计算机科学

计算机科学 ›› 2023, Vol. 50 ›› Issue (9): 210-219.doi: 10.11896/jsjkx.220700023

• 数据库&大数据&数据科学 • 上一篇    下一篇

一种使用伪对应点生成的3D点云配准方法

柏正尧, 许祝, 张奕涵   

  1. 云南大学信息学院 昆明 650500
  • 收稿日期:2022-07-04 修回日期:2022-12-09 出版日期:2023-09-15 发布日期:2023-09-01
  • 通讯作者: 柏正尧(baizhy@ynu.edu.cn)
  • 基金资助:
    云南省重大科技专项课题(202002AD080001);云南大学专业学位研究生实践创新基金(2021Y168)

Deep Artificial Correspondence Generation for 3D Point Cloud Registration

BAI Zhengyao, XU Zhu, ZHANG Yihan   

  1. School of Information Science and Engineering,Yunnan University,Kunming 650500,China
  • Received:2022-07-04 Revised:2022-12-09 Online:2023-09-15 Published:2023-09-01
  • About author:BAI Zhengyao,born in 1967,Ph.D,professor,master supervisor.His main research interests include signal proces-sing,image processing,pattern recognition and machine learning,etc.
  • Supported by:
    Yunnan Provincial Major Science and Technology Special Plan Projects(202002AD080001) and Practice & Innovation Foundation for Professional Degree Graduates of Yunnan University(2021Y168).

PDF (PC)

961

摘要： 针对三维重建过程中点云配准存在的挑战性问题(如寻找对应点困难等)展开研究,充分利用源点云和目标点云的几何信息,提出了一种基于交叉注意力和伪对应点生成机制的点云配准方法——深度伪对应点生成(DeepACG)。该方法采用三级网络模型,第一级是深度特征编码模块,利用交叉注意力机制交换和增强两片待配准点云之间的上下文和结构信息;第二级是伪对应点生成模块,基于软映射关系加权合成伪对应点;第三级为对应点加权和离群点过滤模块,赋予每个对应点对不同的权重值并剔除概率较低的离群点。在合成和真实数据集上进行大量实验,DeepACG方法在室内真实数据集3DMatch上的配准召回率达到92.61%;在数据集ModelNet40上进行目标未知的局部点云配准实验,旋转矩阵和平移向量的均方根误差分别降至0.016和0.000 09。实验结果表明,DeepACG配准精度高,鲁棒性强,配准误差低于当前主流的配准方法。

关键词: 交叉注意力, 伪对应点生成, 离群点过滤, 三维点云配准

Abstract: To address the challenging problems of point cloud registration in 3D reconstruction(e.g.,difficulty in finding corresponding points,etc.),this paper proposes a point cloud registration method based on cross-attention and artificial correspondence generation mechanism,Deep Artificial Correspondence Generation(DeepACG),by fully utilizing the geometric information of the source and target point clouds.Our method adopts a three-stage network model.The first stage is the deep feature encoding module,which exchanges and enhances the contextual and structural information between two unaligned point clouds using the cross-attention mechanism.The second stage is the artificial correspondence generation module,which synthesizes the artificial correspondences by weighting the soft mapping.The third one is the correspondence weighting and outlier filtering module,which assigns different weights to the correspondence pairs and rejects them with a small probability.Extensive experiments are conducted on both synthetic and real-world datasets.Our method achieves a registration recall of 92.61% on the real-world indoor dataset 3DMatch,and we execute unseen partial registration experiments on ModelNet40,reducing the root mean square error of the rotation matrix and translation vector to 0.016 and 0.000 09,respectively.Experimental results show that DeepACG has higher registration accuracy and robustness,and its alignment error is lower than that of the existing mainstream registration approaches.

