Article(id=1241110428361740642, tenantId=1146029695717560320, journalId=1238841944844054536, issueId=1241110425031471264, articleNumber=null, orderNo=null, doi=10.12347/j.ycyk.20240116002, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1705334400000, receivedDateStr=2024-01-16, revisedDate=1710864000000, revisedDateStr=2024-03-20, acceptedDate=null, acceptedDateStr=null, onlineDate=1773833384838, onlineDateStr=2026-03-18, pubDate=1715702400000, pubDateStr=2024-05-15, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1773833384838, onlineIssueDateStr=2026-03-18, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1773833384838, creator=13701087609, updateTime=1773833384838, updator=13701087609, issue=Issue{id=1241110425031471264, tenantId=1146029695717560320, journalId=1238841944844054536, year='2024', volume='45', issue='3', pageStart='1', pageEnd='122', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1773833384044, creator=13701087609, updateTime=1773834311145, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1241114313633624872, tenantId=1146029695717560320, journalId=1238841944844054536, issueId=1241110425031471264, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1241114313633624873, tenantId=1146029695717560320, journalId=1238841944844054536, issueId=1241110425031471264, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=35, endPage=42, ext={EN=ArticleExt(id=1241110429779415401, articleId=1241110428361740642, tenantId=1146029695717560320, journalId=1238841944844054536, language=EN, title=PolSAR Terrain Classification Based on Transformer-Unet Hybrid Model in the Complex Domain, columnId=1241060179823882575, journalTitle=Journal of Telemetry, Tracking and Command, columnName=Artificial Intelligence Technology, runingTitle=null, highlight=null, articleAbstract=

The traditional deep learning-based Polarimetric Synthetic Aperture Radar (PolSAR) feature classification method extracts image local features by stacking convolutional layers, which makes it difficult to establish long-range dependencies. It is not-ed that Transformer, a deep learning model based on a self-attention mechanism that captures global pixel-to-pixel correlations, has achieved success in image classification tasks. Meanwhile, the PolSAR feature classification task has demonstrated better classification results in the complex domain compared to the real domain. Therefore, Transformer is introduced into the complex domain, and a hybrid model of Transformer and Unet based on the complex domain (CT-Unet) is proposed for PolSAR feature classification. This model combines Transformer with CNN for feature extraction on PolSAR data of complex type. The experimental results of PolSAR feature classification using the Xi'an dataset and German dataset show that the proposed model can effectively improve the accuracy of PolSAR feature classification. Transformer is expected to make up for the shortcomings of convolutional neural net-works in the PolSAR feature classification task.

, 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=Wen XIE, Jiapeng ZHANG, Zhezhe ZHANG, Chenchao SHAN), CN=ArticleExt(id=1241110431956259228, articleId=1241110428361740642, tenantId=1146029695717560320, journalId=1238841944844054536, language=CN, title=基于复数域Transformer-Unet混合模型的PolSAR地物分类, columnId=1241060179987460434, journalTitle=遥测遥控, columnName=人工智能技术, runingTitle=null, highlight=null, articleAbstract=

传统的基于深度学习的极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,PolSAR)地物分类方法,通过堆叠卷积层提取图像局部特征,难以建立长距离依赖关系。基于自注意力机制的深度学习模型Transformer(变换)在图像分类任务中取得了成功,其自注意力机制能够捕获全局像素之间的关联性,同时PolSAR地物分类任务被证实:相比于实数域,其在复数域上表现出更好的分类效果。因此,本文将Transformer引入到复数域中,提出了一种基于复数域的Transformer和Unet(语义分割网络)混合模型(CT-Unet)用于PolSAR地物分类,将Transformer与CNN相结合,对复数类型的PolSAR数据进行特征提取,使用西安数据集和德国数据集进行PolSAR地物分类的实验结果表明:提出的模型能够有效提高PolSAR地物分类的准确性,Transformer有望在PolSAR地物分类任务中弥补卷积神经网络的不足。

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=8CmMiUmW+bCeK4jjP6kYTA==, magXml=n6xK32fnxa+ZH9rytAZvQg==, pdfUrl=null, pdf=HMkIcHFRUtnw5562k7eNDg==, pdfFileSize=5874702, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=EbnXcx/TSL0ifm396v+/4A==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=XsQ1+FHTN95MtwoD0kUuFw==, mapNumber=null, authorCompany=null, fund=null, authors=

谢雯 1989年生,副教授,硕士生导师。

张嘉鹏 1998年生,硕士研究生。

张哲哲 1997年生,硕士研究生。

闪晨超 1999年生,硕士研究生。

, authorsList=谢雯, 张嘉鹏, 张哲哲, 闪晨超)}, authors=[Author(id=1241113370364015571, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, orderNo=0, 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=1241113370452095961, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113370364015571, language=EN, stringName=Wen XIE, firstName=Wen, middleName=null, lastName=XIE, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1241113370510816220, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113370364015571, language=CN, stringName=谢雯, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=西安邮电大学通信与信息工程学院 西安 710121, bio={"content":"

谢雯 1989年生,副教授,硕士生导师。

"}, bioImg=null, bioContent=

谢雯 1989年生,副教授,硕士生导师。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1241113370288518090, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, xref=null, ext=[AuthorCompanyExt(id=1241113370292712395, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China), AuthorCompanyExt(id=1241113370301101004, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=西安邮电大学通信与信息工程学院 西安 710121)])]), Author(id=1241113370624062436, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, 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=1241113370724725735, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113370624062436, language=EN, stringName=Jiapeng ZHANG, firstName=Jiapeng, middleName=null, lastName=ZHANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1241113370804417514, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113370624062436, language=CN, stringName=张嘉鹏, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=西安邮电大学通信与信息工程学院 西安 710121, bio={"content":"

