxi University, Nanning Guangxi 530004, China;
    2. Guangxi Key Laboratory of Multimedia Communications and Network Technology (Guangxi University), Nanning Guangxi 530004, China
  • Received:2023-02-19 Revised:2023-04-02 Published:2023-10-09

PDF (PC)

64

摘要： 针对提高点云分类性能和鲁棒性的需求,本文提出将残差动态图卷积与特征强化相结合的点云分类网络。采用多方向编码方法,在局部邻域中心点的空间多方向上选取近邻点,丰富点云特征;通过残差动态图卷积提取特征,以残差结构对局部特征和全局特征进行深度融合,有效缓解网络退化问题;构造强化空间注意力模块,使得网络在空间域中学习自适应地为不同邻域特征分配权重,增强有用特征,并抑制冗余特征;使用高低层次链接,保留更多特性信息。实验表明:本文模型在ModelNet10、ModelNe40数据集上的总体分类精度分别达到94.81%、93.62%,分类精度更高,鲁棒性更强,优于现有先进方法。

关键词: 点云分类, 残差学习, 动态图卷积, 空间注意力, 鲁棒性

Abstract: In order to improve the performance and robustness of point cloud classification, a point cloud classification network combining residual dynamic graph convolution and feature enhancement is proposed. Multi-direction coding method is adopted to select the nearest neighbor points in the spatial multi-direction of the local neighborhood center point to enrich the point cloud features. The residual dynamic graph convolution is used to extract features, and the residual structure is used to deeply fuse local features and global features to effectively alleviate the network degradation problem. The reinforcement spatial attention module is constructed so that the network can learn to adaptively assign weights to different neighborhood features in the spatial domain, enhance useful features, and suppress redundant features. High-level links are used to retain more feature information. Experiments show that the overall classification accuracy of the above models on the ModelNet10 and ModelNe40 data sets is 94.81% and 93.62% respectively, with higher classification accuracy and stronger robustness, which is superior to the existing advanced methods.

Key words: point cloud classification, residual learning, dynamic graph convolution, spatial attention, robustness