Key words: Cross-attention, Artificial correspondence generation, Outlier filtering, 3D point cloud registration

中图分类号: 

  • TP391
[1]ZERMAS D,IZZAT I,PAPANIKOLOPOULOS N.Fast seg-mentation of 3D point clouds:A paradigm on LiDAR data for autonomous vehicle applications[C]//Proceedings of the IEEE International Conference on Robotics and Automation(ICRA).New York:IEEE,2017:5067-5073.
[2]DESCHAUD J E.IMLS-SLAM:Scan-to-Model Matching Based on 3D Data[C]//Proceedings of the IEEE International Conference on Robotics and Automation(ICRA).Los Alamitos:IEEE Computer Society,2018:2480-2485.
[3]WONG J M,KEE V,LE T,et al.SegICP:Integrated deep semantic segmentation and pose estimation[C]//Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems(IROS) / Workshop on Machine Learning Methods for High-Level Cognitive Capabilities in Robotics.New York:IEEE,2017:5784-5789.
[4]BESL P J,MCKAY N D.A method for registration of 3-Dshapes[C]//Proceedings of the Sensor Fusion Conf:Control Paradigms and Data Structures.Washington:The International Society for Optical Engineering(SPIE),1992:586-606.
[5]YANG J L,LI H D,CAMPBELL D,et al.Go-ICP:A Globally Optimal Solution to 3D ICP Point-Set Registration[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2016,38(11):2241-2254.
[6]DENG H,BIRDAL T,ILIC S.PPFNet:Global Context Aware Local Features for Robust 3D Point Matching[C]//Proceedings of the 31st IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2018:195-205.
[7]CHOY C,PARK J,KOLTUN V.Fully Convolutional Geometric Features[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision(ICCV).New York:IEEE,2019:8957-8965.
[8]LI J,LEE G H.Usip:Unsupervised stable interest point detec-tion from 3d point clouds[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision(ICCV).Los Alamitos:IEEE Computer Society,2019:361-370.
[9]WANG Y,SOLOMON J.PRNet:self-supervised learning forpartial-to-partial registration[C]//Proceedings of the 33rd Conference on Neural Information Processing Systems(NeurIPS).La Jolla,California:Neural Information Processing Systems(NIPS),2019:8814-8826.
[10]WANG Y,SOLOMON J M.Deep Closest Point:Learning Representations for Point Cloud Registration[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision(ICCV).Los Alamitos:IEEE Computer Society,2019:3522-3531.
[11]FENG L,ZHU D M,LI Z X,et al.Sparse Point Cloud Filtering Algorithm Basedon Mask[J].Computer Science,2020,47(12):252-257.
[12]FISCHLER M A,BOLLES R C.Random Sample Consensus:A Paradigm for Model Fitting with Apphcatlons to Image Analysis and Automated Cartography[J].Communications of the ACM,1981,24(6):381-395.
[13]BARATH D,MATAS J.Graph-cut RANSAC[C]//Proceedings of the 31st IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2018:6733-6741.
[14]LI J Y,HU Q W,AI M Y.GESAC:Robust graph enhancedsample consensus for point cloud registration[J].Isprs Journal of Photogrammetry and Remote Sensing,2020,167:363-374.
[15]MACIEL J,COSTEIRA J P.A global solution to sparse correspondence problems[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(2):187-199.
[16]PERERA S,BARNES N.Maximal Cliques Based Rigid BodyMotion Segmentation with a RGB-D Camera[C]//Proceedings of the 11th Asian Conference on Computer Vision.Berlin:Springer,2013:120-133.
[17]ZHOU Q Y,PARK J,KOLTUN V.Fast Global Registration[C]//Proceedings of the 14th European Conference on Computer Vision(ECCV).Cham:Springer International Publishing,2016:766-782.
[18]YANG H,SHI J N,CARLONE L.TEASER:Fast and Certifiable Point Cloud Registration[J].IEEE Transactions on Robotics,2021,37(2):314-333.
[19]CHOY C,DONG W,KOLTUN V.Deep Global Registration[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2020:2511-2520.
[20]PAIS G D,RAMALINGAM S,GOVINDU V M,et al.3DRegNet:A Deep Neural Network for 3D Point Registration[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE:7191-7201.