张嘉鹏 1998年生,硕士研究生。

"}, bioImg=null, bioContent=

张嘉鹏 1998年生,硕士研究生。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1241113370288518090, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, xref=null, ext=[AuthorCompanyExt(id=1241113370292712395, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China), AuthorCompanyExt(id=1241113370301101004, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=西安邮电大学通信与信息工程学院 西安 710121)])]), Author(id=1241113372339532786, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, 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=1241113372436001782, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113372339532786, language=EN, stringName=Zhezhe ZHANG, firstName=Zhezhe, middleName=null, lastName=ZHANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1241113372536665083, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113372339532786, language=CN, stringName=张哲哲, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=西安邮电大学通信与信息工程学院 西安 710121, bio={"content":"

张哲哲 1997年生,硕士研究生。

"}, bioImg=null, bioContent=

张哲哲 1997年生,硕士研究生。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1241113370288518090, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, xref=null, ext=[AuthorCompanyExt(id=1241113370292712395, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China), AuthorCompanyExt(id=1241113370301101004, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=西安邮电大学通信与信息工程学院 西安 710121)])]), Author(id=1241113372637328381, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, 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=1241113372729603073, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113372637328381, language=EN, stringName=Chenchao SHAN, firstName=Chenchao, middleName=null, lastName=SHAN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1241113372847042562, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, authorId=1241113372637328381, language=CN, stringName=闪晨超, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=西安邮电大学通信与信息工程学院 西安 710121, bio={"content":"