中图分类号:  TP391.41

[1] MIRZAEI K, ARASHPOUR M, ASADI E, et al. 3D point cloud data processing with machine learning for construction and infrastructure applications: a comprehensive review[J]. Advanced Engineering Informatics, 2022, 51: 101501. DOI: 10.1016/j.aei.2021.101501.
[2] WANG W G, LAI Q X, FU H Z, et al. Salient object detection in the deep learning era: an in-depth survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(6): 3239-3259. DOI: 10.1109/TPAMI.2021.3051099.
[3] 朱勇建, 罗坚, 秦运柏, 等. 基于光度立体和级数展开法的金属表面缺陷检测方法[J]. 广西师范大学学报(自然科学版), 2020, 38(6): 21-31. DOI: 10.16088/j.issn.1001-6600.2020.06.003.
[4] 薛其威, 伍锡如. 基于多模态特征融合的无人驾驶系统车辆检测[J]. 广西师范大学学报(自然科学版), 2022, 40(2): 37-48. DOI: 10.16088/j.issn.1001-6600.2021072002.
[5] SU H, MAJI S, KALOGERAKIS E, et al. Multi-view convolutional neural networks for 3d shape recognition[C]// 2015 IEEE International Conference on Computer Vision (ICCV). Los Alamitos, CA: IEEE Computer Society, 2015: 945-953. DOI: 10.1109/ICCV.2015.114.
[6] MATURANA D, SCHERER S. VoxNet: a 3D convolutional neural network for real-time object recognition[C]// 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Los Alamitos, CA: IEEE Computer Society, 2015: 922-928. DOI: 10.1109/IROS.2015.7353481.
[7] QI C R, SU H, MO K C, et al. PointNet: deep learning on point sets for 3d classification and segmentation[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2017: 77-85. DOI: 10.1109/CVPR.2017.16.
[8] QI C R, 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 30 (NIPS 2017). Red Hook, NY: Curran Associates Inc., 2017: 5105-5114.
[9] ZHAO H S, JIANG L, FU C W, et al. PointWeb: enhancing local neighborhood features for point cloud processing[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Los Alamitos, CA: IEEE Computer Society, 2019: 5560-5568. DOI: 10.1109/CVPR.2019.00571.
[10] YAN X, ZHENG C D, LI Z, et al. PointASNL: robust point clouds processing using nonlocal neural networks with adaptive sampling[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2020: 5588-5597. DOI: 10.1109/CVPR42600.2020.00563.
[11] LI J X, CHEN B M, LEE G H. SO-Net: self-organizing network for point cloud analysis[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Los Alamitos, CA: IEEE Computer Society, 2018: 9397-9406. DOI: 10.1109/CVPR.2018.00979.
[12] LI Y Y, BU R, SUN M C, et al. PointCNN: convolution on X-transformed points[C]// Advances in Neural Information Processing Systems 31 (NeurIPS 2018). Red Hook, NY: Curran Associates Inc., 2018: 828-838.
[13] WU W X, QI Z A, LI F X. PointConv: deep convolutional networks on 3D point clouds[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2019: 9613-9622. DOI: 10.1109/CVPR.2019.00985.
[14] LIU Y C, FAN B, XIANG S M, et al. Relation-shape convolutional neural network for point cloud analysis[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2019: 8887-8896. DOI: 10.1109/CVPR.2019.00910.
[15] 顾砾, 季怡, 刘纯平. 基于多模态特征融合的三维点云分类方法[J]. 计算机工程, 2021, 47(2): 279-284. DOI: 10.19678/j.issn.1000-3428.0057147.
[16] SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2016: 2818-2826. DOI: 10.1109/CVPR.2016.308.
[17] SIMONOVSKY M, KOMODAKIS N. Dynamic edge-conditioned filters in convolutional neural networks on graphs[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2017: 29-38. DOI: 10.1109/CVPR.2017.11.
[18] WANG Y, SUN Y B, LIU Z W, et al. Dynamic graph CNN for learning on point clouds[J]. ACM Transactions on Graphics, 2019, 38(5): 146. DOI: 10.1145/3326362.
[19] LIN Z H, HUANG S Y, WANG Y C F. Convolution in the cloud: learning deformable kernels in 3D graph convolution networks for point cloud analysis[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2020: 1797-1806. DOI: 10.1109/CVPR42600.2020.00187.