[21]MARON H,DYM N,KEZURER I,et al.Point Registration via Efficient Convex Relaxation[J].ACM Transactions on Graphics(TOG),2016,35(4):1-12.
[22]ROSEN D M,CARLONE L,BANDEIRA A S,et al.SE-Sync:A certifiably correct algorithm for synchronization over the special Euclidean group[J].The International Journal of Robotics Research,2019,38(2/3):95-125.
[23]IZATT G,DAI H,TEDRAKE R.Globally Optimal Object Pose Estimation in Point Clouds with Mixed-Integer Programming[C]//Proceedings of the 18th International Symposium on Robotics Research(ISRR).Cham:Springer,2020:695-710.
[24]AOKI Y,GOFORTH H,SRIVATSAN R A,et al.PointNetLK:Robust & Efficient Point Cloud Registration using PointNet[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2019:7156-7165.
[25]CHARLES R Q,SU H,KAICHUN M,et al.PointNet:Deep Learning on Point Sets for 3D Classification and Segmentation[C]//Proceedings of the IEEE Conference onComputer Vision and Pattern Recognition(CVPR).New York:IEEE,2017:77-85.
[26]LUCAS B D,KANADE T.An Iterative Image RegistrationTechnique with an Application to Stereo Vision[C]//Procee-dings of the 7th International joint Conference on Artificial Intelligence.San Francisco:Morgan Kaufmann Publishers Inc,1981:674-679.
[27]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):1-12.
[28]SHI W K,ZHANG Z C,YU M J,et al.Point Cloud CoarseAlignment Algorithm Based on Feature Detection and Depth Feature Description[J].Computer Science,2020,47(12):252-257.
[29]SHEN Y Q,HUI L,JIANG H B,et al.Reliable Inlier Evaluation for Unsupervised Point Cloud Registration[C]//Procee-dings of the AAAI Conference on Artificial Intelligence.California:Association for the Advancement of Artificial Intelligence(AAAI),2022:2198-2206.
[30]HE K M,ZHANG X Y,REN S Q,et al.Deep Residual Learning for Image Recognition[C]//Proceedings of 2016 IEEE Confe-rence on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2016:770-778.
[31]BAI Z Y,TAO J Y.Super-Resolution Reconstruction of Brain MR Images Using Pseudo-3D Convolutional Network[J].Journal of Computer-Aided Design & Computer Graphics,2022,34(2):208-216.
[32]THOMAS H,QI C R,DESCHAUD J E,et al.KPConv:Flexible and Deformable Convolution for Point Clouds[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision(ICCV).New York:IEEE,2019:6420-6429.
[33]BOULCH A,PUY G,MARLET R.FKAConv:Feature-kernelalignment for point cloud convolution[C]//Proceedings of the Asian Conference on Computer Vision(ACCV) 2020.Cham:Springer,2020:381-399.
[34]ZENG A,SONG S R,NIESSNER M,et al.3DMatch:Learning Local Geometric Descriptors from RGB-D Reconstructions[C]//Proceedings of the 30th IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2017:199-208.
[35]DENG H,BIRDAL T,ILIC S.PPF-FoldNet:UnsupervisedLearning of Rotation Invariant 3D Local Descriptors[C]//Proceedings of the 15th European Conference on Computer Vision(ECCV).Berlin:Springer-Verlag,2018:620-638.
[36]GEIGER A,LENZ P,URTASUN R.Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite[C]//Proceedings of 2012 IEEE Conference on Computer Vision and Pattern Recognition.New York:IEEE,2012:3354-3361.
[37]WU Z R,SONG S R,KHOSLA A,et al.3D ShapeNets:A Deep Representation for Volumetric Shapes[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE,2015:1912-1920.
[38]ZHOU Q Y,PARK J,KOLTUN V.Open3D:A Modern Library for 3D Data Processing[J].arXiv:1801.09847,2018.
[39]CAO A Q,PUY G,BOULCH A,et al.PCAM:Product of Cross-Attention Matrices for Rigid Registration of Point Clouds[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision(ICCV).Los Alamitos:IEEE Computer Society,2021:13209-13218.
[40]RUSINKIEWICZ S.A symmetric objective function for ICP[J].ACM Transactions on Graphics,2019,38(4):1-7.
[41]HEZRONI I,DRORY A,GIRYES R,et al.DeepBBS:Deep Best Buddies for Point Cloud Registration[C]//Proceedings of 2021 International Conference on 3D Vision(3DV).Los Alamitos:IEEE Computer Society,2021:342-351.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发