基于强化局部特征的3D点云分类与分割网络

doi:10.3969/j.issn.0255-8297.2022.02.014

摘要/Abstract

摘要： 在点云的处理过程中，许多深度学习网络未能充分考虑局部点之间的复杂关系，导致大量空间几何信息丢失。针对该问题，提出了一个强化局部特征的网络，用于点云的目标分类和语义分割。该网络通过设计编码单元对点的多方向信息进行编码；通过注意力机制学习采样分组后形成局部点云的特征。同时提出了一种新的多维损失函数，结合使用交叉熵损失函数与中心损失函数作用于分类任务。在数据集ModelNet40和ScanNet上进行了大量实验，结果表明：该网络在点云的目标分类和语义分割任务上表现出较好的性能。

关键词: 点云分类与分割, 编码单元, 注意力机制, 多维损失函数

Abstract: In the process of point cloud processing, many deep learning networks fail to fully consider the complicated relationships between local points, resulting in the loss of a large number of spatial geometric information. To solve this problem, an enriching local features network for point cloud classification and segmentation is proposed. The network designs an encoding unit to encode the multi-directional information of points, applies the attention mechanism to learn features of the local points formed after sampling and grouping, and proposes a new multi-dimensional loss function, which combines cross entropy loss function and the center loss function to act on the classification task. Extensive experiments are carried out on Modelnet40 and ScanNet datasets. The experimental results show that the network performs well in object classification and semantic segmentation tasks of 3D point cloud.

Key words: point cloud classification and segmentation, encoding unit, attention mechanism, multidimensional loss function

中图分类号:

TP391.4

陈丽芳, 魏梦如. 基于强化局部特征的3D点云分类与分割网络[J]. 应用科学学报, 2022, 40(2): 328-337.

CHEN Lifang, WEI Mengru. 3D Point Cloud Classification and Segmentation Network Based on Local Feature Enhancement[J]. Journal of Applied Sciences, 2022, 40(2): 328-337.

参考文献

[1] Li Y, Bu R, Sun M, et al. PointCNN:convolution on X-transformed points[J]. Advances in Neural Information Processing Systems, 2018, 31:820-830.
[2] 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, 2015:922-928.
[3] 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, 2015:945-953.
[4] 刘亚文,张颖.结合上下文特征和图割算法的车载点云聚类方法[J].应用科学学报, 2020, 38(6):924-935. Liu Y W, Zhang Y. Vehicle point cloud clustering based on contextual feature and graph cut[J]. Journal of Applied Sciences, 2020, 38(6):924-935.(in Chinese)
[5] Charles R Q, Hao S, 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, 2017:77-85.
[6] Qi C R, Yi L, Su H, et al. Pointnet++:deep hierarchical feature learning on point sets in a metric space[J]. Advances in Neural Information Processing Systems, 2017, 30:5099-5108.
[7] Jiang M Y, Wu Y R, Zhao T Q, et al. PointSIFT:a SIFT-like network module for 3D point cloud semantic segmentation[DB/OL]. 2018[2021-01-16]. https://arxiv.org/abs/1807.00652.
[8] Klokov R, Lempitsky V. Escape from cells:deep Kd-networks for the recognition of 3D point cloud models[C]//2017 IEEE International Conference on Computer Vision, 2017:863-872.
[9] Simonovsky M, Komodakis N. Dynamic edge-conditioned filters in convolutional neural networks on graphs[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition, 2017:29-38.
[10] 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.
[11] 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 Recognitio, 2019:9613-9622.
[12] 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, 2020:5588-5597.
[13] Hu Q Y, Yang B, Xie L H, et al. RandLA-net:efficient semantic segmentation of large-scale point clouds[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020:11105-11114.
[14] Liu W Y, Wen Y D, Yu Z D, et al. SphereFace:deep hypersphere embedding for face recognition[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition, 2017:6738-6746.
[15] Wang H, Wang Y T, Zhou Z, et al. CosFace:large margin cosine loss for deep face recognition[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018:5265-5274.
[16] Deng J K, Guo J, Xue N N, et al. ArcFace:additive angular margin loss for deep face recognition[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019:4685-4694.
[17] Wen Y D, Zhang K P, Li Z F, et al. A discriminative feature learning approach for deep face recognition[C]//Computer Vision, 2016:499-515.
[18] 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, 2015:1912-1920.
[19] Dai A, Chang A X, Savva M, et al. ScanNet:richly-annotated 3D reconstructions of indoor scenes[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition, 2017:2432-2443.
[20] Qi C R, Su H, Nießner M, et al. Volumetric and multi-view CNNs for object classification on 3D data[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016:5648-5656.