[20] ZHANG K E, HAO M, WANG J, et al. Linked dynamic graph CNN: learning on point cloud via linking hierarchical features[EB/OL]. (2019-04-22)[2023-02-19]. https://arxiv.org/abs/1904.10014. DOI: 10.48550/arXiv.1904.10014.
[21] 梁振明, 翟正利, 周炜. 基于多尺度动态图卷积网络的3D点云分类[J]. 计算机应用与软件, 2021, 38(5): 263-267, 306. DOI: 10.3969/j.issn.1000-386x.2021.05.043.
[22] 魏天琪, 郑雄胜. 基于深度学习的三维点云分类方法研究[J]. 计算机应用研究, 2022, 39(5): 1289-1296. DOI: 10.19734/j.issn.1001-3695.2021.10.0469.
[23] YANG J C, ZHANG Q, NI B B, et al. Modeling point clouds with self-attention and Gumbel subset sampling[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2019: 3318-3327. DOI: 10.1109/CVPR.2019.00344.
[24] 沈露, 杨家志, 周国清, 等. 集自注意力与边卷积的点云分类分割模型[J/OL]. 计算机工程与应用: 1-10[2023-02-19]. http://kns.cnki.net/kcms/detail/11.2127.TP.20221019.1414.006.html.
[25] WU Z R, SONG S R, KHOSLA A, et al. 3D ShapeNets: a deep representation for volumetric shapes[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2015: 1912-1920. DOI: 10.1109/CVPR.2015.7298801.
[26] 王文曦, 李乐林. 深度学习在点云分类中的研究综述[J]. 计算机工程与应用, 2022, 58(1): 26-40. DOI: 10.3778/j.issn.1002-8331.2105-0200.
[27] JIANG M Y, WU Y R, ZHAO T Q, et al. PointSIFT: a SIFT-like network module for 3d point cloud semantic segmentation[EB/OL]. (2018-07-02)[2023-02-19]. https://arxiv.org/abs/1807.00652. DOI: 10.48550/arXiv.1807.00652.
[28] VAN DER MAATEN L, HINTON G. Visualizing data using t-SNE[J]. Journal of Machine Learning Research, 2008, 9: 2579-2605.
[1] 邹艳丽, 汪洋, 刘树生, 姚飞. 带有邻居度信息的容量负载模型下电网级联故障研究[J]. 广西师范大学学报（自然科学版）, 2019, 37(4): 27-36.
[2] 范瑞,蒋品群,曾上游,夏海英,廖志贤,李鹏. 多尺度并行融合的轻量级卷积神经网络设计[J]. 广西师范大学学报（自然科学版）, 2019, 37(3): 50-59.
[3] 唐堂, 魏承赟, 罗晓曙, 丘森辉. 基于附加惯性项人群搜索算法的四旋翼无人机姿态控制研究[J]. 广西师范大学学报（自然科学版）, 2018, 36(4): 12-19.
[4] 许远静, 胡维平. 基于随机森林的不同程度病态嗓音识别[J]. 广西师范大学学报（自然科学版）, 2018, 36(4): 34-41.
[5] 钟海鑫, 丘森辉, 罗晓曙, 唐堂, 杨力, 赵帅. 基于附加惯性项BP神经网络的四旋翼无人机姿态控制研究[J]. 广西师范大学学报（自然科学版）, 2017, 35(2): 24-31.
[6] 钟海鑫,罗晓曙,赵帅,杨力,唐堂. 基于改进精英蚁群系统算法的四旋翼无人机 姿态控制研究[J]. 广西师范大学学报（自然科学版）, 2016, 34(4): 85-92.
[7] 陈锦, 罗晓曙. 一种新的基于小波变换的QPSK信号解调方法[J]. 广西师范大学学报（自然科学版）, 2016, 34(2): 35-45.
[8] 熊顺清, 周卫红. 一种基于非采样Contourlet变换的图像水印算法[J]. 广西师范大学学报（自然科学版）, 2011, 29(2): 195-199.
[9] 李勇, 贾贞. 离散混沌系统在保密通信中的应用[J]. 广西师范大学学报（自然科学版）, 2011, 29(1): 15-19.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
[1] 董淑龙, 马姜明, 辛文杰. 景观视觉评价研究进展与趋势——基于CiteSpace的知识图谱分析[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 1 -13 .
[2] 郭嘉梁, 靳婷. 基于语义增强的多模态情感分析[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 14 -25 .
[3] 吴正清, 曹晖, 刘宝锴. 基于注意力卷积神经网络的中文虚假评论检测[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 26 -36 .
[4] 欧阳舒歆, 王洺钧, 荣垂田, 孙华波. 基于改进LSTM的多维QAR数据异常检测[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 49 -60 .
[5] 李依洋, 曾才斌, 黄在堂. 分数Brown运动驱动的具有壁附着的恒化器模型的随机吸引子[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 61 -68 .
[6] 李鹏博, 李永祥. 外部区域上p-Laplace方程的径向对称解[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 69 -75 .
[7] 吴子弦, 成军, 符坚铃, 周心雯, 谢佳龙, 宁全. 基于PI的Semi-Markovian电力系统事件触发控制设计分析[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 76 -85 .
[8] 程蕾, 闫普选, 杜博豪, 叶思, 邹华红. MOF-2的水相合成及其热稳定和介电性能研究[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 86 -95 .
[9] 刘美余, 张进燕, 周童曦, 廖广凤, 杨新洲, 卢汝梅. 匙羹藤中一个新的C21甾体糖苷及其降血糖活性[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 96 -104 .
[10] 王威, 邓华, 胡乐宁, 李杨. 赤泥-海藻酸钠水凝胶对水中Pb(Ⅱ)的吸附性能[J]. 广西师范大学学报（自然科学版）, 2023, 41(5): 105 -115 .
版权所有 © 广西师范大学学报（自然科学版）编辑部
地址：广西桂林市三里店育才路15号　邮编：541004
电话：0773-5857325　E-mail: gxsdzkb@mailbox.gxnu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发