Abstract
3D point clouds are essential for representing geometric structures in various fields such as autonomous driving and virtual reality. However, real-world data often suffers from incompleteness due to occlusions and noise, and existing completion methods typically rely on paired complete–incomplete training data or are limited to recovering relatively small missing regions, which restricts their effectiveness under high missing-rate scenarios. This paper introduces a GAN-based method for completing 3D point clouds, capable of reconstructing detailed structures from partial inputs. Our end-to-end framework, consisting of an encoder, generator, and discriminator, optimizes topological accuracy and spatial continuity through a multi-term joint loss. Experimental results on the ModelNet40 dataset demonstrate superior performance over traditional and deep learning-based methods, achieving Chamfer Distance (CD = 0.085), Earth Mover’s Distance (EMD = 0.199), and F-Score (0.208). The generated high-quality point clouds support downstream tasks like path planning and robotic grasping. The source code and experimental datasets used in this work are publicly available at: DOI: https://doi.org/10.5281/zenodo.18421141.
Data availability
The datasets analyzed during the current study are publicly available in the ModelNet40 dataset, provided by the Princeton ModelNet repository (https://modelnet.cs.princeton.edu). The source code and representative experimental data required to reproduce the results reported in this paper are permanently archived and publicly available via Zenodo at DOI: 10.5281/zenodo.18421141.
References
Wang, Q. & Kim, M. K. Applications of 3D point cloud data in the construction industry: a fifteen-year review from 2004 to 2018. Adv. Eng. Inform. 39, 306–319. https://doi.org/10.1016/j.aei.2019.02.007 (2019).
Chauhan, A. et al. A survey of deep reinforcement learning techniques for energy-efficient green cloud computing. Cluster Comput. 28, 989. https://doi.org/10.1007/s10586-025-05727-w (2025).
Yang, Z. et al. Mpv-pcqa: multimodal no-reference point cloud quality assessment via point cloud and captured dynamic video. Multimedia Syst. 31, 310. https://doi.org/10.1007/s00530-025-01887-2 (2025).
Xu, T. et al. A miniature A-mode ultrasound system for noninvasive bone surface point cloud acquisition. Ann. Biomed. Eng. https://doi.org/10.1007/s10439-025-03918-5 (2025).
Zhang, S., Hu, S., Zhao, X., Zhang, D. & Tao, B. An accurate 3D reconstruction method for large workpieces based on 3D vision. In: (eds Matsuno, T. et al.) Intelligent robotics and applications. ICIRA 2025. Lecture Notes in Computer Science, vol. 16076. Springer, Singapore https://doi.org/10.1007/978-981-95-2101-2_32. (2026).
Jung, Y. et al. A transfer function design for medical volume data using a knowledge database based on deep image and primitive intensity profile features retrieval. J. Comput. Sci. Technol. 39 (2), 320–335. https://doi.org/10.1007/s11390-024-3419-7 (2024).
Wang, J. et al. Non-rigid point cloud registration via anisotropic hybrid field harmonization. IEEE Trans. Pattern Anal. Mach. Intell. https://doi.org/10.1109/TPAMI.2025.3572584 (2025).
Zhou, Z., Luo, Y. & Sun, T. A quantitative 3D reconstruction evaluation method based on Blender. In: Proceedings of the 10th International Conference on Computer and Communications (ICCC), 761–765 (2024). https://doi.org/10.1109/ICCC62609.2024.10942306
Wang, L. et al. A cascaded graph convolutional network for point cloud completion. Vis. Comput. 41, 659–674. https://doi.org/10.1007/s00371-024-03354-x (2025).
Lu, C. H. & Chen, X. H. Improved iterative Poisson point cloud surface reconstruction. In: Proceedings of the 3rd International Conference on Digital Society and Intelligent Systems (DSInS), 382–385 (2023). https://doi.org/10.1109/DSInS60115.2023.10455616
Li, M., Li, G. & Li, X. PoseNorm-PCN: pose-normalized human point cloud completion from a single front view. Vis. Comput. 42, 52. https://doi.org/10.1007/s00371-025-04295-9 (2026).
Fu, Z. et al. AEDNet: adaptive embedding and multiview-aware disentanglement for point cloud completion. In: (eds Leonardis, A., Ricci, E., Roth, S., Russakovsky, O., Sattler, T. & Varol, G.) Computer Vision—ECCV 2024. ECCV 2024. Lecture Notes in Computer Science, vol. 15069. Springer, Cham https://doi.org/10.1007/978-3-031-73247-8_8. (2025).
Zhang, M. et al. Joint-learning: a robust segmentation method for 3D point clouds under label noise. Computer Animation and Virtual Worlds 36.3, e70038 (2025). https://doi.org/10.1002/cav.70038
Ji, J., Zhao, R. & Lei, M. Latent diffusion transformer for point cloud generation. Vis. Comput. 40, 3903–3917. https://doi.org/10.1007/s00371-024-03396-1 (2024).
