Please wait a minute...
Shopping Cart Email List Chinese
Advanced Search Fig/Tab
Frontiers of Computer Science

ISSN 2095-2228

ISSN 2095-2236(Online)

CN 10-1014/TP

Postal Subscription Code 80-970

2018 Impact Factor: 1.129

Featured Articles  |  Most Downloaded  |  Most Reading
Online Submission Subscribe to Our RSS Content Feed
Front. Comput. Sci.    2024, Vol. 18 Issue (2) : 182703    https://doi.org/10.1007/s11704-023-2471-8
Image and Graphics
Decoupled deep hough voting for point cloud registration
Mingzhi YUAN1,2, Kexue FU1,2, Zhihao LI1,2, Manning WANG1,2()
1. Digital Medical Research Center, School of Basic Medical Sciences, Fudan University, Shanghai 200032, China
2. Shanghai Key Laboratory of Medical Image Computing and Computer Assisted Intervention, Shanghai 200032, China
 Download: PDF(5061 KB)   HTML
 Export: BibTeX | EndNote | Reference Manager | ProCite | RefWorks
Abstract

Estimating rigid transformation using noisy correspondences is critical to feature-based point cloud registration. Recently, a series of studies have attempted to combine traditional robust model fitting with deep learning. Among them, DHVR proposed a hough voting-based method, achieving new state-of-the-art performance. However, we find voting on rotation and translation simultaneously hinders achieving better performance. Therefore, we proposed a new hough voting-based method, which decouples rotation and translation space. Specifically, we first utilize hough voting and a neural network to estimate rotation. Then based on good initialization on rotation, we can easily obtain accurate rigid transformation. Extensive experiments on 3DMatch and 3DLoMatch datasets show that our method achieves comparable performances over the state-of-the-art methods. We further demonstrate the generalization of our method by experimenting on KITTI dataset.
Keywords point cloud registration      robust model fitting      deep learning      hough voting     
Corresponding Author(s): Manning WANG   
Just Accepted Date: 21 February 2023   Issue Date: 10 April 2023
 Cite this article:   
Mingzhi YUAN,Kexue FU,Zhihao LI, et al. Decoupled deep hough voting for point cloud registration[J]. Front. Comput. Sci., 2024, 18(2): 182703.
 URL:  
https://academic.hep.com.cn/fcs/EN/10.1007/s11704-023-2471-8
https://academic.hep.com.cn/fcs/EN/Y2024/V18/I2/182703
Fig.1  (a) The pipline of our method. It takes the putative correspondences as input, and output a rigid transformation for point cloud registration. The green lines and red lines in the input correspondences represent inliers and outliers, respectively. (b) The architecture of our sparse convolution network for refinement. The box denotes a sparse convolution followed by a Batchnorm and a ReLu
Fig.2  (a) The input correspondences in triplet are located far apart, making estimated transformation accurate. (b) The input correspondences in triplet are too close to estimate accurate transformation although the input correspondences in triplet are all inliers (colored in green). We mark the two close correspondences by red circles. (c) Ground truth transformation.
Fig.3  The distribution of the inlier ratio of the putative correspondences generated by FCGF [8] on 3DMatch (blue) [20] and 3DLoMatch (orange) [10]
Registration recall/% RRE/(° ) RTE/cm Time/s
FGR-origin [15] * 42.7 4.08 10.6 0.31
FGR [15] 87.25 2.31 7.18 0.96
