面向大规模点云数据的一站式船体逆向建模轻量化方法

doi:10.11996/JG.j.2095-302X.2025061200

图学学报 ›› 2025, Vol. 46 ›› Issue (6): 1200-1208.DOI: 10.11996/JG.j.2095-302X.2025061200

• 制造产品核心工业软件 • 上一篇下一篇

面向大规模点云数据的一站式船体逆向建模轻量化方法

黄永裕¹(), 杜林¹(), 强以铭², 丁军²

¹ 宁波大学海运学院，浙江宁波 315000
² 中国船舶科学研究中心，江苏无锡 214082

收稿日期:2025-08-15 接受日期:2025-11-05 出版日期:2025-12-30 发布日期:2025-12-27
通讯作者:杜林(1988-)，男，副教授，博士。主要研究方向为船型智能设计等。E-mail：dulin1@nbu.edu.cn
第一作者:黄永裕(2000-)，男，硕士研究生。主要研究方向为船型智能设计。E-mail：nbu_hyy@163.com
基金资助:
船舶结构安全全国重点实验室开放基金(NAKLAS-2025KF004-K);国家自然科学基金青年科学基金(52201368);高等学校学科创新引智计划项目(D21013)

A lightweight framework for one-stop hull reverse modeling from large-amount point cloud data

HUANG Yongyu¹(), DU Lin¹(), QIANG Yiming², DING Jun²

¹ Faculty of Maritime and Transportation, Ningbo University, Ningbo Zhejiang 315000, China
² China Ship Scientific Research Center, Wuxi Jiangsu 214082, China

Received:2025-08-15 Accepted:2025-11-05 Published:2025-12-30 Online:2025-12-27
First author：HUANG Yongyu (2000-), master student. His main research interest covers ship hull design method. E-mail：nbu_hyy@163.com
Supported by:
National Key Laboratory of Ship Structural Safety(NAKLAS-2025KF004-K);National Natural Science Foundation of China Youth Science Fund(52201368);Project of the Program for Disciplines Innovation of Higher Education Institutions(D21013)

摘要/Abstract

摘要：

基于大规模点云的船体逆向建模技术在数字孪生、水动力性能分析方面存在较为实际的需求，目前针对船体逆向建模中大规模点云数据处理问题的特殊方法较少，常规方法多依赖部署训练成本较高的神经网络模型或国外商业软件。所以，开展船体轻量化逆向建模一站式方法研究，具有一定的工程应用价值：首先采用栅格化技术对点云数据进行压缩预处理，在保留船体宏观几何特征的前提下，将数据规模压缩90%以上，显著降低后续算法计算负荷；然后基于船体几何特征设计轮廓线检测与拟合算法，运用最小二乘法实现边界修正，确保完整性与光顺性；最后采用基于斜率异常检测的光顺算法，通过斜率异常侦测、初步修正和深度光顺3个步骤，高效地实现纵向和垂向的全船型线自动光顺处理。该方法经过2组不同船型的验证，与商业计算机辅助设计软件建立的参考模型相比主尺度误差(+0.24%~+2.68%)与排水体积偏差(-1.05%~-0.88%)仅存在微小偏移，均处于可接受范围内，且在常规计算机上5 min内即可完成，在船体逆向建模、船舶数字化以及三维生成式大模型的船型光顺等相关领域具有一定的应用潜力。

关键词: 三维点云数据, 船型光顺, 逆向建模, 栅格化降采样, 船型

Abstract:

There is a practical need for reverse modeling technology of ship hulls based on large-scale point clouds in applications such as digital twins and hydrodynamic performance analysis. Currently, few specialized methods addressed large-scale point cloud data for ship hull modeling, and conventional approaches usually relied on neural network models with high training costs or foreign commercial software. In this case, designing a lightweight, one-stop method for hull reverse modeling is valuable: initially, the ship hull was meshed to compress the point cloud data, reducing the data volume by over 90% while preserving the hull’s macro geometric features, thereby significantly decreasing the computational cost for the following algorithms; secondly, contour line detection and fitting algorithms were designed based on the hull’s geometric features, utilizing the least squares method for contour correction to ensure its completeness and smoothness; finally, a smoothing algorithm based on slope anomaly detection was adopted, which efficiently performed automatic smoothing of the hull’s longitudinal and vertical form lines through three steps: slope anomaly detection, preliminary correction, and deep smoothing. The method was validated on two ship models. Compared to the reference models established using commercial CAD software, the principal dimension errors (+0.24% to +2.68%) and displacement volume deviations (-1.05% to -0.88%) were only minor and all within an acceptable range. The entire process was completed within 5 minutes on a conventional computer. The method demonstrated potential for application in related fields such as hull reverse modeling, ship digitalization, and hull form smoothing for 3D generative large models.

Key words: 3D point cloud data, hull-form fairing, reverse modeling, mesh down-sampling, hull form

中图分类号:

U663.2
TP391

黄永裕, 杜林, 强以铭, 丁军. 面向大规模点云数据的一站式船体逆向建模轻量化方法[J]. 图学学报, 2025, 46(6): 1200-1208.

HUANG Yongyu, DU Lin, QIANG Yiming, DING Jun. A lightweight framework for one-stop hull reverse modeling from large-amount point cloud data[J]. Journal of Graphics, 2025, 46(6): 1200-1208.

图/表 17

图1 船舶点云建模常规方法与本文方法对比

Fig. 1 Comparison of regular methods and conducted research

图2 本文设计算法的简要流程图

Fig. 2 This article designs a brief flowchart of the algorithm

图3 测试船舶模型((a) 船模A；(b) 船模B)

Fig. 3 Testing ship models ((a) Ship model A; (b) Ship model B)

表1 测试船模参数

Table 1 Specification of the testing ship models

模型	L_WL/m	B_WL/m	T/m	V/m³	C_B
A	61.96	10.82	4.12	1964.39	0.7112
B	154.79	23.63	6.00	13457.36	0.6132

表2 测试模型的人工噪声特性

Table 2 Characteristics of artificial noise on testing models

模型	点云数量	坐标轴	μ	σ
A	115 737	X	0.1	2.0e-3×L_A
		Y	0.1	2.0e-2×B_A
		Z	0.1	1.5e-3×D_A
B	170 866	X	0.1	2.0e-3×L_B
		Y	0.1	2.0e-2×B_B
		Z	0.1	1.5e-3×D_B

图4 人工噪声处理效果可视化对比(左舷为原始三维点云，右舷为含人工噪声的点云) ((a) 船模A；(b) 船模B)

Fig. 4 Visual of models with artificial noise, the port is original 3D points, the starboard is the noisy point cloud data ((a) Ship model A; (b) Ship model B)

表3 压缩效果对比

Table 3 Comparison of compression performance

点云数量	原始点云	栅格化后点云	压缩率/%	时间/s	L_WL误差/%	B_WL误差/%
船模A	115 737	2 048 (32×64)	98.2	8.68	1.11	1.36
		8 192 (64×128)	92.9	9.95	1.23	1.84
		32 768 (128×256)	71.7	12.32	2.47	2.16
船模B	170 866	20 248 (32×64)	98.8	12.67	1.16	1.52
		8 192 (64×128)	95.2	13.91	1.38	1.91
		32 768 (128×256)	80.8	16.47	2.13	2.63

图5 2组样本的曲面插值结果对比(左舷为拟合曲面，右舷为含噪点云数据) ((a) 船模A；(b) 船模B)

Fig. 5 Surface interpolation results for two samples (the port is the noisy point cloud-data, the starboard is the fitted surface) ((a) Ship model A; (b) Ship model B)

图6 型值边界定义((a) 正常边界二维型值矩阵；(b) 含噪声的边界二维型值矩阵)

Fig. 6 Contour definition with offset values ((a) Offset matrix of normal contour; (b) Offset matrix of noisy contour)

图7 轮廓线修正与待光顺区域识别示意图((a) 船模A；(b) 船模B)

Fig. 7 Contour correction and unfair area identification ((a) Ship model A; (b) Ship model B)

图8 曲线的凹凸与平直

Fig. 8 Line odd and normal cases

图9 船模A水线修正前后对比示意图((a) 0号水线；(b) 31号水线；(c) 63号水线)

Fig. 9 Sample A waterline before and after correction ((a) No. 0 waterline; (b) No. 31 waterline; (c) No. 63 waterline)

图10 船模A站线修正前后对比图((a) 1号站线；(b) 32号站线；(c) 94号站线；(d) 126号站线)

Fig. 10 Sample A station line before and after correction ((a) The No. 1 station line ; (b) The No. 32 station line; (c) The No. 94 station line; (d) The No. 126 station line)

图11 样本修正前后灰度对比图((a) 船模A；(b) 船模B)

Fig. 11 Grayscale comparison chart before and after sample correction ((a) Ship model A; (b) Ship model B)

图12 FreeCAD传统建模流程中的关键步骤((a) 点云蒙皮后生成的初步曲面模型；(b) 经船体轮廓拟合与冗余曲面切除后的最终模型)

Fig. 12 Key steps in the traditional FreeCAD modeling workflow ((a) Rough surface generated after point cloud skinning; (b) Final surface after contour fitting and redundant trimming)

图13 船模曲面修正前后对比以及横剖面面积曲线比对((a) 船模A；(b) 船模B)

Fig. 13 Comparison of surface correction and sectional area curves of ship model ((a) Ship model A; (b) Ship model B)

表4 原始模型与最终模型参数对比

Table 4 Comparison between the original and modified models

参数	原始模型A	最终模型A	模型A的误差	原始模型B	最终模型B	模型B的误差
L_WL	61.96 m	62.66 m	1.13%	154.79 m	156.92 m	1.38%
B_W_L	10.82 m	11.11 m	2.68%	23.63 m	24.05 m	1.78%
Draft	4.12 m	4.13 m	0.24%	6.00 m	6.05 m	0.83%
Volume	1964.39 m³	1947.03 m³	-0.88%	13457.36 m³	13315.77 m³	-1.05%
C_B	0.711 2	0.677 2	-4.78%	0.613 2	0.583 2	-4.89%

参考文献 26

[1]	刘东生, 陈建林, 费点, 等. 基于深度相机的大场景三维重建[J]. 光学精密工程, 2020, 28(1): 234-243.
	LIU D S, CHEN J L, FEI D, et al. Three-dimensional reconstruction of large-scale scene based on depth camera[J]. Optics and Precision Engineering, 2020, 28(1): 234-243 (in Chinese). DOI URL
[2]	XIAO X W, WANG K, ZHONG Z C, et al. A novel data-driven based high-precision building roof contour full-automatic extraction and structured 3D reconstruction method combining stereo images and LiDAR points[J]. International Journal of Digital Earth, 2025, 18(1): 2484668. DOI URL
[3]	WEI Y, DING Z R, HUANG H C, et al. A non-contact measurement method of ship block using image-based 3D reconstruction technology[J]. Ocean Engineering, 2019, 178: 463-475. DOI URL
[4]	HONG S, KIM J. Three-dimensional visual mapping of underwater ship hull surface using image stitching geometry[J]. Ocean Engineering, 2023, 269: 113575. DOI URL
[5]	QIU S Y, WANG S Z, CHEN X Z, et al. Ship shape reconstruction for three-dimensional situational awareness of smart ships based on neural radiation field[J]. Engineering Applications of Artificial Intelligence, 2024, 136: 108858. DOI URL
[6]	王江安, 庞大为, 黄乐, 等. 基于多尺度特征递归卷积的稠密点云重建网络[J]. 图学学报, 2022, 43(5): 875-883.
	WANG J A, PANG D W, HUANG L, et al. Dense point cloud reconstruction network using multi-scale feature recursive convolution[J]. Journal of Graphics, 2022, 43(5): 875-883 (in Chinese).
[7]	王道累, 丁子健, 杨君, 等. 基于体素网格特征的NeRF大场景重建方法[J]. 图学学报, 2025, 46(3): 502-509.
	WANG D L, DING Z J, YANG J, et al. Large scene reconstruction method based on voxel grid feature of NeRF[J]. Journal of Graphics, 2025, 46(3): 502-509 (in Chinese). DOI
[8]	杨永俊, 杨仁友, 李俞锋, 等. 基于AUV搭载单目-线激光器的水下目标三维重建方法[EB/OL]. (2025-10-20) [2025-10-21]. https://link.cnki.net/urlid/42.1294.TH.20251020.1555.008.
	YANG Y J, YANG R Y, LI Y F, et al. Underwater target 3D reconstruction method based on AUV equipped with monocular line laser[EB/OL]. (2025-10-20) [2025-10-21]. https://link.cnki.net/urlid/42.1294.TH.20251020.1555.008. (in Chinese).
[9]	YU X M, TANG L L, RAO Y M, et al. Point-BERT: pre-training 3D point cloud transformers with masked point modeling[C]// 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2022: 19291-19300.
[10]	HUANG Z T, YU Y K, XU J W, et al. PF-Net: point fractal network for 3D point cloud completion[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2020: 7659-7667.
[11]	何星, 朱哲, 燕雪峰, 等. 融合图像信息的跨模态Transformer点云补全算法[J]. 计算机辅助设计与图形学学报, 2024, 36(7): 1026-1033.
	HE X, ZHU Z, YAN X F, et al. Cross-modal Transformer for point cloud completion[J]. Journal of Computer-Aided Design & Computer Graphics, 2024, 36(7): 1026-1033 (in Chinese).
[12]	LUO S T, HU W. Diffusion probabilistic models for 3D point cloud generation[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2021: 2836-2844.
[13]	张荣国, 毕键昌, 胡静, 等. 选择性状态空间双序列特征强化编解码三维点云形状补全方法[EB/OL]. (2023-03-21) [2025-05-09]. https://link.cnki.net/urlid/11.2925.tp.20250321.0946.002.
	ZHANG R G, BI J C, HU J, et al. Selective-state-space based dual-sequence feature-enhanced coding and decoding method for 3D point cloud shape completion[EB/OL]. (2023-03-21) [2025-05-09]. https://link.cnki.net/urlid/11.2925.tp.20250321.0946.002. (in Chinese).
[14]	ZHANG D, LU X, QIN H, et al. Pointfilter: point cloud filtering via encoder-decoder modeling[J]. IEEE Transactions on Visualization and Computer Graphics, 2021, 27(3): 2015-2027. DOI URL
