基于最优传输特征聚合的温室视觉位置识别方法

S24%TP183; 为实现温室场景中基于视觉的位置识别,解决现有视觉位置识别模型局部特征聚合范式对训练样本归纳偏置的强依赖,以及聚合过程中存在的冗余信息问题,构建了一种基于最优传输局部特征聚合的温室视觉位置识别方法.将温室场景图像局部特征聚合过程视为最优传输问题,根据局部特征集动态生成分配矩阵,解耦模型对归纳偏置的强依赖,同时在分配中引入"垃圾"簇来解决特征冗余.结合卷积神经网络(convolution neural network,CNN)和Transformer的优势,优化设计温室场景图像局部特征提取网络.试验结果表明,在种植作物为番茄的温室场景中,所提方法的位置识别...

Full description

Saved in:
Bibliographic Details
Published in农业工程学报 Vol. 40; no. 22; pp. 161 - 172
Main Authors 侯玉涵, 周云成, 刘泽钰, 张润池, 周金桥
Format Journal Article
LanguageChinese
Published 沈阳农业大学信息与电气工程学院,沈阳 110866 01.11.2024
Subjects
Online AccessGet full text
ISSN1002-6819
DOI10.11975/j.issn.1002-6819.202407155

Cover

Abstract S24%TP183; 为实现温室场景中基于视觉的位置识别,解决现有视觉位置识别模型局部特征聚合范式对训练样本归纳偏置的强依赖,以及聚合过程中存在的冗余信息问题,构建了一种基于最优传输局部特征聚合的温室视觉位置识别方法.将温室场景图像局部特征聚合过程视为最优传输问题,根据局部特征集动态生成分配矩阵,解耦模型对归纳偏置的强依赖,同时在分配中引入"垃圾"簇来解决特征冗余.结合卷积神经网络(convolution neural network,CNN)和Transformer的优势,优化设计温室场景图像局部特征提取网络.试验结果表明,在种植作物为番茄的温室场景中,所提方法的位置识别 top-1 召回率(R@1)为 88.96%,与 NetVLAD、MixVPR 和 EigenPlaces 3 种方法相比,R@1 分别提高 29.67、2.97和2.89个百分点.与NetVLAD和MixVPR的聚合器相比,基于最优传输局部特征聚合的R@1分别提高21.65和1.09个百分点.相较于CNN网络,研究构建的温室场景图像局部特征提取网络在R@1指标上提升了 5.45个百分点.所提方法的实际温室场景位置识别率不低于81.94%,具有一定的实际应用能力.基于最优传输局部特征聚合及全局描述符生成方法对位置识别是有效的,场景图像局部特征提取网络能够提高位置识别性能,研究结果可为温室智能农机装备视觉系统设计提供技术参考.
AbstractList S24%TP183; 为实现温室场景中基于视觉的位置识别,解决现有视觉位置识别模型局部特征聚合范式对训练样本归纳偏置的强依赖,以及聚合过程中存在的冗余信息问题,构建了一种基于最优传输局部特征聚合的温室视觉位置识别方法.将温室场景图像局部特征聚合过程视为最优传输问题,根据局部特征集动态生成分配矩阵,解耦模型对归纳偏置的强依赖,同时在分配中引入"垃圾"簇来解决特征冗余.结合卷积神经网络(convolution neural network,CNN)和Transformer的优势,优化设计温室场景图像局部特征提取网络.试验结果表明,在种植作物为番茄的温室场景中,所提方法的位置识别 top-1 召回率(R@1)为 88.96%,与 NetVLAD、MixVPR 和 EigenPlaces 3 种方法相比,R@1 分别提高 29.67、2.97和2.89个百分点.与NetVLAD和MixVPR的聚合器相比,基于最优传输局部特征聚合的R@1分别提高21.65和1.09个百分点.相较于CNN网络,研究构建的温室场景图像局部特征提取网络在R@1指标上提升了 5.45个百分点.所提方法的实际温室场景位置识别率不低于81.94%,具有一定的实际应用能力.基于最优传输局部特征聚合及全局描述符生成方法对位置识别是有效的,场景图像局部特征提取网络能够提高位置识别性能,研究结果可为温室智能农机装备视觉系统设计提供技术参考.
