Article(id=1251535834483208943, tenantId=1146029695717560320, journalId=1251233871195320423, issueId=1251535833375912679, articleNumber=null, orderNo=null, doi=10.13190/j.jbupt.2025-073, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1752422400000, receivedDateStr=2025-07-14, revisedDate=null, revisedDateStr=null, acceptedDate=null, acceptedDateStr=null, onlineDate=1776318995351, onlineDateStr=2026-04-16, pubDate=null, pubDateStr=null, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1776318995351, onlineIssueDateStr=2026-04-16, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1776318995351, creator=13701087609, updateTime=1776318995351, updator=13701087609, issue=Issue{id=1251535833375912679, tenantId=1146029695717560320, journalId=1251233871195320423, year='2025', volume='48', issue='5', pageStart='1', pageEnd='172', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=1, specialIssue=null, createTime=1776318995087, creator=13701087609, updateTime=1776389324200, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1251830815148163525, tenantId=1146029695717560320, journalId=1251233871195320423, issueId=1251535833375912679, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1251830815148163526, tenantId=1146029695717560320, journalId=1251233871195320423, issueId=1251535833375912679, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=17, endPage=24, ext={EN=ArticleExt(id=1251535834692924144, articleId=1251535834483208943, tenantId=1146029695717560320, journalId=1251233871195320423, language=EN, title=CAD Model Reconstruction Method from Point Clouds for IoT 3D Perception, columnId=1251535834252522218, journalTitle=Journal of Beijing University of Posts and Telecommunications, columnName=PAPERS, runingTitle=null, highlight=null, articleAbstract=

With the growing demand for understanding complex three dimensions (3D) environments in fields like Internet of things (IoT)-driven autonomous navigation and related applications, the automatic reconstruction of structured, editable computer-aided design (CAD) models from point cloud data has become a critical task. However, current research mainly focuses on reconstructing CAD models from CAD command sequences, sketches, and extrusion operations, and commonly faces challenges such as excessive reconstruction steps and strong platform dependence. To this end, a high-precision geometric and topological CAD reconstruction method is proposed, specifically for IoT 3D perception. First, the boundary representation of CAD is decomposed into parametric information of geometric primitives and their topological structure. Then, a primitive variational autoencoder (PVAE) and a topological variational autoencoder (TVAE) are designed to model their geometric features and topological relationships, respectively. Furthermore, PointNet++ is used to extract multi-scale local features from point cloud data and fuse them into global features. A topological decoder and primitive decoders are then used to predict topological tree sequences and primitive parameters, achieving high-precision CAD model reconstruction with clearer boundary details. To validate the method's effectiveness, metrics such as chamfer distance, edge chamfer distance, and structural consistency are used to evaluate its performance on two datasets. The experimental results show that the proposed model outperforms existing methods in terms of topological integrity and geometric reconstruction accuracy,enabling higher-precision CAD model reconstruction.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Shuai ZHAO, Zhen XIA, Junliang CHEN, Bo CHENG, Chenyang DU), CN=ArticleExt(id=1251535845375816051, articleId=1251535834483208943, tenantId=1146029695717560320, journalId=1251233871195320423, language=CN, title=面向物联网三维感知的点云到CAD模型重建方法, columnId=1251535834390934253, journalTitle=北京邮电大学学报, columnName=论文, runingTitle=null, highlight=null, articleAbstract=

随着物联网驱动的自主导航等领域对复杂三维环境理解需求的增长,从点云数据自动重建结构化、可编辑的计算机辅助设计(CAD)模型已成为关键任务。然而,当前研究主要从CAD命令序列、草图与拉伸操作等角度进行重建,普遍面临重建步骤多、平台依赖性强等问题。为此,提出一种高精度几何与拓扑关系的CAD重建方法,即面向物联网三维感知的点云到CAD模型重建方法。首先,将CAD的边界表示解析为几何基元的参数化信息与其拓扑结构,并设计基元变分自编码器(PVAE)与拓扑变分自编码器(TVAE)分别建模其几何特征与拓扑关系;在此基础上,利用PointNet++提取点云数据的多尺度局部特征并融合为全局特征,进而通过拓扑解码器与基元解码器分别预测拓扑树序列和基元参数,实现高精度与更清晰的边界细节的CAD模型重建。为验证方法的有效性,采用倒角距离、边缘倒角距离和结构一致性作为评价指标,在2个数据集上对方法性能进行评估。实验结果表明,所提方法在拓扑结构完整性和几何重建精度方面均优于现有方法,能够实现更高精度的CAD模型重建。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=HdMTV9qjygxsdfuKnPECZQ==, magXml=hXtnmus8grSFdqtuqQoTRg==, pdfUrl=null, pdf=YNJ/8HWawSK/XFkJ+l1POw==, pdfFileSize=1658925, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=qKyF3WKqvF+yfdErrQP22Q==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=QRuhPLtHArn3Rw5+ydXa/A==, mapNumber=null, authorCompany=null, fund=null, authors=

赵帅(1986—),男,教授,博士生导师,邮箱:

, authorsList=赵帅, 夏镇, 陈俊亮, 程渤, 杜晨阳)}, authors=[Author(id=1251535845707166100, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=zhaoshuaiby@bupt.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1251535845824606621, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535845707166100, language=EN, stringName=Shuai ZHAO, firstName=Shuai, middleName=null, lastName=ZHAO, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1251535846017544618, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535845707166100, language=CN, stringName=赵帅, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 网络与交换技术全国重点实验室,北京 100876, bio={"content":"

赵帅(1986—),男,教授,博士生导师,邮箱:

"}, bioImg=null, bioContent=

赵帅(1986—),男,教授,博士生导师,邮箱:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1251535845614891400, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, xref=null, ext=[AuthorCompanyExt(id=1251535845623280009, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China), AuthorCompanyExt(id=1251535845627474314, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 网络与交换技术全国重点实验室,北京 100876)])]), Author(id=1251535846101430705, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1251535846223065530, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846101430705, language=EN, stringName=Zhen XIA, firstName=Zhen, middleName=null, lastName=XIA, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1251535846298563008, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846101430705, language=CN, stringName=夏镇, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 网络与交换技术全国重点实验室,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1251535845614891400, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, xref=null, ext=[AuthorCompanyExt(id=1251535845623280009, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China), AuthorCompanyExt(id=1251535845627474314, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 网络与交换技术全国重点实验室,北京 100876)])]), Author(id=1251535846395032008, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1251535846529249749, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846395032008, language=EN, stringName=Junliang CHEN, firstName=Junliang, middleName=null, lastName=CHEN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1251535846629913053, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846395032008, language=CN, stringName=陈俊亮, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 网络与交换技术全国重点实验室,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1251535845614891400, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, xref=null, ext=[AuthorCompanyExt(id=1251535845623280009, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China), AuthorCompanyExt(id=1251535845627474314, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 网络与交换技术全国重点实验室,北京 100876)])]), Author(id=1251535846709604834, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1251535846814462443, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846709604834, language=EN, stringName=Bo CHENG, firstName=Bo, middleName=null, lastName=CHENG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1251535846910931444, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846709604834, language=CN, stringName=程渤, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 网络与交换技术全国重点实验室,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1251535845614891400, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, xref=null, ext=[AuthorCompanyExt(id=1251535845623280009, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China), AuthorCompanyExt(id=1251535845627474314, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 网络与交换技术全国重点实验室,北京 100876)])]), Author(id=1251535846994817533, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1251535847061926405, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846994817533, language=EN, stringName=Chenyang DU, firstName=Chenyang, middleName=null, lastName=DU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1251535847129035275, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, authorId=1251535846994817533, language=CN, stringName=杜晨阳, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=北京邮电大学 网络与交换技术全国重点实验室,北京 100876, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1251535845614891400, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, xref=null, ext=[AuthorCompanyExt(id=1251535845623280009, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China), AuthorCompanyExt(id=1251535845627474314, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 网络与交换技术全国重点实验室,北京 100876)])])], keywords=[Keyword(id=1251535847233892881, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, orderNo=1, keyword=Internet of things three dimensions perception), Keyword(id=1251535847347139097, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, orderNo=2, keyword=point cloud), Keyword(id=1251535847443608094, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, orderNo=3, keyword=computer-aided design), Keyword(id=1251535847531688487, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, orderNo=4, keyword=boundary representation), Keyword(id=1251535847611380268, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, orderNo=5, keyword=PointNet++), Keyword(id=1251535847720432177, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, orderNo=6, keyword=variational autoencoder), Keyword(id=1251535847816901175, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, orderNo=1, keyword=物联网三维感知), Keyword(id=1251535847921758781, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, orderNo=2, keyword=点云), Keyword(id=1251535847984673344, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, orderNo=3, keyword=计算机辅助设计), Keyword(id=1251535848081142343, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, orderNo=4, keyword=边界表示法), Keyword(id=1251535848160834125, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, orderNo=5, keyword=PointNet++), Keyword(id=1251535848282468949, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, orderNo=6, keyword=变分自编码器)], refs=[Reference(id=1251535850283152039, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2024, volume=73, issue=null, pageStart=1, pageEnd=8, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=KAISER M, BRUSA T, FIECHTER S, journalName=IEEE Transactions on Instrumentation and Measurement, refType=null, unstructuredReference=KAISER M, BRUSA T, FIECHTER S, et al. Method for parametric evaluation of 3D surface imaging systems for applications with moving objects[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 1-8., articleTitle=Method for parametric evaluation of 3D surface imaging systems for applications with moving objects, refAbstract=null), Reference(id=1251535850400592554, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=7167, pageEnd=7174, url=null, language=null, rfNumber=[2], rfOrder=1, authorNames=YANG Y, CUI Z, ZHANG Q, journalName=null, refType=null, unstructuredReference=YANG Y, CUI Z, ZHANG Q, et al. PS6D: Point cloud based symmetry-aware 6D object pose estimation in robot bin-picking[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2024: 7167-7174., articleTitle=PS6D: Point cloud based symmetry-aware 6D object pose estimation in robot bin-picking, refAbstract=null), Reference(id=1251535850476090030, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2025, volume=36, issue=7, pageStart=4963, pageEnd=4975, url=null, language=null, rfNumber=[3], rfOrder=2, authorNames=WANG F, WU J, YANG Z, journalName=Journal of Intelligent Manufacturing, refType=null, unstructuredReference=WANG F, WU J, YANG Z, et al. Industrial vision inspection using digital twins: Bridging CAD models and realistic scenarios[J]. Journal of Intelligent Manufacturing, 2025, 36(7): 4963-4975., articleTitle=Industrial vision inspection using digital twins: Bridging CAD models and realistic scenarios, refAbstract=null), Reference(id=1251535850564170417, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2025, volume=26, issue=8, pageStart=11249, pageEnd=11275, url=null, language=null, rfNumber=[4], rfOrder=3, authorNames=JIA S, GONG X, LIU F, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=JIA S, GONG X, LIU F, et al. AI-powered LiDAR point cloud understanding and processing: An updated survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(8): 11249-11275., articleTitle=AI-powered LiDAR point cloud understanding and processing: An updated survey, refAbstract=null), Reference(id=1251535850681610934, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=27154, pageEnd=27163, url=null, language=null, rfNumber=[5], rfOrder=4, authorNames=MA W, CHEN S, LOU Y, journalName=null, refType=null, unstructuredReference=MA W, CHEN S, LOU Y, et al. Draw step by step: Reconstructing CAD construction sequences from point clouds via multimodal diffusion[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 27154-27163., articleTitle=Draw step by step: Reconstructing CAD construction sequences from point clouds via multimodal diffusion, refAbstract=null), Reference(id=1251535850790662842, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=1766, pageEnd=1776, url=null, language=null, rfNumber=[6], rfOrder=5, authorNames=MA W, XU M, LI X, journalName=null, refType=null, unstructuredReference=MA W, XU M, LI X, et al. MultiCAD: Contrastive representation learning for multi-modal 3D computer-aided design models[C]//Proceedings of the ACM International Conference on Information and Knowledge Management, 2023: 1766-1776., articleTitle=MultiCAD: Contrastive representation learning for multi-modal 3D computer-aided design models, refAbstract=null), Reference(id=1251535850903909054, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=6772, pageEnd=6782, url=null, language=null, rfNumber=[7], rfOrder=6, authorNames=WU