|
|
基于Dental-Net的口内3D扫描点云补全
|
Abstract:
目的：为了解决口腔内3D扫描由于受口腔内复杂环境的影响(牙釉质、口水、唾液、金属牙、牙齿填充物)，导致重建的3D点云存在大量缺失孔洞的问题。方法：本文提出一种基于真实口内3D点云的特征信息，补全3D扫描过程中部分缺失点云的Dental-Net深度学习算法模型，实现了口腔牙齿3D精细化完整建模。Dental-Net整体为编解码器架构，编码器处的特征提取模块负责对不同分辨率下的牙齿点云进行特征提取，将提取到的特征输入到解码器，解码器依序输出补全的缺失点云，过程中结合使用了生成对抗网络思想。结果：使用评估指标倒角距离(Chamfer distance, CD)对最终补全结果进行评价，本文提出的方法在口内3D扫描点云数据集上CD = 0.292。结论：最终点云补全结果说明了本文模型具备精确的细节补全能力和不同缺失条件下的泛化能力。
Objective: To solve the problem of missing holes in 3D point cloud reconstruction due to the influence of complex oral environment (enamel, saliva, metal teeth, tooth fillings). Methods: This paper proposed a Dental Net depth learning algorithm model based on the feature information of the real 3D point cloud in the mouth to complete part of the missing point cloud in the 3D scanning process, and realized the 3D fine and complete modeling of oral teeth. Dental Net is an overall codec architecture. The feature extraction module on the encoder side is responsible for feature extraction of tooth point clouds at different resolutions, inputs the extracted features to the decoder, and the decoder outputs the missing point clouds to complete step by step. In the process of completion, the idea of generation countermeasure network is combined. Results: The evaluation index chamfer distance is used to evaluate the final completion result. By the method proposed in this paper CD = 0.292 on the 3D scanning point cloud data set in the mouth. The final point cloud completion results show that the model in this paper has accurate detail completion ability and generalization ability under different missing conditions.
| [1] | Nakhaei, M., Madani, A.S., Moraditalab, A. and Haghi, H.R. (2015) Three-Dimensional Accuracy of Different Impression Techniques for Dental Implants. Dental Research Journal, 12, 431-437.
https://doi.org/10.4103/1735-3327.166190 |
| [2] | Dereli, A.K., Zeybek, V., Sagtas, E., et al. (2018) Sex Determination with Morphological Characteristics of the Skull by Using 3D Modeling Techniques in Computerized Tomography. Forensic Science, Medicine, and Pathology, 14, 450-459. https://doi.org/10.1007/s12024-018-0029-0 |
| [3] | D’Ettorre, G., Farronato, M., Candida, E., Quinzi, V. and Grippaudo, C. (2022) A Comparison between Stereophotogrammetry and Smartphone Structured Light Technology for Three-Dimensional Face Scanning. The Angle Orthodontist, 92, 358-363. https://doi.org/10.2319/040921-290.1 |
| [4] | Claus, D., Radeke, J., Zint, M., et al. (2018) Generation of 3D Digital Models of the Dental Arches Using Optical Scanning Techniques. Seminars in Orthodontics, 24, 416-429. https://doi.org/10.1053/j.sodo.2018.10.006 |
| [5] | Guo, Y., Wang, H., Hu, Q., et al. (2020) Deep Learning for 3D Point Clouds: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 43, 4338-4364. https://doi.org/10.1109/TPAMI.2020.3005434 |
| [6] | Qi, C.R., Yi, L., Su, H. and Guibas, L.J. (2017) PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space. 31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, 4-9 December 2017. |
| [7] | Tchapmi, L.P., Kosaraju, V., Rezatofighi, H., Reid, I. and Savarese, S. (2019) TopNet: Structural Point Cloud Decoder. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, 15-20 June 2019, 383-392. https://doi.org/10.1109/CVPR.2019.00047 |
| [8] | 王鑫宇. 基于深度学习的三维点云物体识别方法研究[D]: [硕士学位论文]. 郑州: 郑州大学, 2021. |
| [9] | 罗开乾, 朱江平, 张建伟. 三维点云补全方法的现状和发展趋势[J]. 信息记录材料, 2020, 21(5): 179-180. |
| [10] | Yuan, W., Khot, T., Held, D., Mertz, C. and Hebert, M. (2018) PCN: Point Completion Network. 2018 International Conference on 3D Vision (3DV), Verona, 5-8 September 2018, 728-737. https://doi.org/10.1109/3DV.2018.00088 |
| [11] | Qi, C.R., Su, H., Mo, K. and Guibas, L.J. (2017) PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, 21-26 July 2017, 77-85. |
| [12] | 刘心溥, 马燕新, 许可, 等. 嵌入Transformer结构的多尺度点云补全[J]. 中国图像图形学报, 2022, 27(2): 538-549. |
| [13] | Sarmad, M., Lee, H.J. and Kim, Y.M. (2019) RL-GAN-Net: A Reinforcement Learning Agent Controlled GAN Network for Real-Time Point Cloud Shape Completion. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, 15-20 June 2019, 5891-5900. https://doi.org/10.1109/CVPR.2019.00605 |
| [14] | Liang, M.F. and Wang, X.J. (2019) Video Abnormal Behavior Detection Based on Generating Countermeasure Network. Proceedings of 2019 7th International Conference on Machinery, Materials and Computing Technology (ICMMCT 2019), Chongqing, 30 May 2019, 263-267. |
| [15] | Birdal, T. and Ilic, S. (2017) A Point Sampling Algorithm for 3D Matching of Irregular Geometries. 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vancouver, 24-28 September 2017, 6871-6878.
https://doi.org/10.1109/IROS.2017.8206609 |
| [16] | Du, Y., Du, T., You, J., et al. (2021) Study of Falling-down-Type DLP 3D Printing Technology for High-Resolution Hydroxyapatite Scaffolds. International Journal of Applied Ceramic Technology, 19, 268-280.
https://doi.org/10.1111/ijac.13915 |
| [17] | Li, P., Wang, R., Wang, Y. and Tao, W. (2020) Evaluation of the ICP Algorithm in 3D Point Cloud Registration. IEEE Access, 8, 68030-68048. https://doi.org/10.1109/ACCESS.2020.2986470 |
| [18] | Wang, W., You, Y., Liu, W. and Lu, C. (2021) Point Cloud Classification with Deep Normalized Reeb Graph Convolution. Image and Vision Computing, 106, Article ID: 104092. https://doi.org/10.1016/j.imavis.2020.104092 |
| [19] | Miao, Y., Zhang, L., Liu, J., Wang, J. and Liu, F. (2021) An End-to-End Shape-Preserving Point Completion Network. IEEE Computer Graphics and Applications, 41, 20-33. https://doi.org/10.1109/MCG.2021.3065533 |
| [20] | Liu, H. and Tian, X. (2022) An Adaptive Gradient Method with Energy and Momentum. Annals of Applied Mathematics, 38, 183-222. https://doi.org/10.4208/aam.OA-2021-0095 |