Abstract_FL As the foundation for implementing closed-loop detection within the realm of visual SLAM(simultaneous localization and mapping),visual place recognition(VPR)has great potential in various applications of greenhouse robot navigation and other fields.However,the existing VPR cannot fully meet the actual requirements of greenhouse scenes due to the complexity and constant variations in the greenhouse environment.In particular,the local feature aggregation paradigm strongly depends on the induction bias of training samples in VPR models,which leads to the issue of information redundancy during feature aggregation.In this study,a greenhouse VPR was presented,according to the optimal transport of local feature aggregation.The process of aggregating local features into a global descriptor was framed as an optimal transport problem,where the cost matrix was predicted through an MLP(multi-layer perceptron).Thus,a cost matrix was dynamically generated using the local features that was extracted from the greenhouse scene images.Additionally,a'dustbin'cluster was introduced into the cost matrix to allocate the redundant features.Taking the cost matrix as the input,the Sinkhornalgorithm was employed to determine an optimal solution to the assignment matrix.Furthermore,the soft assignment of local features to various clusters was achieved through the assignment matrix.Ultimately,the assignment was concatenated to form a global descriptor for the scene image,which was used for place recognition.A deep neural network(DNN)was optimized and designed to serve as the backbone for local feature extraction of greenhouse scene images,by combining the advantages of CNN(convolutional neural network)and Transformer.Furthermore,cosine similarity was used as the metric function to calculate the similarity measure between scene image global descriptors,so as to perform descriptor matching.A series of experiments were conducted in a tomato greenhouse.The experimental results showed that the improved model achieved better performance.The top-1 recall rate(R@1)for place recognition was achieved at 88.96%,which was 29.67,2.97,and 2.89 percentage points higher than the those of NetVLAD,MixVPR,and EigenPlaces models,respectively.When compared to the aggregators employed in MixVPR and NetVLAD,our aggregator achieved improvements in R@1 by 1.09 and 21.65 percentage points,respectively,showcasing its effectiveness.Compared with the CNN,the improved network achieved an increase of 5.45 percentage points in R@1.There was even more pronounced R@1 improvement(reaching 10.48 percentage points),compared with a Transformer network.Simultaneously,our network resulted in a 1.6-fold increase in computation speed compared to the previous Transformer.In addition,the experiments further demonstrated that the improved model exhibited excellent performance of place recognition and strong robustness when dealing with factors,such as small sampling distance shifts,small viewpoint shifts,and different sunlight intensities.The greenhouse VPR achieved a place recognition rate of no less than 81.94%in actual greenhouses,indicating its practical application potential.The method based on optimal transport of local feature aggregation and global descriptor generation was effective for place recognition,and the image local feature extraction network can boost the performance of place recognition.These findings can provide technical support to the visual systems of intelligent agricultural machinery in the greenhouse.