R, XIAO C, ZHENG C, journalName=null, refType=null, unstructuredReference=WU R, XIAO C, ZHENG C. DeepCAD: A deep generative network for computer-aided design models[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 6772-6782., articleTitle=DeepCAD: A deep generative network for computer-aided design models, refAbstract=null), Reference(id=1251535850987795138, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=16816, pageEnd=16826, url=null, language=null, rfNumber=[8], rfOrder=7, authorNames=LI P, GUO J, ZHANG X, journalName=null, refType=null, unstructuredReference=LI P, GUO J, ZHANG X, et al. SECAD-Net: Self-supervised CAD reconstruction by learning sketch-extrude operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023:16816-16826., articleTitle=SECAD-Net: Self-supervised CAD reconstruction by learning sketch-extrude operations, refAbstract=null), Reference(id=1251535851080069834, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=4671, pageEnd=4680, url=null, language=null, rfNumber=[9], rfOrder=8, authorNames=LI P, GUO J, LI H, journalName=null, refType=null, unstructuredReference=LI P, GUO J, LI H, et al. SfmCAD: Unsupervised CAD reconstruction by learning sketch-based feature modeling operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024:4671-4680., articleTitle=SfmCAD: Unsupervised CAD reconstruction by learning sketch-based feature modeling operations, refAbstract=null), Reference(id=1251535851155567311, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=4713, pageEnd=4722, url=null, language=null, rfNumber=[10], rfOrder=9, authorNames=KHAN M S, DUPONT E, ALI S A, journalName=null, refType=null, unstructuredReference=KHAN M S, DUPONT E, ALI S A, et al. CAD-SIG-Net: CAD language inference from point clouds using layer-wise sketch instance guided attention[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 4713-4722., articleTitle=CAD-SIG-Net: CAD language inference from point clouds using layer-wise sketch instance guided attention, refAbstract=null), Reference(id=1251535851277202132, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2022, volume=41, issue=4, pageStart=1, pageEnd=18, url=null, language=null, rfNumber=[11], rfOrder=10, authorNames=GUO H, LIU S, PAN H, journalName=ACM Transactions on Graphics, refType=null, unstructuredReference=GUO H, LIU S, PAN H, et al. ComplexGen: CAD reconstruction by B-rep chain complex generation[J]. ACM Transactions on Graphics, 2022, 41(4): 1-18., articleTitle=ComplexGen: CAD reconstruction by B-rep chain complex generation, refAbstract=null), Reference(id=1251535851386254039, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2025, volume=44, issue=4, pageStart=1, pageEnd=25, url=null, language=null, rfNumber=[12], rfOrder=11, authorNames=LIU Y, XU D, YU X, journalName=ACM Transactions on Graphics, refType=null, unstructuredReference=LIU Y, XU D, YU X, et al. HoLa: B-rep generation using a holistic latent representation[J]. ACM Transactions on Graphics, 2025, 44(4): 1-25., articleTitle=HoLa: B-rep generation using a holistic latent representation, refAbstract=null), Reference(id=1251535851486917338, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=1, pageEnd=10, url=null, language=null, rfNumber=[13], rfOrder=12, authorNames=QI C R, YI L, SU H, journalName=null, refType=null, unstructuredReference=QI C R, YI L, SU H, et al. PointNet + +: Deep hierarchical feature learning on point sets in a metric space[C]//Advances in Neural Information Processing Systems, 2017: 1-10., articleTitle=PointNet + +: Deep hierarchical feature learning on point sets in a metric space, refAbstract=null), Reference(id=1251535851570803424, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=21327, pageEnd=21336, url=null, language=null, rfNumber=[14], rfOrder=13, authorNames=JONES B T, HU M, KODNONGBUA M, journalName=null, refType=null, unstructuredReference=JONES B T, HU M, KODNONGBUA M, et al. Self-supervised representation learning forCAD[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 21327-21336., articleTitle=Self-supervised representation learning forCAD, refAbstract=null), Reference(id=1251535851650495205, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=9601, pageEnd=9611, url=null, language=null, rfNumber=[15], rfOrder=14, authorNames=KOCH S, MATVEEV A, JIANG Z, journalName=null, refType=null, unstructuredReference=KOCH S, MATVEEV A, JIANG Z, et al. ABC: A big CAD model dataset for geometric deep learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 9601-9611., articleTitle=ABC: A big CAD model dataset for geometric deep learning, refAbstract=null), Reference(id=1251535851751158504, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2021, volume=40, issue=4, pageStart=1, pageEnd=24, url=null, language=null, rfNumber=[16], rfOrder=15, authorNames=WILLIS K D D, PU Y, LUO J, journalName=ACM Transactions on Graphics, refType=null, unstructuredReference=WILLIS K D D, PU Y, LUO J, et al. Fusion 360 gallery: A dataset and environment for programmatic CAD construction from human design sequences[J]. ACM Transactions on Graphics, 2021, 40(4): 1-24., articleTitle=Fusion 360 gallery: A dataset and environment for programmatic CAD construction from human design sequences, refAbstract=null), Reference(id=1251535851839238894, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2020, volume=33, issue=null, pageStart=8776, pageEnd=8786, url=null, language=null, rfNumber=[17], rfOrder=16, authorNames=KANIA K, ZIEBA M, KAJDANOWICZ T, journalName=null, refType=null, unstructuredReference=KANIA K, ZIEBA M, KAJDANOWICZ T. UCSG-Net:Unsupervised discovering of constructive solid geometry tree[C]//Advances in Neural Information Processing Systems, 2020, 33: 8776-8786., articleTitle=UCSG-Net:Unsupervised discovering of constructive solid geometry tree, refAbstract=null), Reference(id=1251535851977650931, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=12478, pageEnd=12487, url=null, language=null, rfNumber=[18], rfOrder=17, authorNames=REN D, ZHENG J, CAI J, journalName=null, refType=null, unstructuredReference=REN D, ZHENG J, CAI J, et al. CSG-Stump: A learning friendly CSG-like representation for interpretable shape parsing[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021:12478-12487., articleTitle=CSG-Stump: A learning friendly CSG-like representation for interpretable shape parsing, refAbstract=null), Reference(id=1251535852132840184, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=482, pageEnd=498, url=null, language=null, rfNumber=[19], rfOrder=18, authorNames=REN D, ZHENG J, CAI J, journalName=null, refType=null, unstructuredReference=REN D, ZHENG J, CAI J, et al. ExtrudeNet: Unsupervised inverse sketch-and-extrude for shape parsing[C]//European Conference on Computer Vision. Springer, 2022: 482-498., articleTitle=ExtrudeNet: Unsupervised inverse sketch-and-extrude for shape parsing, refAbstract=null)], funds=null, companyList=[AuthorCompany(id=1251535845614891400, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, xref=null, ext=[AuthorCompanyExt(id=1251535845623280009, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China), AuthorCompanyExt(id=1251535845627474314, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, companyId=1251535845614891400, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=北京邮电大学 网络与交换技术全国重点实验室,北京 100876)])], figs=[ArticleFig(id=1251535848618013280, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=EAfjl/jlkL4d4/eWXjmMtA==, figureFileBig=qKyF3WKqvF+yfdErrQP22Q==, tableContent=null), ArticleFig(id=1251535848819339882, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=图1, caption=HGTR方法结构图, figureFileSmall=EAfjl/jlkL4d4/eWXjmMtA==, figureFileBig=qKyF3WKqvF+yfdErrQP22Q==, tableContent=null), ArticleFig(id=1251535849079386741, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=xF0zAXDfJmh7sDpHKA0y7g==, figureFileBig=dvnGCKkNXcroAiF/6mSskA==, tableContent=null), ArticleFig(id=1251535849184244344, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=图2, caption=几何基元(线)的VAE方法结构图, figureFileSmall=xF0zAXDfJmh7sDpHKA0y7g==, figureFileBig=dvnGCKkNXcroAiF/6mSskA==, tableContent=null), ArticleFig(id=1251535849301684861, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=TypbLSASZ0YYQYI5ycxEPg==, figureFileBig=CDZYlXAnVQ/k0LdEg4KpQQ==, tableContent=null), ArticleFig(id=1251535849406542465, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=图3, caption=在ABC和Fusion 360数据集上HGTR方法重建CAD可视化结果图, figureFileSmall=TypbLSASZ0YYQYI5ycxEPg==, figureFileBig=CDZYlXAnVQ/k0LdEg4KpQQ==, tableContent=null), ArticleFig(id=1251535849515594372, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
数据集方法CD↓ECD↓NC↑
ABCUCSG-Net1.23321.7860.866
CSG-Stump0.6717.7510.892
ExtrudeNet0.5197.1110.885
SECAD-Net0.5067.2860.884
SfmCAD0.3955.0380.919
HGTR0.3544.8670.934
Fusion 360UCSG-Net2.9505.2770.770
CSG-Stump2.7814.5900.744
ExtrudeNet2.2633.5580.819
SECAD-Net2.0523.2820.803
HGTR1.9652.8840.821
), ArticleFig(id=1251535849649812105, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=表1, caption=