闪晨超 1999年生,硕士研究生。

"}, bioImg=null, bioContent=

闪晨超 1999年生,硕士研究生。

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1241113370288518090, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, xref=null, ext=[AuthorCompanyExt(id=1241113370292712395, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China), AuthorCompanyExt(id=1241113370301101004, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=西安邮电大学通信与信息工程学院 西安 710121)])])], keywords=[Keyword(id=1241113373144838152, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, orderNo=1, keyword=Polarimetric synthetic aperture radar(PolSAR)), Keyword(id=1241113373237112843, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, orderNo=2, keyword=Complex domain), Keyword(id=1241113373316804623, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, orderNo=3, keyword=Transformer), Keyword(id=1241113373392302098, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, orderNo=4, keyword=Unet), Keyword(id=1241113373480382484, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, orderNo=1, keyword=极化合成孔径雷达), Keyword(id=1241113376869380127, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, orderNo=2, keyword=复数域), Keyword(id=1241113376982626340, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, orderNo=3, keyword=Transformer), Keyword(id=1241113377066512421, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, orderNo=4, keyword=Unet)], refs=[Reference(id=1241113379264327821, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2015, volume=36, issue=5, pageStart=62, pageEnd=67, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=叶晖, 李博, 王金华, journalName=遥测遥控, refType=null, unstructuredReference=叶晖, 李博, 王金华, 等. 光学遥感卫星地面图像数据测试的局限及解决方法初探[J]. 遥测遥控, 2015, 36(5):62-67., articleTitle=光学遥感卫星地面图像数据测试的局限及解决方法初探, refAbstract=null), Reference(id=1241113379348213903, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2015, volume=36, issue=5, pageStart=62, pageEnd=67, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=YE Hui, LI Bo, WANG Jinhua, journalName=Journal of Telemetry, Tracking and Command, refType=null, unstructuredReference=YE Hui, LI Bo, WANG Jinhua, et al. Preliminary exploration of a new method for image data testing of optical sat-ellites[J]. Journal of Telemetry, Tracking and Command, 2015, 36(5): 62-67., articleTitle=Preliminary exploration of a new method for image data testing of optical sat-ellites, refAbstract=null), Reference(id=1241113379457265813, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=35, issue=3, pageStart=44, pageEnd=49, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=肖雨彤, 张继贤, 黄国满, journalName=遥感信息, refType=null, unstructuredReference=肖雨彤, 张继贤, 黄国满, 等. FCN与CRF结合的Pol-SAR影像建筑区域提取[J]. 遥感信息, 2020, 35(3):44-49., articleTitle=FCN与CRF结合的Pol-SAR影像建筑区域提取, refAbstract=null), Reference(id=1241113379553734811, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=35, issue=3, pageStart=44, pageEnd=49, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=XIAO Yutong, ZHANG Jixian, HUANG Guoman, journalName=Remote Sensing Information, refType=null, unstructuredReference=XIAO Yutong, ZHANG Jixian, HUANG Guoman, et al. Building area extraction of PolSAR images based on FCN and CRF[J]. Remote Sensing Information, 2020, 35(3): 44-49., articleTitle=Building area extraction of PolSAR images based on FCN and CRF, refAbstract=null), Reference(id=1241113379696341152, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2017, volume=48, issue=12, pageStart=174, pageEnd=182, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=郭交, 尉鹏亮, 周正舒, journalName=农业机械学报, refType=null, unstructuredReference=郭交, 尉鹏亮, 周正舒, 等. 基于时变特征的多时相Pol-SAR农作物分类方法[J]. 农业机械学报, 2017, 48(12):174-182., articleTitle=基于时变特征的多时相Pol-SAR农作物分类方法, refAbstract=null), Reference(id=1241113379801198756, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2017, volume=48, issue=12, pageStart=174, pageEnd=182, url=null, language=null, rfNumber=[3], rfOrder=5, authorNames=GUO Jiao, WEI Pengliang, ZHOU Zhengshu, journalName=Transactions of the Chinese Society for Agricultural Machinery, refType=null, unstructuredReference=GUO Jiao, WEI Pengliang, ZHOU Zhengshu, et al. Crop classification method with differential characteristics based on multi-temporal PolSAR images[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48(12): 174-182., articleTitle=Crop classification method with differential characteristics based on multi-temporal PolSAR images, refAbstract=null), Reference(id=1241113381306953898, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2009, volume=47, issue=8, pageStart=2714, pageEnd=2721, url=null, language=null, rfNumber=[4], rfOrder=6, authorNames=OH Y, HONG S Y, KIM Y, journalName=IEEE Transactions on Geoscience and Remote Sensing, refType=null, unstructuredReference=OH Y, HONG S Y, KIM Y, et al. Polarimetric backscattering coefficients of flooded rice fields at L-and C-bands:Measurements, modeling, and data analysis[J]. IEEE Transactions on Geoscience and Remote Sensing, 2009, 47(8): 2714-2721., articleTitle=Polarimetric backscattering coefficients of flooded rice fields at L-and C-bands:Measurements, modeling, and data analysis, refAbstract=null), Reference(id=1241113381395034287, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2024, volume=73, issue=1, pageStart=426, pageEnd=439, url=null, language=null, rfNumber=[5], rfOrder=7, authorNames=SHARIF M, ATTARCHI S, journalName=Advances in Space Research, refType=null, unstructuredReference=SHARIF M, ATTARCHI S. Estimation of hyrcanian forests coverage using ALOS PALSAR and MODIS satellite data (2007—2017)[J]. Advances in Space Research, 2024, 73(1): 426-439., articleTitle=Estimation of hyrcanian forests coverage using ALOS PALSAR and MODIS satellite data (2007—2017), refAbstract=null), Reference(id=1241113381462143155, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2016, volume=116, issue=null, pageStart=86, pageEnd=100, url=null, language=null, rfNumber=[6], rfOrder=8, authorNames=LIANG D, TANG T, BAN Y, journalName=ISPRS Jour-nal of Photogrammetry and Remote Sensing, refType=null, unstructuredReference=LIANG D, TANG T, BAN Y, et al. Unsupervised polari-metric SAR urban area classification based on model-based decomposition with cross scattering[J]. ISPRS Jour-nal of Photogrammetry and Remote Sensing, 2016, 116:86-100., articleTitle=Unsupervised polari-metric SAR urban area classification based on model-based decomposition with cross scattering, refAbstract=null), Reference(id=1241113381567000762, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[7], rfOrder=9, authorNames=刘振, journalName=null, refType=null, unstructuredReference=刘振. 小样本PolSAR图像地物分类技术研究[D]. 