Tychola, K. A., Vrochidou, E. & Papakostas, G. A. Deep learning based computer vision under the prism of 3D point clouds: a systematic review. Vis. Comput. 40, 8287–8329. https://doi.org/10.1007/s00371-023-03237-7 (2024).
Yao, G. et al. DS-GAN: a dual sub-structure GAN for thermal infrared image colorization using U-Net with ConvNeXt and multi-scale large kernel attention. Vis. Comput. 41, 12441–12459. https://doi.org/10.1007/s00371-025-04165-4 (2025).
Hu, X. et al. Msembgan: Multi-stitch embroidery synthesis via region-aware texture generation. IEEE Trans. Vis. Comput. Graph. 31.9, 5334–5347. https://doi.org/10.1109/TVCG.2024.3447351 (2024).
Chen, Z., Hu, Z., Dai, S. & Zhou, L. KANs vs MLPs in OT-GAN. In: Proceedings of the 5th International Symposium on Computer Engineering and Intelligent Communications (ISCEIC), pp. 356–360 (2024). https://doi.org/10.1109/ISCEIC63613.2024.10810254
Kazhdan, M., Bolitho, M. & Hoppe, H. Poisson surface reconstruction. In: Proceedings of the 4th Eurographics Symposium on Geometry Processing (SGP), pp. 61–70 (2006).
Leng, B. et al. Shape embedding and retrieval in multi-flow deformation. Comput. Vis. Media. 10, 439–451. https://doi.org/10.1007/s41095-022-0315-3 (2024).
Liang, G., Zhao, X., Zhao, J. & Zhou, F. MVCNN: a deep learning-based ocean–land waveform classification network for single-wavelength LiDAR bathymetry. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 16, 656–674. https://doi.org/10.1109/JSTARS.2022.3229062 (2023).
Toikkanen, M., Kwon, D. & Lee, M. ReSGAN: intracranial hemorrhage segmentation with residuals of synthetic brain CT scans. In: (eds de Bruijne, M. et al.) Medical image computing and computer assisted intervention—MICCAI 2021. MICCAI 2021. Lecture Notes in Computer Science, vol. 12901. Springer, Cham https://doi.org/10.1007/978-3-030-87193-2_38. (2021).
Dai, A., Qi, C. R. & Nießner, M. Shape completion using 3D-encoder-predictor CNNs and shape synthesis. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 6545–6554 (2017). https://doi.org/10.1109/CVPR.2017.693
Chung, Y. H. & Chen, Y. L. Three-dimensional image inpainting system using 3D-ED-GAN for efficient vision-based detection for rotor dynamic balance system. IEEE Access 10, 60025–60038. https://doi.org/10.1109/ACCESS.2022.3180339 (2022).
Xu, L. et al. SPDGrNet: A Lightweight and Efficient Image Classification Network for Zea mays Diseases. J. Crop Health. 77, 92. https://doi.org/10.1007/s10343-025-01154-4 (2025).
Li, J. et al. Edge-guided generative network with attention for point cloud completion. Vis. Comput. 41, 785–798. https://doi.org/10.1007/s00371-024-03364-9 (2025).
Tian, Z. et al. Enhanced 3D shoeprint classification via multi-scale PointNet + + with attention mechanisms. Vis. Comput. 42, 120. https://doi.org/10.1007/s00371-025-04337-2 (2026).
Zan, G., Wang, Y. & Gao, P. Improved DGCNN based on Transformer for point cloud segmentation. In: Lu, H., Cai, J. (eds) Artificial Intelligence and Robotics. ISAIR 2023. Communications in Computer and Information Science. Springer, Singapore (2024). (1998). https://doi.org/10.1007/978-981-99-9109-9_27
Chen, J. H. & Hsu, C. C. PointCNN-Hand: 3D hand joints estimate by PointCNN from hand point cloud. In: Proceedings of the 2021 International Conference on System Science and Engineering (ICSSE), 458–463 (2021). https://doi.org/10.1109/ICSSE52999.2021.9538459
Ma, X., Yin, Q., Zhang, X. & Tang, L. FoldingNet-based geometry compression of point cloud with multi descriptions. In: Proceedings of the 2022 IEEE International Conference on Multimedia and Expo Workshops (ICMEW), 1–6 (2022). https://doi.org/10.1109/ICMEW56448.2022.9859339
Gong, B., Nie, Y., Lin, Y., Han, X. & Yu, Y. ME-PCN: point completion conditioned on mask emptiness. In: Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision (ICCV), 12468–12477 (2021). https://doi.org/10.1109/ICCV48922.2021.01226
Wang, X. et al. TopNet: transformer-efficient occupancy prediction network for octree-structured point cloud geometry compression. In: Proceedings of the 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 27305–27314 (2025). https://doi.org/10.1109/CVPR52734.2025.02543
Huang, Z., Yu, Y., Xu, J., Ni, F. & Le, X. PF-Net: point fractal network for 3D point cloud completion. In: Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 7659–7667 (2020). https://doi.org/10.1109/CVPR42600.2020.00768
Li, J., Guo, S., Meng, X., Lai, Z. & Han, S. DPG-Net: densely progressive-growing network for point cloud completion. Neurocomputing 491, 1–13. https://doi.org/10.1016/j.neucom.2022.03.060 (2022).
Liu, M., Sheng, L., Yang, S., Shao, J. & Hu, S. M. Morphing and sampling network for dense point cloud completion. arXiv preprint arXiv:1912.00280 (2019).
Wang, Y., Tan, D. J., Navab, N. & Tombari, F. SoftPool++: an encoder–decoder network for point cloud completion. Int. J. Comput. Vis. 130, 1145–1164. https://doi.org/10.1007/s11263-022-01588-7 (2022).
Wen, X. et al. PMP-Net: point cloud completion by learning multi-step point moving paths. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 7443–7452 (2021). https://doi.org/10.1109/CVPR46437.2021.00736
Liu, Z. & Xue, R. Visual image encryption based on compressed sensing and Cycle-GAN. Vis. Comput. 40, 5857–5870. https://doi.org/10.1007/s00371-023-03140-1 (2024).
Shen, B. et al. Point cloud upsampling generative adversarial network based on residual multi-scale off-set attention. Virtual Real. Intell. Hardw. 5 (1), 81–91. https://doi.org/10.1016/j.vrih.2022.08.016 (2023).
Yadav, N. K., Singh, S. K. & Dubey, S. R. ISA-GAN: inception-based self-attentive encoder–decoder network for face synthesis using delineated facial images. Vis. Comput. 40, 8205–8225. https://doi.org/10.1007/s00371-023-03233-x (2024).
Chen, W. et al. Stacked deep fusion GAN for enhanced text-to-image generation. Vis. Comput. 41, 8947–8960. https://doi.org/10.1007/s00371-025-03908-7 (2025).
Rathnakumari, L. & Rao, G. R. K. Enhancing heart disease prediction through a CNN-GAN hybrid deep learning model. SN Comput. Sci. 7, 47 https://doi.org/10.1007/s42979-025-04601-1. (2026).
Liu, X. et al. Toward the unification of generative and discriminative visual foundation model: a survey. Vis. Comput. 41, 3371–3412. https://doi.org/10.1007/s00371-024-03608-8 (2025).
Brimos, P., Seregkos, P., Karamanou, A., Kalampokis, E. & Tarabanis, K. Deep learning missing value imputation on traffic data using self-attention and GAN-based methods. In: Proceedings of the 2024 Panhellenic Conference on Electronics & Telecommunications (PACET), 1–4 (2024). https://doi.org/10.1109/PACET60398.2024.10497055
Tian, Y. et al. DGL-GAN: discriminator-guided GAN compression. Vis. Comput. 41, 4639–4660. https://doi.org/10.1007/s00371-024-03682-y (2025).
Liang, H. & Wang, R. Research on multi-feature fusion shadow puppet motifs generation based on CSPMotifsGAN and cultural heritage preservation. Comput. Animat. Virtual Worlds 363, e70047 (2025).
Sarker, S. et al. A comprehensive overview of deep learning techniques for 3D point cloud classification and semantic segmentation. Mach. Vis. Appl. 35, 67. https://doi.org/10.1007/s00138-024-01543-1 (2024).
Shishegaran, A., Varaee, H., Rabczuk, T. & Shishegaran, G. High correlated variables creator machine: Prediction of the compressive strength of concrete. Comput. Struct. 247, 106479 (2021). https://arxiv.org/abs/2009.06421
Shishegaran, A. Computational methods for shape prediction for steel plate with stiffener subjected to explosive loads (Institut für Strukturmechanik, 2025). https://doi.org/10.25643/dbt.66715
Funding
This work was supported by the Major Science and Technology Projects (No. 318J009).
Author information
Authors and Affiliations
Contributions
D.Z. was responsible for the overall research framework and study design. S.M. designed the algorithms and performed data analysis and processing. J.S. contributed to the optimization of the model modules. H.H. conducted the experimental design and experimental data analysis. All authors reviewed and approved the final manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Zhao, D., Mao, S., Shao, J. et al. Generative adversarial networks for high-fidelity 3D point cloud completion. Sci Rep (2026). https://doi.org/10.1038/s41598-026-44111-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-026-44111-5