RANSAC-1M [12] 90.94 2.66 8.25 1.43
RANSAC-2M [12] 91.31 2.56 8.07 1.99
RANSAC-4M [12] 91.74 2.52 7.81 3.97
Super4PCS [49] * 21.6 5.25 14.1 4.55
ICP (P2Point) [26] * 6.04 8.25 18.1 0.25
ICP (P2Plane) [26] * 6.59 6.61 15.2 0.27
DGR [17] * 85.2 2.58 7.73 0.70
TEASER [5] 87.80 2.45 7.44 0.26
DHVR-origin [18] * 91.4 2.08 6.61 0.46
DHVR [18] 91.99 2.09 6.73 0.83
Ours 92.61 2.17 7.11 0.31
Tab.1  Quantitative results on 3DMatch [20] with FCGF [8] descriptor. Results with * are from [18]
Registration recall/% RRE/(° ) RTE/cm Time/s
FGR [15] 36.05 3.97 12.28 0.92
RANSAC-1M [12] 46.27 4.96 14.33 1.50
RANSAC-2M [12] 48.34 4.66 14.01 2.56
RANSAC-4M [12] 50.03 4.53 13.55 3.75
TEASER [5] 43.29 4.33 11.23 0.12
DHVR-origin [18] * 56.1 ? ? ?
DHVR [18] 54.41 4.14 12.56 0.58
Ours 54.41 3.82 12.37 0.51
Tab.2  Quantitative results on 3DLoMatch [10] with FCGF [8] descriptor. Results with * are from [18]
Registration recall/% RRE/(° ) RTE/cm Time/s
RANSAC-1M [12] 62.32 4.53 11.69 1.12
RANSAC-2M [12] 63.44 4.41 11.44 2.21
RANSAC-4M [12] 64.96 4.27 11.41 4.39
TEASER [5] 60.64 4.06 10.24 0.10
DHVR-origin [18] * 64.6 ? ? ?
DHVR [18] 63.84 3.90 10.46 0.32
Ours 62.49 3.61 10.33 0.31
Tab.3  Quantitative results on 3DLoMatch [10] with Predator [10] descriptor. Results with * are from [18]
Registration recall/% RRE/(° ) RTE/cm Time/s
FGR [15] 97.84 0.20 10.39 11.79
RANSAC-1M [12] 97.84 0.20 10.37 1.42
RANSAC-2M [12] 98.20 0.20 10.34 2.27
RANSAC-4M [12] 98.37 0.20 10.32 3.77
TEASER [5] 98.92 0.35 14.29 0.21
DHVR [18] 96.76 0.18 8.89 0.62
Ours 98.74 0.18 9.99 0.47
Tab.4  Quantitative results on KITTI [21]
Registration recall/% RRE/(° ) RTE/cm Time/s
wo train 85.95 2.43 7.84 0.61
wo train + decoupled 90.57 2.28 7.38 0.29
with train 91.99 2.09 6.73 0.83
with train + decoupled 92.61 2.17 7.11 0.31
Tab.5  Ablation on decoupled strategy and learnable refinement. Note that “+ decoupled” represents using the decoupled strategy. “with train” represents using a neural network for refinement, while “wo train” represents without refinement
Registration recall/% RRE/(° ) RTE/cm Time/s
10k 89.65 2.69 8.55 0.14
50k 91.68 2.31 7.46 0.20
100k 92.61 2.17 7.11 0.31
200k 92.30 2.10 6.94 0.41
DHVR (200k) 91.99 2.09 6.73 0.83
Tab.6  Registration performance using different numbers of triplets
Fig.4  Visualization of registration using different number of triplets. The green lines and red lines represent the inlier correspondences and outlier correspondences, respectively. The black lines denote the estimated maximum consensus set under different numbers of triplets. The true inlier ratio of input correspondences is 2.75%, while the inlier ratio of the estimated maximum consensus sets under 100k and 200k triplets are 2.97% and 3.69%, respectively. Since the true inlier set is not the maximum consensus set, using more triplets causes a wrong estimation. (a) Input correspondences; (b) estimated maximum consensus set using 100k triplets; (c) estimated maximum consensus set using 200k triplets; (d) ground truth transformation; (e) estimated transformation using 100k triplets; (f) estimated transformation using 100k triplets
Rotation recall/% Registration recall/%
3DMatch [20] 95.44 92.61
3DLoMatch [10] 64.80 54.41
KITTI [21] 99.28 98.74
Tab.7  Recall of rotation and registration on different datasets
Fig.5  Visualization of registration results on 3DMatch (first row) and KITTI (second row)
  
  
  