西安: 西安电子科技大学, 2018., articleTitle=小样本PolSAR图像地物分类技术研究, refAbstract=null), Reference(id=1241113381676052668, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=11, issue=1, pageStart=1, pageEnd=32, url=null, language=null, rfNumber=[8], rfOrder=10, authorNames=PARIKH H, PATEL S, PATEL V, journalName=In-ternational Journal of Image and Data Fusion, refType=null, unstructuredReference=PARIKH H, PATEL S, PATEL V. Classification of SAR and PolSAR images using deep learning: A review[J]. In-ternational Journal of Image and Data Fusion, 2020, 11(1): 1-32., articleTitle=Classification of SAR and PolSAR images using deep learning: A review, refAbstract=null), Reference(id=1241113381751550144, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2021, volume=60, issue=null, pageStart=1, pageEnd=18, url=null, language=null, rfNumber=[9], rfOrder=11, authorNames=CUI Y, LIU F, JIAO L, journalName=IEEE Transactions on Geoscience and Remote Sensing, refType=null, unstructuredReference=CUI Y, LIU F, JIAO L, et al. Polarimetric multipath con-volutional neural network for PolSAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 1-18., articleTitle=Polarimetric multipath con-volutional neural network for PolSAR image classification, refAbstract=null), Reference(id=1241113381860602053, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2022, volume=14, issue=15, pageStart=3737, pageEnd=null, url=null, language=null, rfNumber=[10], rfOrder=12, authorNames=XIE W, JIAO L, HUA W, journalName=Remote Sensing, refType=null, unstructuredReference=XIE W, JIAO L, HUA W. Complex-valued multi-scale fully convolutional network with stacked-dilated convo-lution for PolSAR image classification[J]. Remote Sensing, 2022, 14(15): 3737., articleTitle=Complex-valued multi-scale fully convolutional network with stacked-dilated convo-lution for PolSAR image classification, refAbstract=null), Reference(id=1241113381969653962, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2019, volume=12, issue=12, pageStart=5300, pageEnd=5309, url=null, language=null, rfNumber=[11], rfOrder=13, authorNames=MULLISSA A G, PERSELLO C, STEIN A, journalName=IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, refType=null, unstructuredReference=MULLISSA A G, PERSELLO C, STEIN A. PolSAR-Net: A deep fully convolutional network for polarimetric SAR image classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(12): 5300-5309., articleTitle=PolSAR-Net: A deep fully convolutional network for polarimetric SAR image classification, refAbstract=null), Reference(id=1241113382045151439, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=1055, pageEnd=1059, url=null, language=null, rfNumber=[12], rfOrder=14, authorNames=HUANG H, LIN L, TONG R, journalName=null, refType=null, unstructuredReference=HUANG H, LIN L, TONG R, et al. Unet 3+: A full-scale connected unet for medical image segmentation[C]//ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing(ICASSP). IEEE, 2020: 1055-1059., articleTitle=Unet 3+: A full-scale connected unet for medical image segmentation, refAbstract=null), Reference(id=1241113382116454610, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[13], rfOrder=15, authorNames=WANG W, CHEN W, QIU Q, journalName=arXiv preprint arXiv:2303.06908, refType=null, unstructuredReference=WANG W, CHEN W, QIU Q, et al. Crossformer++: A versatile vision transformer hinging on cross-scale attention[J]. arXiv preprint arXiv:2303.06908, 2023., articleTitle=Crossformer++: A versatile vision transformer hinging on cross-scale attention, refAbstract=null), Reference(id=1241113382196146390, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[14], rfOrder=16, authorNames=DOSOVITSKIY A, BEYER L, KOLESNIKOV A, journalName=null, refType=null, unstructuredReference=DOSOVITSKIY A, BEYER L, KOLESNIKOV A,et al. An image is worth 16x16 words: Transformers for image recognition at scale[C]//International Conference on Learning Representations, 2021. arXiv:, articleTitle=An image is worth 16x16 words: Transformers for image recognition at scale, refAbstract=null), Reference(id=1241113382271643863, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2022, volume=44, issue=null, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[15], rfOrder=17, authorNames=陈辉, 张甜, 陈润斌, journalName=电子与信息学报, refType=null, unstructuredReference=陈辉, 张甜, 陈润斌. 基于轻量级卷积Transformer的图像分类方法及在遥感图像分类中的应用[J]. 电子与信息学报, 2022, 44: 1-9., articleTitle=基于轻量级卷积Transformer的图像分类方法及在遥感图像分类中的应用, refAbstract=null), Reference(id=1241113382338752730, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2022, volume=44, issue=null, pageStart=1, pageEnd=9, url=null, language=null, rfNumber=[15], rfOrder=18, authorNames=CHEN Hui, ZHANG Tian, CHEN Runbin, journalName=Journal of Electronics & Information Technology, refType=null, unstructuredReference=CHEN Hui, ZHANG Tian, CHEN Runbin. Image classi-fication method based on lightweight convolutional transformer and its application in remote sensing[J]. Journal of Electronics & Information Technology, 2022, 44: 1-9., articleTitle=Image classi-fication method based on lightweight convolutional transformer and its application in remote sensing, refAbstract=null), Reference(id=1241113382426833117, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2017, volume=55, issue=12, pageStart=7177, pageEnd=7188, url=null, language=null, rfNumber=[16], rfOrder=19, authorNames=ZHANG Z, WANG H, XU F, journalName=IEEE Transactions on Geoscience and Remote Sensing, refType=null, unstructuredReference=ZHANG Z, WANG H, XU F, et al. Complex-valued con-volutional neural network and its application in polari-metric SAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(12): 7177-7188., articleTitle=Complex-valued con-volutional neural network and its application in polari-metric SAR image classification, refAbstract=null), Reference(id=1241113382523302111, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=99, pageStart=1, pageEnd=5, url=null, language=null, rfNumber=[17], rfOrder=20, authorNames=YU L, HU Y, XIE X, journalName=IEEE Geoscience and Remote Sensing Letters, refType=null, unstructuredReference=YU L, HU Y, XIE X, et al. Complex-valued full convo-lutional neural network for SAR target classification[J]. IEEE Geoscience and Remote Sensing Letters, 2019(99): 1-5., articleTitle=Complex-valued full convo-lutional neural network for SAR target classification, refAbstract=null), Reference(id=1241113382657519844, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=8026, pageEnd=8029, url=null, language=null, rfNumber=[18], rfOrder=21, authorNames=XIE W, SUN H, ZHANG Y, journalName=null, refType=null, unstructuredReference=XIE W, SUN H, ZHANG Y, et al. Polsar image classifi-cation via transfer learning and fully convolutional net-work[C]//IGARSS 2023-2023 IEEE International Geo-science and Remote Sensing Symposium. IEEE, 2023:8026-8029., articleTitle=Polsar image classifi-cation via transfer learning and fully convolutional net-work, refAbstract=null), Reference(id=1241113382720434409, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=9, issue=3, pageStart=525, pageEnd=538, url=null, language=null, rfNumber=[19], rfOrder=22, authorNames=秦先祥, 余旺盛, 王鹏, journalName=雷达学报, refType=null, unstructuredReference=秦先祥, 余旺盛, 王鹏, 等. 基于复值卷积神经网络样本精选的极化SAR图像弱监督分类方法[J]. 雷达学报, 2020, 9(3): 525-538., articleTitle=基于复值卷积神经网络样本精选的极化SAR图像弱监督分类方法, refAbstract=null), Reference(id=1241113382787543279, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=9, issue=3, pageStart=525, pageEnd=538, url=null, language=null, rfNumber=[19], rfOrder=23, authorNames=QIN Xianxiang, YU Wangsheng, WANG Peng, journalName=Journal of Radars, refType=null, unstructuredReference=QIN Xianxiang, YU Wangsheng, WANG Peng, et al. Weakly supervised classification of PolSAR images based on sample refinement with complex-valued convo-lutional neural network[J]. Journal of Radars, 2020, 9(3): 525-538., articleTitle=Weakly supervised classification of PolSAR images based on sample refinement with complex-valued convo-lutional neural network, refAbstract=null), Reference(id=1241113382858846450, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=48, issue=9, pageStart=1729, pageEnd=1734, url=null, language=null, rfNumber=[20], rfOrder=24, authorNames=江泽涛, 秦嘉奇, 张少钦, journalName=电子学报, refType=null, unstructuredReference=江泽涛, 秦嘉奇, 张少钦. 参数池化卷积神经网络图像分类方法[J]. 