Author 张润池
周云成
侯玉涵
周金桥
刘泽钰
AuthorAffiliation 沈阳农业大学信息与电气工程学院,沈阳 110866
AuthorAffiliation_xml – name: 沈阳农业大学信息与电气工程学院,沈阳 110866
Author_FL HOU Yuhan
ZHOU Jinqiao
ZHOU Yuncheng
LIU Zeyu
ZHANG Runchi
Author_FL_xml – sequence: 1
  fullname: HOU Yuhan
– sequence: 2
  fullname: ZHOU Yuncheng
– sequence: 3
  fullname: LIU Zeyu
– sequence: 4
  fullname: ZHANG Runchi
– sequence: 5
  fullname: ZHOU Jinqiao
Author_xml – sequence: 1
  fullname: 侯玉涵
– sequence: 2
  fullname: 周云成
– sequence: 3
  fullname: 刘泽钰
– sequence: 4
  fullname: 张润池
– sequence: 5
  fullname: 周金桥
BookMark eNo9j7tKA0EYhaeIYIx5CgurXf9_Zmd3thIJ3iBgo3WYvYUEmYCDqJ0BCUmhCKJIELQQggixsDEhydPs7TFcUSwOB07xnXNWSEl1VEjIGoKJ6Dp8o222tFYmAlDDFuiaFKgFDnJeIuX_dJlUtW55wJE5ABaWyWbyPI2nN-nTZTx7jGcv-eIuG0ySRTfvDpPbfja8Sr_ekvFrPurlo0E8v87m4_yjl_Tf04dJ-nm_SpYieazD6p9XyNHO9mFtz6gf7O7XtuqGRqDCcH0UlEkIJEpKES0bBQ89HxwIhFtsiXzP5RiFhQQPhIOU-SGjkS-oZzOHVcj6L_dMqkiqZqPdOT1RRWNDXTT9c-_nLqWAgn0DiYVk-g
ClassificationCodes S24%TP183
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 10.11975/j.issn.1002-6819.202407155
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
Discipline Agriculture
DocumentTitle_FL Recognizing visual position in the greenhouse using optimal transport feature aggregation
EndPage 172
ExternalDocumentID nygcxb202422018
GrantInformation_xml – fundername: (国家重点研发计划); (国家重点研发计划)
  funderid: (国家重点研发计划); (国家重点研发计划)
GroupedDBID -04
2B.
4A8
5XA
5XE
92G
92I
93N
ABDBF
ABJNI
ACGFO
ACGFS
ACUHS
AEGXH
AIAGR
ALMA_UNASSIGNED_HOLDINGS
CCEZO
CHDYS
CW9
EOJEC
FIJ
IPNFZ
OBODZ
PSX
RIG
TCJ
TGD
TUS
U1G
U5N
ID FETCH-LOGICAL-s1028-9c1823a0da1a221146185ebc070d89700fcb951fe51f85d87123ce32fc82b6373
ISSN 1002-6819
IngestDate Thu May 29 04:08:37 EDT 2025
IsPeerReviewed false
IsScholarly true
Issue 22
Keywords Transformer
视觉位置识别
温室
深度神经网络
特征聚合
greenhouse
optimal transport
feature aggregation
visual place recognition
最优传输
deep neural network
Language Chinese
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-s1028-9c1823a0da1a221146185ebc070d89700fcb951fe51f85d87123ce32fc82b6373
PageCount 12
ParticipantIDs wanfang_journals_nygcxb202422018
PublicationCentury 2000
PublicationDate 2024-11-01
PublicationDateYYYYMMDD 2024-11-01
PublicationDate_xml – month: 11
  year: 2024
  text: 2024-11-01
  day: 01
PublicationDecade 2020
PublicationTitle 农业工程学报
PublicationTitle_FL Transactions of the Chinese Society of Agricultural Engineering
PublicationYear 2024
Publisher 沈阳农业大学信息与电气工程学院,沈阳 110866
Publisher_xml – name: 沈阳农业大学信息与电气工程学院,沈阳 110866
SSID ssib051370041
ssj0041925
ssib001101065
ssib023167668
Score 2.483521
Snippet S24%TP183; 为实现温室场景中基于视觉的位置识别,解决现有视觉位置识别模型局部特征聚合范式对训练样本归纳偏置的强依赖,以及聚合过程中存在的冗余信息问题,构建了一种基...
