广义无监督函数映射学习的三维形状密集对应方法

TP391.41; 提出了一种新颖的广义无监督函数映射学习的三维形状密集对应方法.首先,基于多层感知器(multilayer perceptron,MLP)和残差网络,直接学习深度点特征.其次,计算点云的近似测地线距离,并对其进行特征分解,建立特征嵌入空间,引入注意力机制,有效学习广义基函数表示.再次,结合点特征与广义基函数生成三维形状的深度特征表示.最后,建立无监督的函数映射网络框架,获取形状之间的密集对应表示.提出的三元正则优化机制,联合重构损失、特征损失和形状匹配的距离损失,在特征域和空间域上有效提升了学习性能及形状对应的精度.实验结果表明,广义基函数表示与无监督函数映射学习机制适用于任...

Full description

Saved in:
Bibliographic Details
Published in浙江大学学报（理学版） Vol. 50; no. 6; pp. 736 - 744
Main Authors 窦丰, 马会文, 谢昕洋, 杨万文, 石雪, 韩丽, 林彬
Format Journal Article
LanguageChinese
Published 辽宁师范大学 计算机与人工智能学院,辽宁 大连 116081 25.11.2023
Subjects
shape correspondence
functional maps
deep learning
无监督学习
深度学习
函数映射
形状对应
unsupervised learning
Online AccessGet full text
ISSN1008-9497
DOI10.3785/j.issn.1008-9497.2023.06.008