电子学报, 2020, 48(9): 1729-1734., articleTitle=参数池化卷积神经网络图像分类方法, refAbstract=null), Reference(id=1241113382946926836, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2020, volume=48, issue=9, pageStart=1729, pageEnd=1734, url=null, language=null, rfNumber=[20], rfOrder=25, authorNames=JIANG Zetao, QIN Jiaqi, ZHANG Shaoqin, journalName=Acta Electronica Sinica, refType=null, unstructuredReference=JIANG Zetao, QIN Jiaqi, ZHANG Shaoqin. Parame-terized pooling convolution neural network for image classification[J]. Acta Electronica Sinica, 2020, 48(9):1729-1734., articleTitle=Parame-terized pooling convolution neural network for image classification, refAbstract=null), Reference(id=1241113383035007226, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2022, volume=51, issue=3, pageStart=483, pageEnd=491, url=null, language=null, rfNumber=[21], rfOrder=26, authorNames=杜圣杰, 贾晓芬, 黄友锐, journalName=红外与激光工程, refType=null, unstructuredReference=杜圣杰, 贾晓芬, 黄友锐, 等. 用于CNN模型图像分类任务的高效激活函数[J]. 红外与激光工程, 2022, 51(3): 483-491., articleTitle=用于CNN模型图像分类任务的高效激活函数, refAbstract=null), Reference(id=1241113383110504701, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2022, volume=51, issue=3, pageStart=483, pageEnd=491, url=null, language=null, rfNumber=[21], rfOrder=27, authorNames=DU Shengjie, JIA Xiaofen, HUANG Yourui, journalName=Infrared and Laser Engi-neering, refType=null, unstructuredReference=DU Shengjie, JIA Xiaofen, HUANG Yourui, et al. High efficient activation function design for CNN model image classification task[J]. Infrared and Laser Engi-neering, 2022, 51(3): 483-491., articleTitle=High efficient activation function design for CNN model image classification task, refAbstract=null), Reference(id=1241113383194390785, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2018, volume=47, issue=7, pageStart=7, pageEnd=14, url=null, language=null, rfNumber=[22], rfOrder=28, authorNames=张腊梅, 陈泽茜, 邹斌, journalName=红外与激光工程, refType=null, unstructuredReference=张腊梅, 陈泽茜, 邹斌. 基于3D卷积神经网络的Pol-SAR图像精细分类[J]. 红外与激光工程, 2018, 47(7):7-14., articleTitle=基于3D卷积神经网络的Pol-SAR图像精细分类, refAbstract=null), Reference(id=1241113383336997124, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2018, volume=47, issue=7, pageStart=7, pageEnd=14, url=null, language=null, rfNumber=[22], rfOrder=29, authorNames=ZHANG Lamei, CHEN Zexi, ZOU Bin, journalName=Infrared and Laser Engineering, refType=null, unstructuredReference=ZHANG Lamei, CHEN Zexi, ZOU Bin. Fine classification of polarimetric SAR images based on 3D convolutional neural network[J]. Infrared and Laser Engineering, 2018, 47(7): 7-14., articleTitle=Fine classification of polarimetric SAR images based on 3D convolutional neural network, refAbstract=null), Reference(id=1241113383429271818, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=2023, volume=45, issue=1, pageStart=87, pageEnd=110, url=null, language=null, rfNumber=[23], rfOrder=30, authorNames=HAN K, WANG Y, CHEN H, journalName=IEEE Transactions on Pattern Analysis and Machine Intelligence, refType=null, unstructuredReference=HAN K, WANG Y, CHEN H, et al. A survey on vision transformer[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(1): 87-110., articleTitle=A survey on vision transformer, refAbstract=null), Reference(id=1241113383529935112, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=10.1109/IGARSS.1997.606460, pmid=null, pmcid=null, year=1997, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[24], rfOrder=31, authorNames=LOU Y, KIM Y, ZYL J V, journalName=null, refType=null, unstructuredReference=LOU Y, KIM Y, ZYL J V, et al. The NASA/JPL airborne synthetic aperture radar's 1996 pacRim deployment[C]//International Geoscience and Remote Sensing Sympo-sium. IEEE, 1997., articleTitle=The NASA/JPL airborne synthetic aperture radar's 1996 pacRim deployment, refAbstract=null), Reference(id=1241113383613821196, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, doi=null, pmid=null, pmcid=null, year=1992, volume=1, issue=null, pageStart=649, pageEnd=651, url=null, language=null, rfNumber=[25], rfOrder=32, authorNames=VAN ZYL J, CARANDE R, LOU Y, journalName=null, refType=null, unstructuredReference=VAN ZYL J, CARANDE R, LOU Y, et al. The NASA/JPL three-frequency polarimetric AIRSAR system[C]//IGARSS'92 International Geoscience and Remote Sensing Symposium. IEEE, 1992, 1: 649-651., articleTitle=The NASA/JPL three-frequency polarimetric AIRSAR system, refAbstract=null)], funds=[Fund(id=1241113378626793582, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, awardId=61901365, language=CN, fundingSource=国家自然科学基金(61901365), fundOrder=null, country=null), Fund(id=1241113378727456880, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, awardId=62071379, language=CN, fundingSource=国家自然科学基金(62071379), fundOrder=null, country=null), Fund(id=1241113378802954355, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, awardId=2019JQ-377, language=CN, fundingSource=陕西省自然科学基金(2019JQ-377), fundOrder=null, country=null), Fund(id=1241113378870063224, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, awardId=19JK0805, language=CN, fundingSource=陕西省教育厅专项科研计划(19JK0805), fundOrder=null, country=null), Fund(id=1241113378937172093, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, awardId=xyt2016-01, language=CN, fundingSource=西安邮电大学西邮新星团队项目(xyt2016-01), fundOrder=null, country=null), Fund(id=1241113379050418305, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, awardId=null, language=CN, fundingSource=陕西高校青年创新团队, fundOrder=null, country=null)], companyList=[AuthorCompany(id=1241113370288518090, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, xref=null, ext=[AuthorCompanyExt(id=1241113370292712395, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China), AuthorCompanyExt(id=1241113370301101004, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, companyId=1241113370288518090, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=西安邮电大学通信与信息工程学院 西安 710121)])], figs=[ArticleFig(id=1241113377297199151, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, label=Fig. 1, caption=Network architecture of CT-Unet, figureFileSmall=YV9mPafJxrEZCK0MbvWHsA==, figureFileBig=QVnhyYjlgFlVejri5FCoiA==, tableContent=null), ArticleFig(id=1241113377385279536, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, label=图1, caption=CT-Unet网络模型结构图, figureFileSmall=YV9mPafJxrEZCK0MbvWHsA==, figureFileBig=QVnhyYjlgFlVejri5FCoiA==, tableContent=null), ArticleFig(id=1241113377515302966, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, label=Fig. 2, caption=The process of multi-head self-attention mechanism, figureFileSmall=D92Z00EPZgBB/ZeE6yju/w==, figureFileBig=4yQtdYk4SdIZ7oUCk1aj6g==, tableContent=null), ArticleFig(id=1241113377624354875, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, label=图2, caption=多头自注意力机制的过程, figureFileSmall=D92Z00EPZgBB/ZeE6yju/w==, figureFileBig=4yQtdYk4SdIZ7oUCk1aj6g==, tableContent=null), ArticleFig(id=1241113377725018177, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, label=Fig. 3, caption=The classification results of the Xi'an dataset, figureFileSmall=QfRzynDYOhRO7eYmVvKioQ==, figureFileBig=5JHpQT3fcVgj5o776KQG2A==, tableContent=null), ArticleFig(id=1241113377817292872, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, label=图3, caption=西安地区数据集分类结果图, figureFileSmall=QfRzynDYOhRO7eYmVvKioQ==, figureFileBig=5JHpQT3fcVgj5o776KQG2A==, tableContent=null), ArticleFig(id=1241113377896984651, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, label=Fig. 4, caption=The classification results of the Germany dataset, figureFileSmall=6SaD4/f08K3XF8EDtXK3sQ==, figureFileBig=fh3OmbzMdw3AOoyJlJydIg==, tableContent=null), ArticleFig(id=1241113378001842256, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, label=图4, caption=德国ESAR数据集分类结果图, figureFileSmall=6SaD4/f08K3XF8EDtXK3sQ==, figureFileBig=fh3OmbzMdw3AOoyJlJydIg==, tableContent=null), ArticleFig(id=1241113378102505557, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, label=Table 1, caption=