  
1 X, Huang G, Mei J, Zhang R Abbas . A comprehensive survey on point cloud registration. 2021, arXiv preprint arXiv: 2103.02690
2 J, Yang K, Xian P, Wang Y Zhang . A performance evaluation of correspondence grouping methods for 3D rigid data matching. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43( 6): 1859–1874
3 K, Fu S, Liu X, Luo M Wang . Robust point cloud registration framework based on deep graph matching. In: Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, 8889–8898
4 G K L, Tam Z Q, Cheng Y K, Lai F C, Langbein Y, Liu D, Marshall R R, Martin X F, Sun P L Rosin . Registration of 3D point clouds and meshes: a survey from rigid to nonrigid. IEEE Transactions on Visualization and Computer Graphics, 2013, 19( 7): 1199–1217
5 H, Yang J, Shi L Carlone . TEASER: fast and certifiable point cloud registration. IEEE Transactions on Robotics, 2021, 37( 2): 314–333
6 J, Luo M, Yuan K, Fu M, Wang C Zhang . Deep graph matching based dense correspondence learning between non-rigid point clouds. IEEE Robotics and Automation Letters, 2022, 7( 3): 5842–5849
7 M, Yuan X, Huang K, Fu Z, Li M Wang . Boosting 3D point cloud registration by transferring multi-modality knowledge. 2023, arXiv preprint arXiv: 2302.05210
8 C, Choy J, Park V Koltun . Fully convolutional geometric features. In: Proceedings of 2019 IEEE/CVF International Conference on Computer Vision. 2019, 8957–8965
9 H, Wang Y, Liu Z, Dong W Wang . You only hypothesize once: point cloud registration with rotation-equivariant descriptors. In: Proceedings of the 30th ACM International Conference on Multimedia. 2022, 1630–1641
10 S, Huang Z, Gojcic M, Usvyatsov A, Wieser K Schindler . PREDATOR: registration of 3D point clouds with low overlap. In: Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021, 4265–4274
11 X, Bai Z, Luo L, Zhou H, Fu L, Quan C L Tai . D3Feat: joint learning of dense detection and description of 3D local features. In: Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020, 6358–6366
12 M A, Fischler R C Bolles . Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Communications of the ACM, 1981, 24( 6): 381–395
13 Leordeanu M, Hebert M. A spectral technique for correspondence problems using pairwise constraints. In: Proceedings of the 10th IEEE International Conference on Computer Vision. 2005, 1482–1489
14 D, Barath J Matas . Graph-cut RANSAC. In: Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018, 6733–6741
15 Q Y, Zhou J, Park V Koltun . Fast global registration. In: Proceedings of the 14th European Conference on Computer Vision. 2016, 766–782
16 K M, Yi E, Trulls Y, Ono V, Lepetit M, Salzmann P Fua . Learning to find good correspondences. In: Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018, 2666–2674
17 C, Choy W, Dong V Koltun . Deep global registration. In: Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020, 2511–2520
18 J, Lee S, Kim M, Cho J Park . Deep Hough voting for robust global registration. In: Proceedings of 2021 IEEE/CVF International Conference on Computer Vision. 2021, 15974–15983
19 C, Choy J, Gwak S Savarese . 4D spatio-temporal ConvNets: minkowski convolutional neural networks. In: Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019, 3070–3079
20 A, Zeng S, Song M, Nießner M, Fisher J, Xiao T Funkhouser . 3DMatch: learning local geometric descriptors from RGB-D reconstructions. In: Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. 2017, 199–208
21 A, Geiger P, Lenz R Urtasun . Are we ready for autonomous driving? The KITTI vision benchmark suite. In: Proceedings of 2012 IEEE Conference on Computer Vision and Pattern Recognition. 2012, 3354–3361
22 H, Wang D, Huang Y Wang . GridNet: efficiently learning deep hierarchical representation for 3D point cloud understanding. Frontiers of Computer Science, 2022, 16( 1): 161301
23 Y, Xian J, Xiao Y Wang . A fast registration algorithm of rock point cloud based on spherical projection and feature extraction. Frontiers of Computer Science, 2019, 13( 1): 170–182