2个数据集上不同方法实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
数据集方法CD↓ECD↓NC↑
ABCUCSG-Net1.23321.7860.866
CSG-Stump0.6717.7510.892
ExtrudeNet0.5197.1110.885
SECAD-Net0.5067.2860.884
SfmCAD0.3955.0380.919
HGTR0.3544.8670.934
Fusion 360UCSG-Net2.9505.2770.770
CSG-Stump2.7814.5900.744
ExtrudeNet2.2633.5580.819
SECAD-Net2.0523.2820.803
HGTR1.9652.8840.821
), ArticleFig(id=1251535849716920973, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
方法CD↓ECD↓NC↑
HGTR+PointNet0.3615.6120.864
HGTR+MLP0.4126.0690.822
HGTR0.3544.8670.934
), ArticleFig(id=1251535849817584276, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=表2, caption=

ABC数据集上的消融实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
方法CD↓ECD↓NC↑
HGTR+PointNet0.3615.6120.864
HGTR+MLP0.4126.0690.822
HGTR0.3544.8670.934
), ArticleFig(id=1251535849888887446, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
采样方式CD↓ECD↓NC↑
非均匀采样0.3754.7530.939
HGTR(均匀采样)0.3544.8670.934
), ArticleFig(id=1251535849976967833, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=表3, caption=

ABC数据集上不同采样方式的实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
采样方式CD↓ECD↓NC↑
非均匀采样0.3754.7530.939
HGTR(均匀采样)0.3544.8670.934
), ArticleFig(id=1251535850081825439, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
噪声比例/%CD↓ECD↓NC↑
HGTR0.3544.8670.934
10.3604.9610.929
30.3745.0990.919
50.3855.1540.913
80.4205.5060.907
100.4826.1130.896
), ArticleFig(id=1251535850157322915, tenantId=1146029695717560320, journalId=1251233871195320423, articleId=1251535834483208943, language=CN, label=表4, caption=

ABC数据集上不同噪声比例的实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
噪声比例/%CD↓ECD↓NC↑
HGTR0.3544.8670.934
10.3604.9610.929
30.3745.0990.919
50.3855.1540.913
80.4205.5060.907
100.4826.1130.896
)], attaches=null, journal=Journal(id=1251231493079478365, delFlag=0, nameCn=北京邮电大学学报, nameEn=Journal of Beijing University of Posts and Telecommunications, nameHistory1=null, nameHistory2=null, issn=1007-5321, eissn=, cn=11-3570/TN, coden=null, periodic=双月刊, language=CN, oaType=1, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=, officePhone=, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=3oVT0g3QLp1rkvAIIuTmqQ==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Beijing University of Posts and Telecommunications, journalRemark=null, publicationField=null, createdTime=1776246434710, updatedTime=1776251818275, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=J, firstLetterEn=J, subjectCode=Natural Sciences, subjectName=自然科学, subjectCodeEn=Natural Sciences, subjectNameEn=null, picCn=3oVT0g3QLp1rkvAIIuTmqQ==, picEn=cZgqTxrxv5qAJeszFMxaRQ==, jcr=null, cjcr=null, exts=[JournalExt(id=1251254073522078690, language=CN, name=北京邮电大学学报, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1776251818304, updatedTime=1776251818304, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://journal.bupt.edu.cn/journalx_yddxzrb/authorLogOn.action, submissionEditorUrl=https://journal.bupt.edu.cn/journalx_yddxzrb/editorLogOn.action, submissionReviewUrl=https://journal.bupt.edu.cn/journalx_yddxzrb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1251254073559827427, language=EN, name=Journal of Beijing University of Posts and Telecommunications, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1776251818313, updatedTime=1776251818313, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://journal.bupt.edu.cn/journalx_yddxzrb/authorLogOn.action, submissionEditorUrl=https://journal.bupt.edu.cn/journalx_yddxzrb/editorLogOn.action, submissionReviewUrl=https://journal.bupt.edu.cn/journalx_yddxzrb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1251233871195320423, websiteList=[Website(id=1251257283355820070, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1251233871195320423, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/bjyddxxb/CN, language=CN, createTime=1776252583588, createBy=18614031015, updateTime=1776252856823, updateBy=18614031015, name=北京邮电大学学报-中文, tplId=1146099689490845704, title=北京邮电大学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1251258558722031723, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=articleTextType, value=kx, createTime=1776252887659, updateTime=1776252887659, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558696865896, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=banner, value=null, createTime=1776252887653, updateTime=1776252887653, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558743003246, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=grayFlag, value=0, createTime=1776252887664, updateTime=1776252887664, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558684282983, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=logo, value=https://castjournals.cast.org.cn/joweb/bjyddxxb/CN/file/pic?fileId=TrlkYZ8NzFuGUlsjyFExDA==, createTime=1776252887650, updateTime=1776252887650, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558755586160, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=minRunFlag, value=0, createTime=1776252887667, updateTime=1776252887667, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558713643114, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/bjyddxxb/CN/file/pic, createTime=1776252887657, updateTime=1776252887657, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558747197551, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=silenceFlag, value=0, createTime=1776252887665, updateTime=1776252887665, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558705254505, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1776252887655, updateTime=1776252887655, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558726226028, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=themeColor, value=null, createTime=1776252887660, updateTime=1776252887660, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258558734614637, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283355820070, code=themeStyle, value=null, createTime=1776252887662, updateTime=1776252887662, creator=18614031015, updator=18614031015)]), Website(id=1251257283590701108, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1251233871195320423, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/bjyddxxb/EN, language=EN, createTime=1776252583644, createBy=18614031015, updateTime=1776252852864, updateBy=18614031015, name=北京邮电大学学报-英文, tplId=1146101810881728533, title=Journal of Beijing University of Posts and Telecommunications, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1251258589290119285, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=articleTextType, value=kx, createTime=1776252894947, updateTime=1776252894947, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589269147762, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=banner, value=null, createTime=1776252894942, updateTime=1776252894942, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589311090808, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=grayFlag, value=0, createTime=1776252894952, updateTime=1776252894952, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589260759153, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=logo, value=https://castjournals.cast.org.cn/joweb/bjyddxxb/EN/file/pic?fileId=TrlkYZ8NzFuGUlsjyFExDA==, createTime=1776252894940, updateTime=1776252894940, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589323673722, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=minRunFlag, value=0, createTime=1776252894955, updateTime=1776252894955, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589281730676, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/bjyddxxb/EN/file/pic, createTime=1776252894945, updateTime=1776252894945, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589315285113, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=silenceFlag, value=0, createTime=1776252894953, updateTime=1776252894953, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589277536371, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1776252894944, updateTime=1776252894944, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589294313590, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=themeColor, value=null, createTime=1776252894948, updateTime=1776252894948, creator=18614031015, updator=18614031015), WebsiteProps(id=1251258589302702199, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1251257283590701108, code=themeStyle, value=null, createTime=1776252894950, updateTime=1776252894950, creator=18614031015, updator=18614031015)])], journalTitle=北京邮电大学学报, weixinUrl=null, journalUrl=https://journal.bupt.edu.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Beijing University of Posts and Telecommunications, journalPhotoCn=3oVT0g3QLp1rkvAIIuTmqQ==, journalPhotoEn=cZgqTxrxv5qAJeszFMxaRQ==, journalFirstLetter=J, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/bjyddxxb/CN/10.13190/j.jbupt.2025-073, detailUrlEn=https://castjournals.cast.org.cn/joweb/bjyddxxb/EN/10.13190/j.jbupt.2025-073, pdfUrlCn=https://castjournals.cast.org.cn/joweb/bjyddxxb/CN/PDF/10.13190/j.jbupt.2025-073, pdfUrlEn=https://castjournals.cast.org.cn/joweb/bjyddxxb/EN/PDF/10.13190/j.jbupt.2025-073, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
面向物联网三维感知的点云到CAD模型重建方法
收藏切换
PDF下载
赵帅 , 夏镇 , 陈俊亮 , 程渤 , 杜晨阳
北京邮电大学学报 | 论文 2025,48(5): 17-24
收起
收藏切换
北京邮电大学学报 | 论文 2025, 48(5): 17-24
面向物联网三维感知的点云到CAD模型重建方法
全屏
赵帅 , 夏镇, 陈俊亮, 程渤, 杜晨阳
作者信息
  • 北京邮电大学 网络与交换技术全国重点实验室,北京 100876
  • 赵帅(1986—),男,教授,博士生导师,邮箱:

CAD Model Reconstruction Method from Point Clouds for IoT 3D Perception
Shuai ZHAO , Zhen XIA, Junliang CHEN, Bo CHENG, Chenyang DU
Affiliations
  • State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China
doi: 10.13190/j.jbupt.2025-073
文章导航
收藏切换

随着物联网驱动的自主导航等领域对复杂三维环境理解需求的增长,从点云数据自动重建结构化、可编辑的计算机辅助设计(CAD)模型已成为关键任务。然而,当前研究主要从CAD命令序列、草图与拉伸操作等角度进行重建,普遍面临重建步骤多、平台依赖性强等问题。为此,提出一种高精度几何与拓扑关系的CAD重建方法,即面向物联网三维感知的点云到CAD模型重建方法。首先,将CAD的边界表示解析为几何基元的参数化信息与其拓扑结构,并设计基元变分自编码器(PVAE)与拓扑变分自编码器(TVAE)分别建模其几何特征与拓扑关系;在此基础上,利用PointNet++提取点云数据的多尺度局部特征并融合为全局特征,进而通过拓扑解码器与基元解码器分别预测拓扑树序列和基元参数,实现高精度与更清晰的边界细节的CAD模型重建。为验证方法的有效性,采用倒角距离、边缘倒角距离和结构一致性作为评价指标,在2个数据集上对方法性能进行评估。实验结果表明,所提方法在拓扑结构完整性和几何重建精度方面均优于现有方法,能够实现更高精度的CAD模型重建。

物联网三维感知  /  点云  /  计算机辅助设计  /  边界表示法  /  PointNet++  /  变分自编码器

With the growing demand for understanding complex three dimensions (3D) environments in fields like Internet of things (IoT)-driven autonomous navigation and related applications, the automatic reconstruction of structured, editable computer-aided design (CAD) models from point cloud data has become a critical task. However, current research mainly focuses on reconstructing CAD models from CAD command sequences, sketches, and extrusion operations, and commonly faces challenges such as excessive reconstruction steps and strong platform dependence. To this end, a high-precision geometric and topological CAD reconstruction method is proposed, specifically for IoT 3D perception. First, the boundary representation of CAD is decomposed into parametric information of geometric primitives and their topological structure. Then, a primitive variational autoencoder (PVAE) and a topological variational autoencoder (TVAE) are designed to model their geometric features and topological relationships, respectively. Furthermore, PointNet++ is used to extract multi-scale local features from point cloud data and fuse them into global features. A topological decoder and primitive decoders are then used to predict topological tree sequences and primitive parameters, achieving high-precision CAD model reconstruction with clearer boundary details. To validate the method's effectiveness, metrics such as chamfer distance, edge chamfer distance, and structural consistency are used to evaluate its performance on two datasets. The experimental results show that the proposed model outperforms existing methods in terms of topological integrity and geometric reconstruction accuracy,enabling higher-precision CAD model reconstruction.