The classification accuracy of the Xi'an dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
FCNUnetTF-CVFCN模型1模型2CT-Unet
草地0.971 50.984 70.989 90.969 10.984 90.991 7
城市0.963 90.950 10.974 20.982 20.978 30.988 9
河流0.926 80.958 50.968 90.956 20.969 40.977 3
OA0.962 10.968 50.981 10.971 80.980 20.988 5
MIou0.913 30.932 10.956 50.940 50.956 40.972 4
Kappa0.931 00.946 60.966 40.953 80.966 30.978 9
GFLOPs0.023 60.032 70.084 0420.655 1437.223 71 545.490 7
Params11 849 74010 898 48042 153 92815 400 48015 505 74627 813 130
), ArticleFig(id=1241113378198974552, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, label=表1, caption=

西安数据集分类准确度

, figureFileSmall=null, figureFileBig=null, tableContent=
FCNUnetTF-CVFCN模型1模型2CT-Unet
草地0.971 50.984 70.989 90.969 10.984 90.991 7
城市0.963 90.950 10.974 20.982 20.978 30.988 9
河流0.926 80.958 50.968 90.956 20.969 40.977 3
OA0.962 10.968 50.981 10.971 80.980 20.988 5
MIou0.913 30.932 10.956 50.940 50.956 40.972 4
Kappa0.931 00.946 60.966 40.953 80.966 30.978 9
GFLOPs0.023 60.032 70.084 0420.655 1437.223 71 545.490 7
Params11 849 74010 898 48042 153 92815 400 48015 505 74627 813 130
), ArticleFig(id=1241113378303832159, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=EN, label=Table 2, caption=

The classification accuracy of the Germany dataset.