24 J, Xiao S, Liu L, Hu Y Wang . Filtering method of rock points based on BP neural network and principal component analysis. Frontiers of Computer Science, 2018, 12( 6): 1149–1159
25 M, Yuan Z, Li Q, Jin X, Chen M Wang . PointCLM: a contrastive learning-based framework for multi-instance point cloud registration. In: Proceedings of the 17th European Conference on Computer Vision. 2022, 595–611
26 P J, Besl N D McKay . Method for registration of 3-D shapes. In: Proceedings of SPIE 1611, Sensor Fusion IV: Control Paradigms and Data Structures. 1992, 586–606
27 S, Granger X Pennec . Multi-scale EM-ICP: a fast and robust approach for surface registration. In: Proceedings of the 7th European Conference on Computer Vision. 2002, 418–432
28 D, Chetverikov D, Stepanov P Krsek . Robust Euclidean alignment of 3D point sets: the trimmed iterative closest point algorithm. Image and Vision Computing, 2005, 23( 3): 299–309
29 J, Yang H, Li D, Campbell Y Jia . Go-ICP: a globally optimal solution to 3D ICP point-set registration. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38( 11): 2241–2254
30 Y, Aoki H, Goforth R A, Srivatsan S Lucey . PointNetLK: robust & efficient point cloud registration using PointNet. In: Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019, 7156–7165
31 X, Huang G, Mei J Zhang . Feature-metric registration: a fast semi-supervised approach for robust point cloud registration without correspondences. In: Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020, 11363–11371
32 R B, Rusu N, Blodow M Beetz . Fast point feature histograms (FPFH) for 3D registration. In: Proceedings of 2009 IEEE International Conference on Robotics and Automation. 2009, 3212–3217
33 Y, Wang J M Solomon . Deep closest point: learning representations for point cloud registration. In: Proceedings of 2019 IEEE/CVF International Conference on Computer Vision. 2019, 3522–3531
34 Z J, Yew G H Lee . RPM-Net: robust point matching using learned features. In: Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020, 11821–11830
35 T, Papadopoulo M I A Lourakis . Estimating the Jacobian of the singular value decomposition: theory and applications. In: Proceedings of the 6th European Conference on Computer Vision. 2000, 554–570
36 Y, Li T Harada . Lepard: learning partial point cloud matching in rigid and deformable scenes. In: Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022, 5544–5554
37 Á P, Bustos T J Chin . Guaranteed outlier removal for rotation search. In: Proceedings of 2015 IEEE International Conference on Computer Vision. 2015, 2165–2173
38 Á, Parra T J, Chin F, Neumann T, Friedrich M Katzmann . A practical maximum clique algorithm for matching with pairwise constraints. 2019, arXiv preprint arXiv: 1902.01534
39 Á P, Bustos T J Chin . Guaranteed outlier removal for point cloud registration with correspondences. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40( 12): 2868–2882
40 J, Shi H, Yang L Carlone . ROBIN: a graph-theoretic approach to reject outliers in robust estimation using invariants. In: Proceedings of 2021 IEEE International Conference on Robotics and Automation. 2021, 13820–13827
41 H M, Le T T, Do T, Hoang N M Cheung . SDRSAC: semidefinite-based randomized approach for robust point cloud registration without correspondences. In: Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019, 124–133
42 J, Yang K, Xian Y, Xiao Z Cao . Performance evaluation of 3D correspondence grouping algorithms. In: Proceedings of 2017 International Conference on 3D Vision. 2017, 467–476
43 R Q, Charles H, Su K, Mo L J Guibas . PointNet: deep learning on point sets for 3D classification and segmentation. In: Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. 2017, 77–85
44 L, Sun L Deng . TriVoC: efficient voting-based consensus maximization for robust point cloud registration with extreme outlier ratios. IEEE Robotics and Automation Letters, 2022, 7( 2): 4654–4661
