基于动态图卷积的加权点云分类网络

doi:10.3778/j.issn.1002-8331.2103-0395

摘要/Abstract

摘要： 传统的直接处理点云的PointNet类深度学习网络大多只考虑了点云的全局特征而忽视了点云局部特征，动态图卷积网络DGCNN通过构建[k]近邻图完成了对局部特征的弥补。然而现有的DGCNN使用简单的边缘特征作为局部特征的输入，没有对局部特征进行更深入的研究，且仅使用最大池化处理点云无序性问题，这造成了一定的信息损失。提出加权点云分类网络WDGCNN，使用特征拼接思想优化网络结构以实现多层次特征的融合、通过对[k]近邻图构成的边缘特征设计恰当的加权函数以弱化远点的干扰，相对加强近点的特征、采用最大池化和平均池化相结合的对称函数弥补单独使用最大池化造成的全局信息损失的新方法，实现了模型优化。实验结果表明，在通用点云分类数据集ModelNet40上，WDGCNN相比于DGCNN分类准确率由91.61%达到了93.22%，验证了新方法的有效性。

关键词: 图像处理, 三维点云分类, 深度学习, 图卷积, [k]近邻, 池化

Abstract: Most of the traditional deep learning networks which directly deal with point clouds only consider the global features of point clouds and ignore the local features of point clouds. Dynamic graph convolution network（DGCNN） completes the local feature compensation by constructing [k]-nearest neighbor graph. However, the existing DGCNN uses simple edge features as the input of local features, and does not conduct further research on local features, In this paper, weighted dynamic graph convolution network（WDGCNN）, a weighted point cloud classification network is proposed, which uses the idea of feature mosaicking to optimize the network structure to achieve multi-level feature fusion, and designs an appropriate weighting function for the edge features of [k]-nearest neighbor graph to weaken the interference of remote points and relatively strengthen the features of near points. A new pooling method is adopted to compensate for the global information loss caused by the use of maximum pooling alone. Experimental results show that compared with DGCNN, the classification accuracy of WDGCNN increases from 91.61% to 93.22% on the general point cloud classification dataset ModelNet40, which verifies the effectiveness of the new method.

Key words: image processing, three-dimensional point cloud classification, deep learning, graph convolution, [k]-nearest neighbor, pooling

孙一珺, 胡辉. 基于动态图卷积的加权点云分类网络[J]. 计算机工程与应用, 2022, 58(20): 240-246.

SUN Yijun, HU Hui. Weighted Point Cloud Classification Network Based on Dynamic Graph Convolution[J]. Computer Engineering and Applications, 2022, 58(20): 240-246.

参考文献

[1] QI C R，LIU W，WU C，et al.Frustum pointnets for 3d object detection from RGB-D data[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Los Alamitos：IEEE Computer Society Press，2018：918-927.
[2] QI C R，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.Los Alamitos：IEEE Computer Society Press，2017：652-660.
[3] WANG Y，SUN Y，LIU Z，et al.Dynamic graph CNN for learning on point clouds[J].ACM Transactions on Graphics，2019，38（5）：1-12.
[4] GUO Y，WANG H，HU Q，et al.Deep learning for 3d point clouds：a survey[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2021，43（12）：4338-4364.
[5] GRIFFITHS D，BOEHM J.A review on deep learning techniques for 3D sensed data classification[J].Remote Sensing，2019，11（12）：1499.
[6] SU H，MAJI S，KALOGERAKIS E，et al.Multi-view convolutional neural networks for 3d shape recognition[C]//Proceedings of the IEEE International Conference on Computer Vision.Los Alamitos：IEEE Computer Society Press，2015：945-953.
[7] MA C，GUO Y，YANG J，et al.Learning multi-view representation with LSTM for 3-D shape recognition and retrieval[J].IEEE Transactions on Multimedia，2018，21（5）：1169-1182.
[8] 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.Los Alamitos：IEEE Computer Society Press，2015：922-928.
[9] 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.Los Alamitos：IEEE Computer Society Press，2017：5099-5108.
[10] YANG J，ZHANG Q，NI B，et al.Modeling point clouds with self-attention and gumbel subset sampling[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Los Alamitos：IEEE Computer Society Press，2019：3323-3332.
[11] DUAN Y，ZHENG Y，LU J，et al.Structural relational reasoning of point clouds[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Los Alamitos：IEEE Computer Society Press，2019：949-958.
[12] LI Y，BU R，SUN M，et al.Pointcnn：convolution on x-transformed points[C]//Advances in Neural Information Processing Systems.Los Alamitos：IEEE Computer Society Press，2018：820-830.
[13] THOMAS H，QI C R，DESCHAUD J E，et al.Kpconv：flexible and deformable convolution for point clouds[C]//Proceedings of the IEEE International Conference on Computer Vision.Los Alamitos：IEEE Computer Society Press，2019：6411-6420.
[14] 马京晖，潘巍，王茹.基于[K]-means聚类的三维点云分类[J].计算机工程与应用，2020，56（17）：181-186.
MA Jinghui，PAN Wei，WANG Ru.3D point cloud classification based on [K]-means clustering[J].Computer Engineering and Applications，2020，56（17）：181-186.
[15] LI G，MULLER M，THABET A，et al.DeepGCNs：can GCNs go as deep as CNNs?[C]//Proceedings of the IEEE International Conference on Computer Vision.Los Alamitos：IEEE Computer Society Press，2019：9267-9276.
[16] HE K，ZHANG X，REN S，et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Los Alamitos：IEEE Computer Society Press，2016：770-778.
[17] HUANG G，LIU Z，VAN DER MAATEN L，et al.Densely connected convolutional networks[C]//Proceedings of the IEEE Conference on Computer Vsion and Pattern Recognition.Los Alamitos：IEEE Computer Society Press，2017：4700-4708.
[18] YU F，KOLTUN V.Multi-scale context aggregation by dilated convolutions[C]//International Conference on Learning Representations（ICLR），2016.
[19] LIU Y，FAN B，XIANG S，et al.Relation-shape convolutional neural network for point cloud analysis[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Los Alamitos：IEEE Computer Society Press，2019：8895-8904.
[20] JOSEPH-RIVLIN M，ZVIRIN A，KIMMEL R.Momen（e）t：flavor the moments in learning to classify shapes[C]//Proceedings of the IEEE International Conference on Computer Vision Workshops，2019.
[21] LIANG P，CHEW C M，LEE G H.PointAtrousGraph：deep hierarchical encoder-decoder with point atrous convolution for unorganized 3D points[C]//2020 IEEE International Conference on Robotics and Automation（ICRA），2020.
[22] WOO S，PARK J，LEE J Y，et al.CBAM：convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision（ECCV），2018：3-19.
[23] SHILANE P，MIN P，KAZHDAN M，et al.The Princeton shape benchmark[C]//Proceedings of the Shape Modeling Applications.Los Alamitos：IEEE Computer Society Press，2004：167-178.
[24] 白静，徐浩钧.MSP-Net：多尺度点云分类网络[J].计算机辅助设计与图形学学报，2019，31（11）：1917-1924.
BAI Jing，XU Haojun.MSP-Net：multi-scale point cloud classification network[J].Journal of Computer-Aided Design & Computer Graphics，2019，31（11）：1917-1924.
[25] MAATEN L，HINTON G.Visualizing data using t-SNE[J].Journal of Machine Learning Research，2008，9：2579-2605.