, figureFileSmall=null, figureFileBig=null, tableContent=
FCNUnetTF-CVFCN模型1模型2CT-Unet
建筑区0.906 80.923 00.975 00.932 80.955 20.984 4
林区0.939 80.973 70.967 70.970 40.980 90.962 9
开放区0.981 40.989 00.986 00.992 50.988 10.991 2
OA0.954 90.969 60.979 80.973 40.978 50.984 2
MIou0.891 80.926 80.953 70.932 80.950 90.960 1
Kappa0.914 00.942 80.964 30.947 80.962 10.969 3
GFLOPs0.033 40.045 70.090 12 891.865 63 005.774 425 403.761 3
Params11 849 74010 898 48042 153 92815 400 48015 505 74627 813 130
), ArticleFig(id=1241113378408689764, tenantId=1146029695717560320, journalId=1238841944844054536, articleId=1241110428361740642, language=CN, label=表2, caption=

德国ESAR数据集分类准确度

, figureFileSmall=null, figureFileBig=null, tableContent=
FCNUnetTF-CVFCN模型1模型2CT-Unet
建筑区0.906 80.923 00.975 00.932 80.955 20.984 4
林区0.939 80.973 70.967 70.970 40.980 90.962 9
开放区0.981 40.989 00.986 00.992 50.988 10.991 2
OA0.954 90.969 60.979 80.973 40.978 50.984 2
MIou0.891 80.926 80.953 70.932 80.950 90.960 1
Kappa0.914 00.942 80.964 30.947 80.962 10.969 3
GFLOPs0.033 40.045 70.090 12 891.865 63 005.774 425 403.761 3
Params11 849 74010 898 48042 153 92815 400 48015 505 74627 813 130
)], attaches=null, journal=Journal(id=1238822691516493831, delFlag=0, nameCn=遥测遥控, nameEn=Journal of Telemetry, Tracking and Command, nameHistory1=null, nameHistory2=null, issn=2095-1000, eissn=null, cn=11-1780/TP, coden=null, periodic=1, language=CN, oaType=null, 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=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=uUqWBk5pV4VWPULNoAHkuQ==, journalPrice=null, startedYear=null, abbrevIsoEn=Journal of Telemetry, Tracking and Command, journalRemark=null, publicationField=null, createdTime=1773287945871, updatedTime=1773299995823, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=J, firstLetterEn=J, subjectCode=Engineering, subjectName=null, subjectCodeEn=Engineering, subjectNameEn=null, picCn=uUqWBk5pV4VWPULNoAHkuQ==, picEn=RDDslHHvDBDNnJMjk6BxdA==, jcr=null, cjcr=null, exts=[JournalExt(id=1238873232732909900, language=CN, name=遥测遥控, 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=1773299995835, updatedTime=1773299995835, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://manuscript.spacejournal.cn/ycyk, submissionEditorUrl=https://manuscript.spacejournal.cn/ycyk, submissionReviewUrl=https://manuscript.spacejournal.cn/ycyk, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1238873232779047245, language=EN, name=Journal of Telemetry, Tracking and Command, 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=1773299995846, updatedTime=1773299995846, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://manuscript.spacejournal.cn/ycyk, submissionEditorUrl=https://manuscript.spacejournal.cn/ycyk, submissionReviewUrl=https://manuscript.spacejournal.cn/ycyk, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1238841944844054536, websiteList=[Website(id=1238843830011744952, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1238841944844054536, 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/ycyk/CN, language=CN, createTime=1773292985680, createBy=18614031015, updateTime=1773293577303, updateBy=18614031015, name=遥测遥控-中文, tplId=1146099689490845704, title=遥测遥控, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1238846390617568150, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=articleTextType, value=kx, createTime=1773293596176, updateTime=1773293596176, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390596596627, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=banner, value=null, createTime=1773293596171, updateTime=1773293596171, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390638539673, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=grayFlag, value=0, createTime=1773293596181, updateTime=1773293596181, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390588208018, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=logo, value=https://castjournals.cast.org.cn/joweb/ycyk/CN/file/pic?fileId=g9S2GGWx7iZjf51o9vpmmA==, createTime=1773293596169, updateTime=1773293596169, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390651122587, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=minRunFlag, value=0, createTime=1773293596184, updateTime=1773293596184, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390613373845, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/ycyk/CN/file/pic, createTime=1773293596175, updateTime=1773293596175, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390646928282, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=silenceFlag, value=0, createTime=1773293596183, updateTime=1773293596183, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390604985236, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1773293596173, updateTime=1773293596173, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390625956759, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=themeColor, value=null, createTime=1773293596178, updateTime=1773293596178, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846390634345368, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830011744952, code=themeStyle, value=null, createTime=1773293596180, updateTime=1773293596180, creator=18614031015, updator=18614031015)]), Website(id=1238843830078853819, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1238841944844054536, 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/ycyk/EN, language=EN, createTime=1773292985696, createBy=18614031015, updateTime=1773293566698, updateBy=18614031015, name=遥测遥控-英文, tplId=1146101810881728533, title=Journal of Telemetry, Tracking and Command, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1238846419235304352, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=articleTextType, value=kx, createTime=1773293602999, updateTime=1773293602999, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419214332829, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=banner, value=null, createTime=1773293602994, updateTime=1773293602994, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419256275875, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=grayFlag, value=0, createTime=1773293603004, updateTime=1773293603004, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419205944220, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=logo, value=https://castjournals.cast.org.cn/joweb/ycyk/EN/file/pic?fileId=g9S2GGWx7iZjf51o9vpmmA==, createTime=1773293602992, updateTime=1773293602992, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419268858789, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=minRunFlag, value=0, createTime=1773293603007, updateTime=1773293603007, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419231110047, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/ycyk/EN/file/pic, createTime=1773293602998, updateTime=1773293602998, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419264664484, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=silenceFlag, value=0, createTime=1773293603006, updateTime=1773293603006, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419222721438, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1773293602996, updateTime=1773293602996, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419243692961, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=themeColor, value=null, createTime=1773293603001, updateTime=1773293603001, creator=18614031015, updator=18614031015), WebsiteProps(id=1238846419252081570, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1238843830078853819, code=themeStyle, value=null, createTime=1773293603003, updateTime=1773293603003, creator=18614031015, updator=18614031015)])], journalTitle=遥测遥控, weixinUrl=null, journalUrl=https://ycyk.spacejournal.cn/, iacademicId=null, status=1, seqNo=null, journalTitleEn=Journal of Telemetry, Tracking and Command, journalPhotoCn=uUqWBk5pV4VWPULNoAHkuQ==, journalPhotoEn=RDDslHHvDBDNnJMjk6BxdA==, 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/ycyk/CN/10.12347/j.ycyk.20240116002, detailUrlEn=https://castjournals.cast.org.cn/joweb/ycyk/EN/10.12347/j.ycyk.20240116002, pdfUrlCn=https://castjournals.cast.org.cn/joweb/ycyk/CN/PDF/10.12347/j.ycyk.20240116002, pdfUrlEn=https://castjournals.cast.org.cn/joweb/ycyk/EN/PDF/10.12347/j.ycyk.20240116002, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于复数域Transformer-Unet混合模型的PolSAR地物分类
收藏切换
PDF下载
谢雯 , 张嘉鹏 , 张哲哲 , 闪晨超
遥测遥控 | 人工智能技术 2024,45(3): 35-42
收起
收藏切换
遥测遥控 | 人工智能技术 2024, 45(3): 35-42
基于复数域Transformer-Unet混合模型的PolSAR地物分类
全屏
谢雯, 张嘉鹏, 张哲哲, 闪晨超
作者信息
  • 西安邮电大学通信与信息工程学院 西安 710121
  • 谢雯 1989年生,副教授,硕士生导师。