45 J C Gower . Generalized procrustes analysis. Psychometrika, 1975, 40( 1): 33–51
46 O, Ronneberger P, Fischer T Brox . U-Net: convolutional networks for biomedical image segmentation. In: Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. 2015, 234–241
47 A, Paszke S, Gross S, Chintala G, Chanan E, Yang Z, DeVito Z, Lin A, Desmaison L, Antiga A Lerer . Automatic differentiation in pytorch. In: Proceedings of the 31st Conference on Neural Information Processing Systems. 2017
48 Q Y, Zhou J, Park V Koltun . Open3D: a modern library for 3D data processing. 2018, arXiv preprint arXiv: 1801.09847
49 N, Mellado D, Aiger N J Mitra . Super 4PCS fast global pointcloud registration via smart indexing. Computer Graphics Forum, 2014, 33( 5): 205–215
50 J, Straub O, Freifeld G, Rosman J J, Leonard J W Fisher . The manhattan frame model—manhattan world inference in the space of surface normals. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40( 1): 235–249
[1] FCS-22471-OF-MY_suppl_1 Download
[1] Jingyu LIU, Shi CHEN, Li SHEN. A comprehensive survey on graph neural network accelerators[J]. Front. Comput. Sci., 2025, 19(2): 192104-.
[2] Lingling ZHAO, Shitao SONG, Pengyan WANG, Chunyu WANG, Junjie WANG, Maozu GUO. A MLP-Mixer and mixture of expert model for remaining useful life prediction of lithium-ion batteries[J]. Front. Comput. Sci., 2024, 18(5): 185329-.
[3] Enes DEDEOGLU, Himmet Toprak KESGIN, Mehmet Fatih AMASYALI. A robust optimization method for label noisy datasets based on adaptive threshold: Adaptive-k[J]. Front. Comput. Sci., 2024, 18(4): 184315-.
[4] Hengyu LIU, Tiancheng ZHANG, Fan LI, Minghe YU, Ge YU. A probabilistic generative model for tracking multi-knowledge concept mastery probability[J]. Front. Comput. Sci., 2024, 18(3): 183602-.
[5] Mingzhen LI, Changxi LIU, Jianjin LIAO, Xuegui ZHENG, Hailong YANG, Rujun SUN, Jun XU, Lin GAN, Guangwen YANG, Zhongzhi LUAN, Depei QIAN. Towards optimized tensor code generation for deep learning on sunway many-core processor[J]. Front. Comput. Sci., 2024, 18(2): 182101-.
[6] Hanadi AL-MEKHLAFI, Shiguang LIU. Single image super-resolution: a comprehensive review and recent insight[J]. Front. Comput. Sci., 2024, 18(1): 181702-.
[7] Yufei ZENG, Zhixin LI, Zhenbin CHEN, Huifang MA. Aspect-level sentiment analysis based on semantic heterogeneous graph convolutional network[J]. Front. Comput. Sci., 2023, 17(6): 176340-.
[8] Yamin HU, Hao JIANG, Zongyao HU. Measuring code maintainability with deep neural networks[J]. Front. Comput. Sci., 2023, 17(6): 176214-.
[9] Muazzam MAQSOOD, Sadaf YASMIN, Saira GILLANI, Maryam BUKHARI, Seungmin RHO, Sang-Soo YEO. An efficient deep learning-assisted person re-identification solution for intelligent video surveillance in smart cities[J]. Front. Comput. Sci., 2023, 17(4): 174329-.
[10] Tian WANG, Jiakun LI, Huai-Ning WU, Ce LI, Hichem SNOUSSI, Yang WU. ResLNet: deep residual LSTM network with longer input for action recognition[J]. Front. Comput. Sci., 2022, 16(6): 166334-.
[11] Yi WEI, Mei XUE, Xin LIU, Pengxiang XU. Data fusing and joint training for learning with noisy labels[J]. Front. Comput. Sci., 2022, 16(6): 166338-.
[12] Donghong HAN, Yanru KONG, Jiayi HAN, Guoren WANG. A survey of music emotion recognition[J]. Front. Comput. Sci., 2022, 16(6): 166335-.
[13] Pinzhuo TIAN, Yang GAO. Improving meta-learning model via meta-contrastive loss[J]. Front. Comput. Sci., 2022, 16(5): 165331-.
[14] Tian WANG, Shiye LEI, Youyou JIANG, Choi CHANG, Hichem SNOUSSI, Guangcun SHAN, Yao FU. Accelerating temporal action proposal generation via high performance computing[J]. Front. Comput. Sci., 2022, 16(4): 164317-.
[15] Kaimin WEI, Tianqi LI, Feiran HUANG, Jinpeng CHEN, Zefan HE. Cancer classification with data augmentation based on generative adversarial networks[J]. Front. Comput. Sci., 2022, 16(2): 162601-.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed