Research on Multi-Scale Point Cloud Completion Method Based on Local Neighborhood Dynamic Fusion
Source abstract
Abstract
Point cloud completion reconstructs incomplete, sparse inputs into complete 3D shapes. However, in the current 3D completion task, it is difficult to effectively extract the local details of an incomplete one, resulting in poor restoration of local details and low accuracy of the completed point clouds. To address this problem, this paper proposes a multi-scale point cloud completion method based on local neighborhood dynamic fusion (LNDF: adaptive aggregation of multi-scale local features through dynamic range and weight adjustment). Firstly, the farthest point sampling (FPS) strategy is applied to the original incomplete and defective point clouds for down-sampling to obtain three types of point clouds at different scales. When extracting features from point clouds of different scales, the local neighborhood aggregation of key points is dynamically adjusted, and the Transformer architecture is integrated to further enhance the correlation of local feature extraction information. Secondly, by combining the method of generating point clouds layer by layer in a pyramid-like manner, the local details of the point clouds are gradually enriched from coarse to fine to achieve point cloud completion. Finally, when designing the decoder, inspired by the concept of generative adversarial networks (GANs), an attention discriminator designed in series with a feature extraction layer and an attention layer is added to further optimize the completion performance of the network. Experimental results show that LNDM-Net reduces the average Chamfer Distance (CD) by 5.78% on PCN and 4.54% on ShapeNet compared to SOTA. The visualization of completion results demonstrates the superior performance of our method in both point cloud completion accuracy and local detail preservation. When handling diverse samples and incomplete point clouds in real-world 3D scenarios from the KITTI dataset, the approach exhibits enhanced generalization capability and completion fidelity.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: 3D Shape Modeling and Analysis · Remote Sensing and LiDAR Applications · 3D Surveying and Cultural Heritage
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
School of Smart Manufacturing, Jianghan University, Wuhan 430056, China
State Key Laboratory of Precision Blasting, Jianghan University, Wuhan 430056, China
Funding
Natural Science Foundation of Hubei Province
References (27)
Show source-provided references
1. Huang, X., Cheng, X., Geng, Q., Cao, B., Zhou, D., Wang, P., Lin, Y., and Yang, R. (2018, January 18–22). The apolloscape dataset for autonomous driving. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, Salt Lake City, UT, USA.10.1109/CVPRW.2018.00141 ↗
2. Deep learning for 3D point clouds: A survey10.1109/TPAMI.2020.3005434 ↗
3. Multi-scale Transformer based point cloud completion network10.11834/jig.210510 ↗
4. Pauly, M., Mitra, N.J., Wallner, J., Pottmann, H., and Guibas, L.J. (2008). Discovering structural regularity in 3d geometry. ACM SIGGRAPH 2008 Papers, Association for Computing Machinery.10.1145/1399504.1360642 ↗
5. Speciale, P., Oswald, M.R., Cohen, A., and Pollefeys, M. (2016, January 11–14). A symmetry prior for convex variational 3d reconstruction. Proceedings of the European Conference on Computer Vision, Amsterdam, The Netherlands.10.1007/978-3-319-46484-8_19 ↗
6. Li, Y., Wu, X., Chrysathou, Y., Sharf, A., Cohen-Or, D., and Mitra, N.J. (2011). Globfit: Consistently fitting primitives by discovering global relations. ACM SIGGRAPH 2011 Papers, Association for Computing Machinery.10.1145/1964921.1964947 ↗
7. Qi, C.R., Su, H., Kaichun, M., and Guibas, L.J. (2017, January 21–26). PointNet: Deep learning on point sets for 3D classification and segmentation. Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA.
8. Qi, C.R., Li, Y., Su, H., and Guibas, L.J. (2017, January 4–9). PointNet++: Deep hierarchical feature learning on point sets in a metric space. Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, CA, USA.
9. Achlioptas, P., Diamanti, O., Mitliagkas, I., and Guibas, L. (2018, January 10–15). Learning representations and generative models for 3d point clouds. Proceedings of the International Conference on Machine Learning, Stockholm, Sweden.
10. Pct: Point cloud transformer10.1007/s41095-021-0229-5 ↗
11. Dynamic graph CNN for learning on point clouds
12. Yuan, W.T., Khot, T., Held, D., Mertz, C., and Hebert, M. (2018, January 5–8). PCN: Point Completion Network. Proceedings of the 2018 International Conference on 3D Vision (3DV), Verona, Italy.10.1109/3DV.2018.00088 ↗
13. Yu, X., Rao, Y., Wang, Z., Liu, Z., Lu, J., and Zhou, J. (2021). PoinTr: Diverse Point Cloud Completion with Geometry-Aware Transformers, ICCV.10.1109/ICCV48922.2021.01227 ↗
14. Yu, X., Rao, Y., Wang, Z., Liu, Z., Lu, J., and Zhou, J. (2023). AdaPoinTr: Diverse Point Cloud Completion with Adaptive Geometry-Aware Transformers. IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), IEEE Computer Society.10.1109/TPAMI.2023.3309253 ↗
15. 3D Point Caps++: Learning 3D Representations with Capsule Networks10.1007/s11263-022-01632-6 ↗
16. Zhao, Y., Birdal, T., Deng, H., and Tombari, F. (2019, January 15–20). 3D Point Capsule Networks. Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA.10.1109/CVPR.2019.00110 ↗
17. Huang, Z., Yu, Y., Xu, J., Ni, F., and Le, X. (2020, January 13–19). PF-Net: Point Fractal Network for 3D Point Cloud Completion. Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA.10.1109/CVPR42600.2020.00768 ↗
18. VConv-DAE: Deep Volumetric Shape Learning Without Object Labels
19. Yang, Y., Feng, C., Shen, Y., and Tian, D. (2018, January 18–22). Foldingnet: Point cloud auto-encoder via deep grid deformation. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA.10.1109/CVPR.2018.00029 ↗
20. Tchapmi, L.P., Kosaraju, V., Rezatofighi, H., Reid, I., and Savarese, S. (2019, January 15–20). TopNet: Structural Point Cloud Decoder. Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA.10.1109/CVPR.2019.00047 ↗
21. Zhang, Q.-L., and Yang, Y.-B. (2021, January 6–11). SA-Net: Shuffle Attention for Deep Convolutional Neural Networks. Proceedings of the ICASSP 2021—2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Toronto, ON, Canada.10.1109/ICASSP39728.2021.9414568 ↗
22. SoftPoolNet: Shape Descriptor for Point Cloud Completion and Classification
23. Xiang, P., Wen, X., Liu, Y.S., Cao, Y.P., Wan, P., Zheng, W., and Han, Z. (2021, January 10–17). SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer. Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada.10.1109/ICCV48922.2021.00545 ↗
24. Guo, S., Fu, Y., Qian, Z., Rong, Z., and Wu, Y. (2022, January 18–22). Cross-Attention-Based Feature Extraction Network for 3D Point Cloud Registration. Proceedings of the 2022 IEEE International Conference on Multimedia and Expo (ICME), Taipei, Taiwan.10.1109/ICME52920.2022.9858925 ↗
25. Dosovitskiy, A. (2020). An image is worth 16x16 words: Transformers for image recognition at scale. arXiv.
26. Zhao, H., Jiang, L., Jia, J., Torr, P.H., and Koltun, V. (2021, January 11–17). Point transformer. Proceedings of the IEEE/CVF international Conference on Computer Vision, Montreal, BC, Canada.10.1109/ICCV48922.2021.01595 ↗
27. Xie, H., Yao, H., Zhou, S., Mao, J., Zhang, S., and Sun, W. (2020, January 23–28). GRNet: Gridding residual network for dense point cloud completion. Proceedings of the European Conference on Computer Vision, Glasgow, UK.10.1007/978-3-030-58545-7_21 ↗