Internet of things three dimensions perception  /  point cloud  /  computer-aided design  /  boundary representation  /  PointNet++  /  variational autoencoder
赵帅, 夏镇, 陈俊亮, 程渤, 杜晨阳. 面向物联网三维感知的点云到CAD模型重建方法. 北京邮电大学学报, 2025 , 48 (5) : 17 -24 . DOI: 10.13190/j.jbupt.2025-073
Shuai ZHAO, Zhen XIA, Junliang CHEN, Bo CHENG, Chenyang DU. CAD Model Reconstruction Method from Point Clouds for IoT 3D Perception[J]. Journal of Beijing University of Posts and Telecommunications, 2025 , 48 (5) : 17 -24 . DOI: 10.13190/j.jbupt.2025-073
随着物联网(IoT,Internet of things)感知对复杂三维(3D,three dimensions)环境理解中的需求日益增加,将Photoneo、MechMind等设备[1-2]感知到的3D数据(点云)转换为结构化、可编辑的计算机辅助设计(CAD,computer-aided design)模型,已成为数字孪生[3]、自主导航[4]等物联网驱动应用的关键任务。在复杂的物联网应用中,包括仿真、规划和装配等下游任务都是在CAD模型的基础上进行的。因此,从点云数据到CAD模型的重建不仅能够为下游任务提供可验证的3D数据,还能够生成可编辑的CAD模型,为后续应用提供基础支持。
在CAD重建研究中,现有方法主要通过3类途径实现:重构CAD设计命令序列[5-7],草图与拉伸构成[8-10],或者几何参数化信息与拓扑关系[11-12]。对于CAD设计命令序列的重构,通常将其视为1个长序列,并对每条命令的类型及对应参数进行预测。例如,Ma等[5]提出一种基于多模态扩散的方法,从点云生成CAD构造序列,实现CAD重建。然而,由于同一几何体可能对应多种有效命令序列,其模型生成结果依赖数据中的规律。此外,对于复杂CAD模型,过长的命令序列会导致模型训练困难,从而难以满足创新性设计场景的需求。为克服这一局限,后续研究转向模拟实际建模过程的分步重构方法。例如,Li等[8]提出通过学习2D草图与拉伸操作,从原始几何重建可编辑的CAD模型。然而,这些方法存在显著的软件依赖性。由于不同CAD软件在草图设计和拉伸操作上存在建模差异,跨平台使用时需要适配相应的设计约束逻辑。随后,Liu等[12]将拓扑关系与几何参数化信息结合为整体编码,并支持多模态输入来生成CAD模型。然而,这些重建方法通常将几何参数化信息与拓扑结构解耦处理。在重建拓扑时,仍需独立建模不同几何基元之间的关联关系,这导致CAD重建过程依赖多步迭代,从而降低重建效率并可能影响模型一致性。目前,从点云到CAD的重建方法通常需要多步迭代,这可能导致CAD模型的误差随步骤数量增加而累积。此外,在CAD重建中,还存在几何基元参数精度低以及拓扑结构不完整的问题。
针对上述问题,提出了面向高精度几何与拓扑重构的点云到CAD高效重建模型(HGTR,efficient reconstruction model from point clouds to CAD for high-precision geometric and topological recovery)。该模型能够同时重建几何基元与拓扑结构,从而生成具有更高精度和更清晰边界的CAD模型。具体而言,为提高CAD重建精度,首先将CAD的边界表示(BRep,boundary representation)解析为不同基元(点、线、环、面)的参数化信息及其拓扑结构。然后,针对不同基元的参数化信息,设计基元变分自编码器(PVAE,primitive variational autoencoder)进行训练;同时,将拓扑结构构建为拓扑树序列,用于训练拓扑变分自编码器(TVAE,topology variational autoencoder);最后,使用PointNet++ [13]对点云数据进行编码,提取多尺度局部特征并融合为全局特征。该全局特征随后输入拓扑解码器与基元解码器,分别预测拓扑结构和基元参数信息,从而实现高精度CAD重建。主要贡献总结如下:
1)为了能重建出更高精度的几何基元,将CAD模型解析为几何基元的参数化信息及对应的拓扑结构,并设计不同的变分自编码器;
2)将拓扑结构构建为拓扑树序列,设计基于Transformer的变分自编码器以获取节点间的关系,从而实现更完整的拓扑结构重建;
3)实验结果表明,HGTR方法在大型CAD模型数据集(ABC,a big CAD model dataset)和Fusion 360数据集上能重建出完整的CAD拓扑结构,并重建出更高精度和边界细节更清晰的CAD模型。
HGTR方法由3个模块组成:基于PointNet++的点云编码模块、拓扑结构重建模块和几何参数化重建模块。该方法基于深度学习,将传感器感知到的稀疏且不规则点云数据高效重建为具有精确几何参数和拓扑结构的CAD模型,从而实现复杂物体的精确几何和拓扑重建,并满足物联网感知驱动的数字孪生等应用需求。该方法的整体结构如图1所示。
点云到CAD模型的高效重建问题旨在将传感器感知采集的三维点云数据转化为具备准确几何基元参数和完整拓扑结构的CAD模型。具体而言,给定1组由传感器感知的点云集P ={p1p2,…,pN},pi作为输入,方法需要重建出基元(点、线、环和面)参数化信息与拓扑结构,并重构出CAD的BRep信息[14]
HGTR方法通过几何基元参数化和拓扑结构重建相结合的方式能有效降低重建误差。CAD的BRep可解构为几何基元参数化信息和对应的拓扑结构,其中几何基元包括点、线、环和面,具体解构CAD得到的参数化信息如下。
1)点:V ={xyz}表示点的参数化信息由三维坐标组成。
2)线:E ={Coedeaexerrmbe}表示线的参数化信息,其中C为曲线类型,oe为线的原点坐标,de为线的方向向量,ae为线的轴向量,xe定义局部坐标系的X轴方向,通常与轴向量ae垂直,r为半径,rm为椭圆的短半轴半径,be为方向布尔值。
3)环:L ={beE1E2,…,EN}表示环的参数化信息,其中be为方向布尔值,{E1E2,…,EN}表示环中包含N条曲线。
4)面:F ={Sofnfafxfrrmαbf}表示面的参数化信息,其中S为面的类型,of为面的原点坐标,nf为面的法向量,af为轴向量的法向量,xf为局部坐标系的X轴的方向,r为主半径,rm为次半径,α为半角角度,bf为方向布尔值。
CAD的拓扑结构根据解构到的参数化信息可表示为T ={TF,FTF,LTL,ETE,V},其中TF,F为面和面的关联关系,TF,L为面和环的关联关系,TL,E为环和线的关联关系,TE,V为线和点的关联关系。
CAD几何基元参数化信息的重建对精度要求较高。为此,对BRep解构出几何基元(点、线、环和面)分别使用不同的基元变分自编码器PVAE拟合其数据集中不同基元的数据分布,从而为后续基于点云编码得到的全局特征解码提供支持,图2展示了以线为例的VAE方法结构。在CAD拓扑结构的重建过程中,将不同几何基元之间的关联关系构建为拓扑树序列,并采用拓扑变分自编码器对该序列进行建模与拟合。
基于Transformer的VAE被用于学习CAD几何基元(点、线、环和面)的高效潜在表示。该结构通过自注意力机制建模基元参数间的相关性与层次关系,相较传统多层感知机(MLP,multilayer perceptron)或循环神经网络具有更强的高维特征表达能力。结合VAE的潜在空间连续建模,可精确重建几何参数并生成合理的新基元,为CAD重建提供高质量表示。
1)基元编码器。给定线的参数化信息Ei,首先使用线性层映射其潜在特征ho,随后将该潜在特征输入至多层标准Transformer,利用自注意力机制捕捉线的几何信息在不同维度的相互依赖关系,最终输出对应的上下文特征hencdenc为
其中Φenc(·)为Transformer编码器。基于上下文特征,分别利用2个独立的线性层得到近似分布φϕzE|Ei)的均值μ与对数方差ln σ2
为了使训练过程中实现梯度对采样步骤的反向传播,不直接从近似分布中直接采样,而是采用重参数化技巧,从标准正态分布N(0,I)中采样1个随机噪声ε,并与均值μ和方差σ结合得到隐向量为
其中☉表示逐元素相乘,通过这种方式隐向量zE即包含来自编码器确定性信息(μσ)及随机性(ε),并保证整个过程可微,实现梯度反向传播。
2)基元解码器。为了精确重构线的参数化信息,将重采样的隐向量线zE输入至多层标准Transformer结构的解码器,将其映射回高维特征空间,以获得上下文特征向量hdec,最后使用线性层将其重构为与原始线的参数化信息相同维度的
为建模CAD模型中基元间复杂拓扑关系,将图形式拓扑结构序列化为可学习序列,用Transformer捕捉全局依赖并建模潜在拓扑约束。基于此构建基于Transformer的VAE结构,编码器提取拓扑特征,潜变量实现连续分布建模,获得结构一致性且可生成的拓扑表示,为CAD重建提供结构先验。
1)拓扑编码器。解析BRep信息得到CAD的拓扑结构,将其构建为1个拓扑树序列{},其中每个元素表示拓扑关系或间隔符。类型包括:TF,FTF,LTL,ETE,V及几何基元(VELF)。同时,引入特殊符号分隔符(SEP,separator token)和填充符(PAD,padding token),〈SEP〉用于区分关系类型,〈PAD〉用于填充。词表由填充符、拓扑关系和几何基元组成。
为了捕捉拓扑序列位置上的顺序信息,引入位置编码。对于序列中的第i个位置,嵌入维度索引k,其位置编码Φposik定义为
对输入序列进行嵌入映射,与位置编码进行逐元素相加得
随后通过多头自注意力机制建模拓扑依赖得