[15]	徐晨, 倪蓉蓉, 赵耀. 融合稀疏点云补全的3D目标检测算法[J]. 图学学报, 2021, 42(1): 37-43.
	XU C, NI R R, ZHAO Y. 3D object detection algorithm combined with sparse point cloud completion[J]. Journal of Graphics, 2021, 42(1): 37-43 (in Chinese). DOI
[16]	冯肖维, 杜光皓, 赵一平, 等. 点云鲁棒低秩联合估计重构[J]. 计算机辅助设计与图形学学报, 2023, 35(9): 1344-1352.
	FENG X W, DU G H, ZHAO Y P, et al. Point cloud reconstruction based on robust low-rank collaborative estimation[J]. Journal of Computer-Aided Design & Computer Graphics, 2023, 35(9): 1344-1352 (in Chinese).
[17]	时志鹏, 冯肖维, 赵一平. 改进邻域漂移的多假设检验点云降噪[J]. 计算机工程, 2024, 50(6): 276-286. DOI
	SHI Z P, FENG X W, ZHAO Y P. Improved multi-hypothesis test point cloud noise reduction for neighborhood drift[J]. Computer Engineering, 2024, 50(6): 276-286 (in Chinese). DOI
[18]	CARR J C, BEATSON R K, CHERRIE J B, et al. Reconstruction and representation of 3D objects with radial basis functions[C]// The 28th Annual Conference on Computer Graphics and Interactive Techniques. New York: ACM, 2001: 67-76.
[19]	王文培, 张惊雷, 贾鑫. 基于边占用分割距离场的点云表面重建[J/OL]. (2025-02-14) [2025-05-09]. https://link.cnki.net/urlid/11.2925.TP.20250214.1532.026.
	WANG W P, ZHANG J L, JIA X. Surface reconstruction of point clouds based on edge occupancy segmentation distance field[EB/OL]. (2025-02-14) [2025-05-09]. https://link.cnki.net/urlid/11.2925.TP.20250214.1532.026. (in Chinese).
[20]	刘圣军, 陶珊珊, 王海波, 等. 基于平面路径的扫掠面高精度重建[J]. 图学学报, 2025, 46(2): 425-436. DOI
	LIU S J, TAO S S, WANG H B, et al. High-precision reconstruction of swept surfaces with a planar path[J]. Journal of Graphics, 2025, 46(2): 425-436 (in Chinese). DOI
[21]	DUAN W Y, ZHANG P X, HUANG L M, et al. Ship hull surface reconstruction from scattered points cloud using an RBF neural network mapping technology[J]. Computers & Structures, 2023, 281: 107012. DOI URL
[22]	LAFARGE F, ALLIEZ P. Surface reconstruction through point set structuring[J]. Computer Graphics Forum, 2013, 32(2pt2): 225-234.
[23]	BERGER M, TAGLIASACCHI A, SEVERSKY L M, et al. A survey of surface reconstruction from point clouds[J]. Computer Graphics Forum, 2017, 36(1): 301-329. DOI URL
[24]	杜林, 李广年, 郑彭军. 船体曲面特征的计算机视觉认知与生成机制研究[J]. 船舶力学, 2022, 26(7): 949-961.
	DU L, LI G N, ZHENG P J. Ship hull feature recognition and creation functionality investigation based on computer vision technology[J]. Journal of Ship Mechanics, 2022, 26(7): 949-961 (in Chinese).
[25]	舒跃辉, 杜林, 李广年. 基于卷积生成对抗网络的船型数据噪声处理方法研究[J]. 中国造船, 2023, 64(4): 164-175.
	SHU Y H, DU L, LI G N. Investigation of ship hull surface data processing based on generative adversarial networks[J]. Shipbuilding of China, 2023, 64(4): 164-175 (in Chinese).
[26]	黄永裕, 何灵伟, 杨沂铮, 等. 三维生成式大模型在船型概念设计中的可行性研究[J]. 中国舰船研究, 2025, 20(6): 1-14.
	HUANG Y Y, HE L W, YANG Y Z, et al. An investigation of AI-based 3D model generators for ship hull design at the conceptual design stage[J]. Chinese Journal of Ship Research, 2025, 20(6): 1-14 (in Chinese).

面向大规模点云数据的一站式船体逆向建模轻量化方法

A lightweight framework for one-stop hull reverse modeling from large-amount point cloud data

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 26

相关文章 3

编辑推荐

Metrics

本文评价

[1]	刘德丰, 陈伟政, 白亚强, 刘凯, 王琦. 基于条件生成模型的船型概念方案正向设计方法探索[J]. 图学学报, 2025, 46(6): 1209-1215.
[2]	黄祥, 王红星, 顾徐, 孟悦, 王浩羽. 一种新的基于特殊离群样本优化的三维点云特征选择算法[J]. 图学学报, 2022, 43(5): 884-891.
[3]	李胜男，林晓，陈言，马利庄. 基于点云的球面三维逆向建模[J]. 图学学报, 2013, 34(3): 49-52.