Partial Point Cloud Registration Based on Dilated Graph Convolution and Outlier Filtering

doi:10.3778/j.issn.1002-8331.2111-0428

Abstract

Abstract: Due to some unfavorable factors, such as irregular structure, noise and outliers in non-Euclidean space, how to accurately register partial point clouds is still a challenging task. For this task, an effective network is proposed for partial point cloud registration. In order to effectively extract the fine-grained features of local point cloud, a dilated graph convolution block with dense connection is devised by setting different expansion rates to increase the receptive field. The dense connection can effectively utilize the features and enhance the information transfer between features. Moreover, in the proposed network, an outlier filtering module based on multi-layer perceptron is explored to filter out mismatched point pairs by using context normalization. In this network, the transformation parameters for point cloud matching are derived via a singular value decomposition module. Experimental results on three widely used datasets, i.e. ModelNet40, ShapeNetCore and Real Data, demonstrate the effectiveness of the proposed network.

Key words: dilated graph convolution, dense connection, outlier filtering, point cloud registration

摘要： 由于点云在非欧几里德空间中，受到结构不规则、噪声、离群点等不利因素的影响，如何准确配准残缺点云，仍然是一个具有挑战性的任务。针对此任务，提出了一种有效的残缺点云配准网络。为了有效提取局部点云的细粒度特征，设计了一个密集膨胀图卷积模块，通过设置不同的膨胀率增大感受野，该模块中的密集连接形式，能够在有效利用特征的同时，加强特征间的信息传递。在所提出的网络结构中，基于多层感知器的离群点过滤模块，通过利用上下文标准化过滤掉不匹配的点对。在该网络中，匹配点云所需要的转换参数，利用奇异值分解模块获取。在三个广泛使用的数据集ModelNet40、ShapeNetCore与Real Data上的实验结果，验证了所提出网络的有效性。

关键词: 膨胀图卷积, 密集连接, 离群点过滤, 点云配准

SUN Zhanli, ZHANG Yuxin, CHEN Xia. Partial Point Cloud Registration Based on Dilated Graph Convolution and Outlier Filtering[J]. Computer Engineering and Applications, 2022, 58(22): 186-194.

孙战里, 张玉欣, 陈霞. 基于膨胀图卷积与离群点过滤的残缺点云配准[J]. 计算机工程与应用, 2022, 58(22): 186-194.