然后通过前馈神经网络对多头自注意力输出进行非线性变换,并结合残差连接与层归一化得到编码器输出,可表示为
其中ΦNorm(·)表示层归一化。为近似分布φϕzT|),通过2个独立的线性层分别处理得到其均值μ与对数方差ln σ2。随后采样其拓扑树序列的隐向量zT
2)拓扑解码器。为实现对拓扑树序列的有效重建,使用基于Transformer架构的解码器处理隐向量zT生成拓扑树序列表示。解码器由多层Transformer与序列映射层组成,首先将隐向量zT输入到Transformer中通过注意力机制建模拓扑元素间的依赖关系,然后通过线性层与softmax将表示映射至序列空间,并生成对应的拓扑序列指令概率分布。
对于实际场景物体,通过Photoneo或MechMind等传感器感知获得点云P ={p1p2,…,pN},pi,其中N表示采样点数。为弥补点云与CAD中BRep参数化和拓扑结构信息的差异,采用PointNet + +方法对点云编码,提取其局部细节和全局轮廓的多维度几何特征。
1)点集抽象模块。首先,使用最远点采样(FPS,farthest point sampling)从P筛选出N′个质心{c1c2,…,cN′},对每个质心ci通过半径r查询得到局部点集Gi。为保证平移不变性,对局部点归一化处理得
将归一化后的坐标及对应的附加特征(第(L-1)层集合抽象模块最后的输出)进行拼接,并通过线性层与最大池化层提取该层的局部特征为
其中:vi表示聚合到质心ci的局部特征,为最大池化操作,‖为拼接操作。
2)全局特征提取。每一层中包含多个点集抽象模块,每个模块将局部点集的点云聚合到质心,经过L层集合抽象处理后可得到NL个点及其高阶特征的集合。为了提取输入点云集的全局特征,首先将每个点通过线性层映射到统一的语义空间,然后使用最大池化层获取其全局特征为
其中:H为最终经PointNet + +编码点云后的全局特征表示,dm为重建CAD模块的输入维度。
为了从点云全局特征H重建CAD,使用2种解码器分别重建CAD的拓扑树序列和几何基元的参数化信息。对于拓扑树序列的重建,首先通过线性层将点云特征映射至拓扑树序列的语义空间为
其中zT的维度保持一致。随后将输入拓扑解码器,通过标准的Transformer层获得拓扑树序列潜在特征,并通过线性层与softmax还原至拓扑树序列相应维度,以得到与对应的拓扑序列概率分布为
对于几何基元参数的重建,首先利用不同线性层将点云全局特征映射至不同基元的语义空间,以获得相应的隐向量g∈{VELF},随后将这些隐向量输入到几何基元解码器中提取潜在特征,最后通过不同的线性层获得几何基元参数化信息为
其中表示重建后的几何基元参数化信息,与重建后的拓扑结构结合用于重构CAD。
训练过程主要分为2个阶段,第1个阶段分别训练拓扑结构与几何基元参数化的解码器,其损失由各自的VAE损失决定。以线为例,通过最大化数据的边际对数似然ln ψEi),即优化其证据下界(ELBO,evidence lower bound),以学习最优的编解码器参数(ϕθ)为
其中:E[…]为重构损失,其最小化的目标是使重构后的几何参数化信息尽可能接近原始输入;DKL[…]为KL散度,其目标是约束编码器输出的分布尽可能与先验分布ψz)相似。
第2阶段训练的目标是实现点云到CAD的重建,损失函数由拓扑结构重建损失和几何基元(点、线、环和面)参数重建损失组成。拓扑序列被离散化为“词表索引”,重建等价于分类任务,采用交叉熵(CE,cross entropy)损失,如式(15a)所示;其次,几何基元(点、边、环和面)参数的重建,其参数为连续值,因此采用均方误差(MSE,mean squared error)作为损失函数,如式(15b)所示。最终,整体损失由2部分组合而成,可表示为
其中:Ltotal为方法总损失函数,Ltopo为重建CAD拓扑结构的CE损失,其中表示预测的拓扑序列分布,V表示真实的拓扑序列。Lgg∈{VELF}为重建CAD几何参数化信息的MSE损失,其中yi分别表示预测和真实的几何参数。
1)数据集。为了验证所提方法在CAD重建任务中的有效性,在ABC数据集[15]与Fusion 360数据集[16]进行实验验证。ABC数据集包括6000个CAD模型,其中训练集数量为5000,测试集数量为1000。Fusion 360数据集包括3700个CAD模型,其中训练集数量为3000,测试集数量为700。针对2个数据集,移除面数量超过50个的CAD模型。
2)评价指标。通过3种通用的定量评价指标倒角距离(CD,chamfer distance)、边缘倒角距离(ECD,edge chamfer distance)、一致性(NC,normal consistency)评价所提方法在CAD模型重建任务的效果进行评估。
倒角距离用于衡量预测点集P和真实点集Q的对称距离,则CD定义为
边缘倒角距离用于评估预测边缘点集Pe和真实边缘点集Qe的结构特征的还原能力,ECD定义为
一致性衡量预测法向量与真实法向量间的一致性,NC定义为
3)超参数设置。实验基于PyTorch框架实现,并在搭载NVIDIA Tesla A800 80G显卡的高性能服务器上进行训练。TVAE的编解码器中Transformer层数均为6层,隐藏层大小为256;PVAE的编解码器中Transformer层数均为4层,隐藏层大小为128;训练过程中使用Adam优化器,初始学习率为1×10 -4,训练周期为100个Epoch,批量大小(batch size)设为16,输入点云大小为8192。
4)对比方法。为验证HGTR方法的重建效果,将在2个数据集上与构造实体几何树的无监督发现(UCSG,unsupervised discovering of constructive solid geometry tree)-Net[17],CSG-Stump[18],ExtrudeNet[19],素描-拉伸式计算机辅助设计(SECAD,sketch-extrude computer-aided design)-Net[8],SfmCAD[9]方法进行对比与分析,进一步评估其有效性和可行性。
不同重建方法在数据集ABC和Fusion 360上的实验结果如表1所示,主要从3个评价指标CD、ECD和NC上进行评价,这些指标综合衡量了CAD重建模型在几何精度和表面光滑方法的表现。
表1结果显示HGTR方法在2个数据集上的3个评价指标均取得最优表现。在ABC数据集上,HGTR的CD达到0.354,NC为0.934,相较于性能最好的基准方法(SfmCAD)分别提升10.38%和1.63%;在Fusion 360数据集上,HGTR的CD达到1.965,NC为0.821,相较于性能最好的基准方法(SECAD-Net)分别提升4.24%和2.24%;这些结果表明,HGTR方法能重建出几何精度更高、边界细节更清晰的CAD模型,同时NC指标最优表示该方法在法向一致性上表现最为出色,能重建出更光滑、连续的表面。相比之下,早期使用构造实体几何(CSG,constructive solid geometry)的UCSG-Net和CSG-Stump的CD和ECD上表现较差,尤其是在ABC数据集上,说明其在复杂结构的CAD模型的几何拟合与边界重建能力相对较弱;ExtrudeNet、SECAD-Net和SfmCAD方法相较于CSG方法性能有明显提升,主要原因在于,这类方法通常基于CAD的构建过程,通过草图绘制与拉伸操作实现CAD模型重建,在几何精度上都有一定程度提升。与上述方法相比,HGTR通过同时重建几何基元的参数化信息和拓扑结构,实现了更精细化的CAD重建,因此能够获得更优的性能,充分证明了HGTR在CAD重建任务中的有效性。
HGTR方法采用基于Transformer的变分自编码器作为几何参数化信息的编解码器,同时针对拓扑树序列使用基于Transformer的变分自编码器来获取拓扑关系。为了验证所提出的方法中不同模块的作用和有效性,进行了以下消融实验,包括使用PointNet替换PointNet + +编码点云定义为HGTR + PointNet;几何基元与拓扑结构的变分自编码器的方法核心结构Transformer替换为MLP定义为HGTR + MLP。将上述组合的方法在ABC数据集上进行实验,其实验结果如表2所示。实验结果显示,HGTR +PointNet相较于HGTR方法未提取点云数据中的局部特征和多尺度特征,使其难以直接获得点云中的细节特征,导致方法性能有所下降。HGTR +MLP将Transformer为基础结构的编码器换成MLP使其能以获得节点间的深层次交互关系,例如面与环的交互关系等,使方法重建CAD的几何精度和边界细节更差。因此,HGTR各个模块对于重建CAD的几何信息和拓扑关系都有不同程度的影响,验证了其算法的有效性。
为了验证点云采样方式对于算法性能的影响,进行不同采样方式实验,包括非均匀采样与均匀采样,其实验结果如表3所示。
实验结果显示,HGTR方法在均匀采样上的CD表现更优,表明其在整体几何轮廓的拟合上更加稳定,能保证全局结构的一致性。而非均匀采样在ECD和NC上表现更优,表明方法能重建出更好的边界特征。在数字孪生等关注全局一致性的物联网下游任务中,HGTR采用均匀采样更适合;在装配等关注局部特征的场景中,采用非均匀采样更适合。
为了评估HGTR方法在输入点云存在噪声情况下的鲁棒性,设计不同噪声比例的数据进行实验验证,具体而言,在原始点云中分别引入1%,3%,5%,8%和10%的高斯噪声,在ABC数据集上进行实验,其实验结果如表4所示。
实验结果表明,随着噪声比例增加,HGTR方法在CD和ECD指标均有所上升,而NC有所下降,表明噪声对算法的几何精度和法向一致性产生了一定影响。具体而言,HGTR方法在较低比例(1%~5%)的噪声下,性能下降幅度较小,表明HGTR有较强的鲁棒性和复杂噪声环境下的适应能力。
为了更加直观地展示HGTR方法在重建CAD任务中的表现,在ABC和Fusion 360数据集上进行了可视化分析,结果如图3所示,对比了点云、重建CAD模型和数据集中真实CAD模型。结果表明,所提出的HGTR方法能够从稀疏的点云输入中有效还原复杂拓扑结构关系和几何基元参数,所生成的CAD模型在拓扑结构和几何形状上与真实模型一致。上述结果表明,HGTR不仅具备出色的全局几何重建能力,还能够对局部细节实现准确捕捉,从而体现出较强的重建性能,为物联网复杂场景下的CAD重建提供有效支撑。