Cover

Abstract TP391.41; 提出了一种新颖的广义无监督函数映射学习的三维形状密集对应方法.首先,基于多层感知器(multilayer perceptron,MLP)和残差网络,直接学习深度点特征.其次,计算点云的近似测地线距离,并对其进行特征分解,建立特征嵌入空间,引入注意力机制,有效学习广义基函数表示.再次,结合点特征与广义基函数生成三维形状的深度特征表示.最后,建立无监督的函数映射网络框架,获取形状之间的密集对应表示.提出的三元正则优化机制,联合重构损失、特征损失和形状匹配的距离损失,在特征域和空间域上有效提升了学习性能及形状对应的精度.实验结果表明,广义基函数表示与无监督函数映射学习机制适用于任意三维形状,突破了现有方法只适用于连续二维流形的局限,在任意三维形状匹配中取得了更优的性能.
AbstractList TP391.41; 提出了一种新颖的广义无监督函数映射学习的三维形状密集对应方法.首先,基于多层感知器(multilayer perceptron,MLP)和残差网络,直接学习深度点特征.其次,计算点云的近似测地线距离,并对其进行特征分解,建立特征嵌入空间,引入注意力机制,有效学习广义基函数表示.再次,结合点特征与广义基函数生成三维形状的深度特征表示.最后,建立无监督的函数映射网络框架,获取形状之间的密集对应表示.提出的三元正则优化机制,联合重构损失、特征损失和形状匹配的距离损失,在特征域和空间域上有效提升了学习性能及形状对应的精度.实验结果表明,广义基函数表示与无监督函数映射学习机制适用于任意三维形状,突破了现有方法只适用于连续二维流形的局限,在任意三维形状匹配中取得了更优的性能.
Author 谢昕洋
马会文
窦丰
杨万文
韩丽
石雪
林彬
AuthorAffiliation 辽宁师范大学 计算机与人工智能学院,辽宁 大连 116081
AuthorAffiliation_xml – name: 辽宁师范大学 计算机与人工智能学院,辽宁 大连 116081
Author_FL DOU Feng
YANG Wanwen
MA Huiwen
SHI Xue
XIE Xinyang
HAN Li
LIN Bin
Author_FL_xml – sequence: 1
  fullname: DOU Feng
– sequence: 2
  fullname: MA Huiwen
– sequence: 3
  fullname: XIE Xinyang
– sequence: 4
  fullname: YANG Wanwen
– sequence: 5
  fullname: SHI Xue
– sequence: 6
  fullname: HAN Li
– sequence: 7
  fullname: LIN Bin
Author_xml – sequence: 1
  fullname: 窦丰
– sequence: 2
  fullname: 马会文
– sequence: 3
  fullname: 谢昕洋
– sequence: 4
  fullname: 杨万文
– sequence: 5
  fullname: 石雪
– sequence: 6
  fullname: 韩丽
– sequence: 7
  fullname: 林彬
BookMark eNo9j8tKw0AYhWdRwVr7FoKrjP8_08llqcUbFNzouiRmRhpkCgaxuC6VuhOsikUICu6kBUWMUnyZTIJvYURxdeBbnHO-BVLRXS0JWUKg3HHFSkQ7cawpAriW1_AcyoBxCjYtQYVU__k8qcdxJwBAEMg8t0rWTPqZpcP8OinGF8Xdgzmb5aNpfpOYad88PWZpUtz2s7dh8fFiZvfF-auZDL7GAzNJzftlfpXmz6NFMqf8w1jW_7JG9jbWd5tbVmtnc7u52rJiBOZaCEHoOxIdR4YMuRABKiyvMJvxBpNMKJcDqH1PoUAlA-SISklpi0Ao7jFeI8u_vSe-Vr4-aEfd4yNdLrZPo7DXC36cwS5V-Tcxl2g5
ClassificationCodes TP391.41
ContentType Journal Article
Copyright Copyright © Wanfang Data Co. Ltd. All Rights Reserved.
Copyright_xml – notice: Copyright © Wanfang Data Co. Ltd. All Rights Reserved.
DBID 2B.
4A8
92I
93N
PSX
TCJ
DOI 10.3785/j.issn.1008-9497.2023.06.008
DatabaseName Wanfang Data Journals - Hong Kong
WANFANG Data Centre
Wanfang Data Journals
万方数据期刊 - 香港版
China Online Journals (COJ)
China Online Journals (COJ)
DatabaseTitleList
DeliveryMethod fulltext_linktorsrc
DocumentTitle_FL Unsupervised generalized functional map learning for arbitrary 3D shape dense correspondence
EndPage 744
ExternalDocumentID zjdxxb202306008
GrantInformation_xml – fundername: 辽宁省科技厅应用基础研究计划项目(2023JH2/101300190)
GroupedDBID 2B.
4A8
92I
93N
ALMA_UNASSIGNED_HOLDINGS
GROUPED_DOAJ
PSX
TCJ
ID FETCH-LOGICAL-s1028-10bda7e177ed21355b1f1b00262342e25f8300fc9f151feb1311ffee65b5f3923
ISSN 1008-9497
IngestDate Thu May 29 04:06:15 EDT 2025
IsPeerReviewed true
IsScholarly true
Issue 6
Keywords shape correspondence
functional maps
deep learning
无监督学习
深度学习
函数映射
形状对应
unsupervised learning
Language Chinese
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-s1028-10bda7e177ed21355b1f1b00262342e25f8300fc9f151feb1311ffee65b5f3923
PageCount 9
ParticipantIDs wanfang_journals_zjdxxb202306008
PublicationCentury 2000
PublicationDate 2023-11-25
PublicationDateYYYYMMDD 2023-11-25
PublicationDate_xml – month: 11
  year: 2023
  text: 2023-11-25
  day: 25
PublicationDecade 2020
PublicationTitle 浙江大学学报（理学版）
PublicationTitle_FL Journal of Zhejiang University(Science Edition)
PublicationYear 2023
Publisher 辽宁师范大学 计算机与人工智能学院,辽宁 大连 116081
Publisher_xml – name: 辽宁师范大学 计算机与人工智能学院,辽宁 大连 116081
SSID ssib001051298
ssib051373732
ssib002258177
ssib004369313
ssib008679801
ssib002476865
ssib023167501
ssib059160192
ssib000948450
ssib002040240
ssib001104615