SourceID wanfang
SourceType Aggregation Database
StartPage 161
Title 基于最优传输特征聚合的温室视觉位置识别方法
URI https://d.wanfangdata.com.cn/periodical/nygcxb202422018
Volume 40
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVEBS
  databaseName: EBSCOhost Academic Search Ultimate
  issn: 1002-6819
  databaseCode: ABDBF
  dateStart: 20140101
  customDbUrl: https://search.ebscohost.com/login.aspx?authtype=ip,shib&custid=s3936755&profile=ehost&defaultdb=asn
  isFulltext: true
  dateEnd: 99991231
  titleUrlDefault: https://search.ebscohost.com/direct.asp?db=asn
  omitProxy: true
  ssIdentifier: ssj0041925
  providerName: EBSCOhost
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnR1Na9RAdGi3IHoQP_GbBZ2TZN1MMsnMSSa7WYqgpxZ6K0k2WU9baLegPVmQ0h4UQRQpgh6EIkI9eLFl21-zXz_D9yazH7b1E3aHl5mX9zEvk_cmzJsh5I7rZw7zZWTxSMIEBS4t6UnXiu00yziTPsswwfnhI2923n2wwBemptsTq5ZWW3EpWTsxr-R_rAp1YFfMkv0Hy46IQgXAYF8owcJQ_pWNaciprNFA0dDFUoQ09Kis4PIFrKlQKQygoEbQIKTSoaFPhaSBxNuhRtjYBKVUmmCZCoE4cClcJBgIqjSyCqlyEVkBBW8ISM2iSkUV7wIA0LCppnE4UlOBFszTTKF0aH7m5TAs1mhaciQljCSBTxVHmgrEC7QAQNxDCkJB0_Bh0TeFyBA1C7VEwAR-EyigmI2E8p4CGKkAp_IkClYIIyEoEkoqGQ1-QjF9qennsgRAtzz59YS5Jo1w-LxrLKa7VSIDoH6iwtC50KFjPaGpRlUuqq3108hoZjAPwFxTLiP8u_4CppIqBsPpuCR3MUMjP5LGOCf0Xp4wLsZ4r3yzKzNK8xRv44vsfJd7E9bY-QlJxz2m9Ll2mciiNGJRws6C8DPfQ_nIluTNp43kSYwYDOJHMU1mmO95rEBmVFANauNw3MYvDiN_wXDXBW88veW2g4crjJZk4YIErlcnGCFOkdtDEe_9WkCdbtfMomZjIjKcO0fOmildUeXj8zyZWnt8gZxRjWWzrU16kdzvftjv7L_svX_Wab_rtD8ODl_3t_a6h-uD9e3uq83-9vPe98_d3U-DnY3Bzlbn4EX_YHfwdaO7-aX3dq_37c0lMl8L5yqzljm6xFrBiN2SCczbnahcj-yIMcz8h7g4jRNwsHUhQessiWFuk6XwF7wufAggk9RhWSJY7Dm-c5kUmkvN9Aop4gkJ0GrzOo_gXSqFX5eJSBl421jGZfsqKRr1F82raWXxiIGu_RnlOjk9Hh43SKG1vJrehHC7Fd8yVv0B3G6dUQ
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%9F%BA%E4%BA%8E%E6%9C%80%E4%BC%98%E4%BC%A0%E8%BE%93%E7%89%B9%E5%BE%81%E8%81%9A%E5%90%88%E7%9A%84%E6%B8%A9%E5%AE%A4%E8%A7%86%E8%A7%89%E4%BD%8D%E7%BD%AE%E8%AF%86%E5%88%AB%E6%96%B9%E6%B3%95&rft.jtitle=%E5%86%9C%E4%B8%9A%E5%B7%A5%E7%A8%8B%E5%AD%A6%E6%8A%A5&rft.au=%E4%BE%AF%E7%8E%89%E6%B6%B5&rft.au=%E5%91%A8%E4%BA%91%E6%88%90&rft.au=%E5%88%98%E6%B3%BD%E9%92%B0&rft.au=%E5%BC%A0%E6%B6%A6%E6%B1%A0&rft.date=2024-11-01&rft.pub=%E6%B2%88%E9%98%B3%E5%86%9C%E4%B8%9A%E5%A4%A7%E5%AD%A6%E4%BF%A1%E6%81%AF%E4%B8%8E%E7%94%B5%E6%B0%94%E5%B7%A5%E7%A8%8B%E5%AD%A6%E9%99%A2%2C%E6%B2%88%E9%98%B3+110866&rft.issn=1002-6819&rft.volume=40&rft.issue=22&rft.spage=161&rft.epage=172&rft_id=info:doi/10.11975%2Fj.issn.1002-6819.202407155&rft.externalDocID=nygcxb202422018
thumbnail_s http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=http%3A%2F%2Fwww.wanfangdata.com.cn%2Fimages%2FPeriodicalImages%2Fnygcxb%2Fnygcxb.jpg