References

[1] 唐维军，徐琨，柳有权，等.自动驾驶汽车虚拟测试中的树木点云生成方法[J].计算机工程与应用，2021，57（11）：185-192.
TANG W J，XU K，LIU Y Q，et al.Tree point cloud generation method in virtual test of autonomous vehicle[J].Computer Engineering and Applications，2021，57（11）：185-192.
[2] 钱宝鑫，肖志勇，宋威.改进的卷积神经网络在肺部图像上的分割应用[J].计算机科学与探索，2020，14（8）：1358-1367.
QIAN B X，XIAO Z Y，SONG W.Application of improved convolutional neural network in lung image segmentation[J].Journal of Frontiers of Computer Science and Technology，2020，14（8）：1358-1367.
[3] 郝雯，王映辉，宁小娟，等.面向点云的三维物体识别方法综述[J].计算机科学，2017，44（9）：11-16.
HAO Q，WANG Y H，NING X J，et al.Overview of 3D object recognition methods oriented to point clouds[J].Computer Science，2017，44（9）：11-16.
[4] BESL P J，MCKAY N D.A method for registration of 3-D shapes[J].Transactions on Pattern Analysis and Machine Intelligence，1992，14（2）：239-256.
[5] H?HNEL D，BURGARD W.Probabilistic matching for 3D scan registration[C]//Proceedings of the VDI-Conference Robotik，2002.
[6] AGAMENNONI G，FONTANA S，SIEGWART R Y，et al.Point clouds registration with probabilistic data association[C]//2016 International Conference on Intelligent Robots and Systems，Daejeon，2016：4092-4098.
[7] YANG J，LI H，CAMPBELL D，et al.Go-ICP：a globally optimal solution to 3D ICP point-set registration[J].Transactions on Pattern Analysis and Machine Intelligence，2015，38（11）：2241-2254.
[8] ZHOU Q Y，PARK J，KOLTUN V.Fast global registration[C]//14th European Conference on Computer Vision.Cham：Springer，2016：766-782.
[9] BOUAZIZ S，TAGLIASACCHI A，PAULY M.Sparse iterative closest point[J].Computer Graphics Forum，2013，32（5）：113-123.
[10] 王庆闪，张军，刘元盛，等.基于NDT与ICP结合的点云配准算法[J].计算机工程与应用，2020，56（7）：88-95.
WANG Q S，ZHANG J，LIU Y S，et al.Point cloud registration algorithm based on combination of NDT and ICP[J].Computer Engineering and Applications，2020，56（7）：88-95.
[11] AOKI Y，GOFORTH H，SRIVATSAN R A，et al.PointNetLK：robust & efficient point cloud registration using PointNet[C]//2019 IEEE Conference on Computer Vision and Pattern Recognition，Long Beach，2019：7163-7172.
[12] QI C R，SU H，MO K，et al.PointNet：deep learning on point sets for 3D classification and segmentation[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition，Hawaii，2017：652-660.
[13] LUCAS B D，KANADE T.An iterative image registration technique with an application to stereo vision[C]//7th International Joint Conference on Artificial Intelligence，Vancouver，1981：121-130.
[14] WANG Y，SOLOMON J M.Deep closest point：learning representations for point cloud registration[C]//2019 IEEE/CVF International Conference on Computer Vision，Seoul，2019：3523-3532.
[15] WANG Y，SUN Y B，LIU Z W，et al.Dynamic graph CNN for learning?on?point?clouds[J].Transactions on Graphics，2019，38（5）：1-12.
[16] PAPADOPOULO T，LOURAKIS M I A.Estimating the Jacobian of the singular value decomposition：theory and applications[C]//16th European Conference on Computer Vision.Berlin：Springer，2000：554-570.
[17] WANG Y，SOLOMON J M.PRNet：self-supervised learning for partial-to-partial registration[C]//Conference on Neural Information Processing Systems，Vancouver，2019：8814-8826.
[18] JANG E，GU S，POOLE B.Categorical reparameterization with Gumbel-Softmax[J].arXiv：1611.01144，2016.
[19] LI J，ZHANG C，XU Z，et al.Iterative distance-aware simi-larity matrix convolution with mutual-supervised point elimination for efficient point cloud registration[C]//16th European Conference on Computer Vision.Cham：Springer，2020：378-394.
[20] HUANG X S，MEI G F，ZHANG J.Feature-metric registration：a fast semi-supervised approach for robust point cloud registration without correspondences[C]//2020 IEEE Conference on Computer Vision and Pattern Recognition，Hawaii，2020：11366-11374.
[21] WEI H，QIAO Z，LIU Z，et al.End-to-end 3D point cloud learning for registration task using virtual correspondences[C]//2020 IEEE/RSJ International Conference on Intelligent Robots and Systems，Hawaii，2020：2678-2683.
[22] LI G，MULLER M，THABET A，et al.DeepGCNs：can GCNs go as deep as CNNs?[C]//2019 IEEE/CVF International Conference on Computer Vision，Seoul，2019：9267-9276.
[23] HUANG G，LIU Z，VAN DER MAATEN L，et al.Densely connected?convolutional?networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition，Hawaii，2017：2261-2269.
[24] YI K M，TRULLS E，ONO Y，et al.Learning to find good correspondences[C]//2018 IEEE Conference on Computer Vision and Pattern Recognition，Salt Lake，2018：2666-2674.
[25] SUN W，JIANG W，TRULLS E，et al.ACNe：attentive context normalization for robust permutation-equivariant learning[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition，Seattle，2020：11286-11295.
[26] ULYANOV D，VEDALDI A，LEMPITSKY V.Instance normalization：the missing ingredient for fast stylization[J].arXiv：1607.08022，2016.
[27] IOFFE S，SZEGEDY C.Batch normalization：accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning.Lille：IMLS，2015：448-456.
[28] WU Z，SONG S，KHOSLA A，et al.3D ShapeNets：a deep representation for volumetric shapes[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition，Boston，2015：1912-1920.
[29] CHANG A X，FUNKHOUSER T，GUIBAS L，et al.Shape-
Net：an information-rich 3D model repository[J].arXiv：1512.03012，2015.
[30] TURK G，LEVOY M.Zippered polygon meshes from range images[C]//21st Annual Conference on Computer Graphics and Interactive Techniques，Orlando，1994：311-318.