针对物联网驱动的数字孪生与自主导航等复杂三维环境理解需求,提出一种结合PointNet++与PVAE-TVAE的点云到CAD高精度重建方法。该方法通过协同建模几何基元参数化和拓扑关系,实现复杂物体的几何与拓扑重建,在几何精度和结构完整性上均显著提升。在2个公开CAD重建数据集上进行系统性评估,在倒角距离、边缘倒角距离和一致性3个评价指标上均优于现有方法,验证了所提方法的有效性。但当前方法在几何基元的覆盖范围上仍存在一定局限,尤其对非均匀有理B样条(NURBS,non-uniform rationalB-splines)曲面等复杂几何的支持不足,未来研究将致力于扩展方法对复杂几何基元的重建能力,同时在图编辑距离、流形性测试等更多指标上,对比更多直接生成BRep方法验证该方法的有效性。此外,还将针对更复杂遮挡场景和真实扫描数据开展实验,以提升方法在物联网感知和工程应用中的实用价值。
参考文献 引证文献
排序方式:
[1]
KAISER M, BRUSA T, FIECHTER S, et al. Method for parametric evaluation of 3D surface imaging systems for applications with moving objects[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 1-8.
[2]
YANG Y, CUI Z, ZHANG Q, et al. PS6D: Point cloud based symmetry-aware 6D object pose estimation in robot bin-picking[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2024: 7167-7174.
[3]
WANG F, WU J, YANG Z, et al. Industrial vision inspection using digital twins: Bridging CAD models and realistic scenarios[J]. Journal of Intelligent Manufacturing, 2025, 36(7): 4963-4975.
[4]
JIA S, GONG X, LIU F, et al. AI-powered LiDAR point cloud understanding and processing: An updated survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(8): 11249-11275.
[5]
MA W, CHEN S, LOU Y, et al. Draw step by step: Reconstructing CAD construction sequences from point clouds via multimodal diffusion[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 27154-27163.
[6]
MA W, XU M, LI X, et al. MultiCAD: Contrastive representation learning for multi-modal 3D computer-aided design models[C]//Proceedings of the ACM International Conference on Information and Knowledge Management, 2023: 1766-1776.
[7]
WU R, XIAO C, ZHENG C. DeepCAD: A deep generative network for computer-aided design models[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 6772-6782.
[8]
LI P, GUO J, ZHANG X, et al. SECAD-Net: Self-supervised CAD reconstruction by learning sketch-extrude operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023:16816-16826.
[9]
LI P, GUO J, LI H, et al. SfmCAD: Unsupervised CAD reconstruction by learning sketch-based feature modeling operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024:4671-4680.
[10]
KHAN M S, DUPONT E, ALI S A, et al. CAD-SIG-Net: CAD language inference from point clouds using layer-wise sketch instance guided attention[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 4713-4722.
[11]
GUO H, LIU S, PAN H, et al. ComplexGen: CAD reconstruction by B-rep chain complex generation[J]. ACM Transactions on Graphics, 2022, 41(4): 1-18.
[12]
LIU Y, XU D, YU X, et al. HoLa: B-rep generation using a holistic latent representation[J]. ACM Transactions on Graphics, 2025, 44(4): 1-25.
[13]
QI C R, YI L, SU H, et al. PointNet + +: Deep hierarchical feature learning on point sets in a metric space[C]//Advances in Neural Information Processing Systems, 2017: 1-10.
[14]
JONES B T, HU M, KODNONGBUA M, et al. Self-supervised representation learning forCAD[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 21327-21336.
[15]
KOCH S, MATVEEV A, JIANG Z, et al. ABC: A big CAD model dataset for geometric deep learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 9601-9611.
[16]
WILLIS K D D, PU Y, LUO J, et al. Fusion 360 gallery: A dataset and environment for programmatic CAD construction from human design sequences[J]. ACM Transactions on Graphics, 2021, 40(4): 1-24.
[17]
KANIA K, ZIEBA M, KAJDANOWICZ T. UCSG-Net:Unsupervised discovering of constructive solid geometry tree[C]//Advances in Neural Information Processing Systems, 2020, 33: 8776-8786.
[18]
REN D, ZHENG J, CAI J, et al. CSG-Stump: A learning friendly CSG-like representation for interpretable shape parsing[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021:12478-12487.
[19]
REN D, ZHENG J, CAI J, et al. ExtrudeNet: Unsupervised inverse sketch-and-extrude for shape parsing[C]//European Conference on Computer Vision. Springer, 2022: 482-498.
2025年第48卷第5期
PDF下载
88
41
引用本文
BibTeX
文章信息
doi: 10.13190/j.jbupt.2025-073
  • 接收时间:2025-07-14
  • 首发时间:2026-04-16
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-07-14
基金
作者信息
    北京邮电大学 网络与交换技术全国重点实验室,北京 100876
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/bjyddxxb/CN/10.13190/j.jbupt.2025-073
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