ssib000969734
ssib008143637
ssib006704891
ssj0002507526
Score 2.37582
Snippet TP391.41; 提出了一种新颖的广义无监督函数映射学习的三维形状密集对应方法.首先,基于多层感知器(multilayer perceptron,MLP)和残差网络,直接学习深度点特征.其次,计算点...
SourceID wanfang
SourceType Aggregation Database
StartPage 736
Title 广义无监督函数映射学习的三维形状密集对应方法
URI https://d.wanfangdata.com.cn/periodical/zjdxxb202306008
Volume 50
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVAON
  databaseName: DOAJ Directory of Open Access Journals
  issn: 1008-9497
  databaseCode: DOA
  dateStart: 20220101
  customDbUrl:
  isFulltext: true
  dateEnd: 99991231
  titleUrlDefault: https://www.doaj.org/
  omitProxy: true
  ssIdentifier: ssj0002507526
  providerName: Directory of Open Access Journals
– providerCode: PRVEBS
  databaseName: Inspec with Full Text
  issn: 1008-9497
  databaseCode: ADMLS
  dateStart: 20190301
  customDbUrl:
  isFulltext: true
  dateEnd: 99991231
  titleUrlDefault: https://www.ebsco.com/products/research-databases/inspec-full-text
  omitProxy: false
  ssIdentifier: ssib001051298
  providerName: EBSCOhost
– providerCode: PRVHPJ
  databaseName: ROAD: Directory of Open Access Scholarly Resources
  issn: 1008-9497
  databaseCode: M~E
  dateStart: 19990101
  customDbUrl:
  isFulltext: true
  dateEnd: 99991231
  titleUrlDefault: https://road.issn.org
  omitProxy: true
  ssIdentifier: ssib059160192
  providerName: ISSN International Centre
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnR3LahRBcAgbES-iqPgMOaRPYeO8-nWc2cwSxHgxgdzCzs6M4mEFk0DIOUTiTTAqBiEoeJMEFHGV4M9kdvEvrKp5bJuIL3a3qa2qqeqqnpmu6unusawp4SUije2kmeF8HR_Hb5QW3Wba8f0EMoo4obk583fE3KJ_a4kvjTVOGbOW1lbjme7GL9eV_E-rAg7aFVfJ_kPL1kIBATC0L5TQwlD-VRuziLNQs7DNIh8BpVkkmJYssFkkmQ6ZdgiYZYGHzJD5h7PEAwfaBChipr_KRyAAZlEKLOUERAKMIhUgJGQhMYO0wEWMClgo6PA2UwBo1K4qDIhC5oBpn5QKwkDpseLll1V8TEjOdEF1mG6TBJ8F0qhbDQjUG3AWgYoWU4pqa1d6Cx6JdUZSwVMPfCAlCCpbFThgRAHL4dsiSgvtLyo96rRZhAeQ7eRD8CdWGALzcMQiyPGq8pz8WUo51OJ6uOawWJZNFweJjqidwIYIQn3ynSIboAQbvJNOmUYacAfU4ADoQl0LvV5WICIgIAxIlOQ5gc5GjEK5pdbCcxpJYKLbOlmnaaMKCs9AHU07jrCLN-OU3RxOetF-MTO66geLDYDL693s1IrFIVV8JIv9Oo93vZ5UnLpeVDBTK5hBP9IOubYahRz1RNCNB8n6euxSIkxr9sddKYTbsMaD2fnbd41EwFe--fxdCy3NjQ1tDFyN5784dcEIdF3oq1zfXPHNlSPNgQpIu83n_Z7QnhHoCgk9nbGrlII0QxiBPW5bqeya7uL-Enz0nzuehE-dSHBIyjC3qkeDIRGRnN4NWXvutDVVuvXm75xKiw17Wad3z4iLF85ZZ8uEdjIo7k7nrbGN-xesMO9_O-pvD17sDXefDl-_zR8fDnYOBi_38oPN_P27o_7e8NXm0eft4deP-eGb4ZNP-f7W992tfL-ff3k2eN4ffNi5aC22o4XWXLN8W0tzBZMUiOfipCNT8GqauA6kMbGTOTTG43q-m7o8U55tZ12dQZKRQYjoOU6WpangMc8gS_MuWY3ew1562ZpUKrFj3-7GEudtSK_DeRbDD10EQtMr1mRp83J5N15ZPnYmXf0zyzXrzOgiv241Vh-tpTcgw1iNJ8rTb4JG6H4AIaHU2g
linkProvider EBSCOhost
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=%E5%B9%BF%E4%B9%89%E6%97%A0%E7%9B%91%E7%9D%A3%E5%87%BD%E6%95%B0%E6%98%A0%E5%B0%84%E5%AD%A6%E4%B9%A0%E7%9A%84%E4%B8%89%E7%BB%B4%E5%BD%A2%E7%8A%B6%E5%AF%86%E9%9B%86%E5%AF%B9%E5%BA%94%E6%96%B9%E6%B3%95&rft.jtitle=%E6%B5%99%E6%B1%9F%E5%A4%A7%E5%AD%A6%E5%AD%A6%E6%8A%A5%EF%BC%88%E7%90%86%E5%AD%A6%E7%89%88%EF%BC%89&rft.au=%E7%AA%A6%E4%B8%B0&rft.au=%E9%A9%AC%E4%BC%9A%E6%96%87&rft.au=%E8%B0%A2%E6%98%95%E6%B4%8B&rft.au=%E6%9D%A8%E4%B8%87%E6%96%87&rft.date=2023-11-25&rft.pub=%E8%BE%BD%E5%AE%81%E5%B8%88%E8%8C%83%E5%A4%A7%E5%AD%A6+%E8%AE%A1%E7%AE%97%E6%9C%BA%E4%B8%8E%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%E5%AD%A6%E9%99%A2%2C%E8%BE%BD%E5%AE%81+%E5%A4%A7%E8%BF%9E+116081&rft.issn=1008-9497&rft.volume=50&rft.issue=6&rft.spage=736&rft.epage=744&rft_id=info:doi/10.3785%2Fj.issn.1008-9497.2023.06.008&rft.externalDocID=zjdxxb202306008
thumbnail_s http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=http%3A%2F%2Fwww.wanfangdata.com.cn%2Fimages%2FPeriodicalImages%2Fzjdxxb%2Fzjdxxb.jpg