    张嘉鹏 1998年生,硕士研究生。

    张哲哲 1997年生,硕士研究生。

    闪晨超 1999年生,硕士研究生。

PolSAR Terrain Classification Based on Transformer-Unet Hybrid Model in the Complex Domain
Wen XIE, Jiapeng ZHANG, Zhezhe ZHANG, Chenchao SHAN
Affiliations
  • School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an 710121, China
出版时间: 2024-05-15 doi: 10.12347/j.ycyk.20240116002
文章导航
收藏切换

传统的基于深度学习的极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,PolSAR)地物分类方法,通过堆叠卷积层提取图像局部特征,难以建立长距离依赖关系。基于自注意力机制的深度学习模型Transformer(变换)在图像分类任务中取得了成功,其自注意力机制能够捕获全局像素之间的关联性,同时PolSAR地物分类任务被证实:相比于实数域,其在复数域上表现出更好的分类效果。因此,本文将Transformer引入到复数域中,提出了一种基于复数域的Transformer和Unet(语义分割网络)混合模型(CT-Unet)用于PolSAR地物分类,将Transformer与CNN相结合,对复数类型的PolSAR数据进行特征提取,使用西安数据集和德国数据集进行PolSAR地物分类的实验结果表明:提出的模型能够有效提高PolSAR地物分类的准确性,Transformer有望在PolSAR地物分类任务中弥补卷积神经网络的不足。

极化合成孔径雷达  /  复数域  /  Transformer  /  Unet

The traditional deep learning-based Polarimetric Synthetic Aperture Radar (PolSAR) feature classification method extracts image local features by stacking convolutional layers, which makes it difficult to establish long-range dependencies. It is not-ed that Transformer, a deep learning model based on a self-attention mechanism that captures global pixel-to-pixel correlations, has achieved success in image classification tasks. Meanwhile, the PolSAR feature classification task has demonstrated better classification results in the complex domain compared to the real domain. Therefore, Transformer is introduced into the complex domain, and a hybrid model of Transformer and Unet based on the complex domain (CT-Unet) is proposed for PolSAR feature classification. This model combines Transformer with CNN for feature extraction on PolSAR data of complex type. The experimental results of PolSAR feature classification using the Xi'an dataset and German dataset show that the proposed model can effectively improve the accuracy of PolSAR feature classification. Transformer is expected to make up for the shortcomings of convolutional neural net-works in the PolSAR feature classification task.

Polarimetric synthetic aperture radar(PolSAR)  /  Complex domain  /  Transformer  /  Unet
谢雯, 张嘉鹏, 张哲哲, 闪晨超. 基于复数域Transformer-Unet混合模型的PolSAR地物分类. 遥测遥控, 2024 , 45 (3) : 35 -42 . DOI: 10.12347/j.ycyk.20240116002
Wen XIE, Jiapeng ZHANG, Zhezhe ZHANG, Chenchao SHAN. PolSAR Terrain Classification Based on Transformer-Unet Hybrid Model in the Complex Domain[J]. Journal of Telemetry, Tracking and Command, 2024 , 45 (3) : 35 -42 . DOI: 10.12347/j.ycyk.20240116002
遥感卫星[1]技术发展已久,其中的极化合成孔径雷达(PolSAR)技术以其多极化、多频率和多角度的特点,能够提供更丰富和详细的地物表面信息,显著提高了地形分类的准确性,对城市规划[2]、农业监测[3]、军事探测[4]等相关领域具有重要意义和实际应用价值。通过深入研究PolSAR在地物分类中的应用,可以为提高遥感数据解译精度提供技术支持[5]。PolSAR地物分类方法经过不断发展大致可分为三类:第一类是基于PolSAR数据散射机制[6]的算法,通过分解PolSAR数据的相干矩阵、散射矩阵等得到有用的地物分类特征;第二类是基于统计分布[7]的算法,通常根据不同的分布对PolSAR数据进行模型的搭建,而传统的PolSAR地物分类方法通常需要从PolSAR数据中手工提取特征,这导致了模型的泛化性能受限等问题。随着深度学习技术的发展,第三类即基于深度学习的方法被应用在了PolSAR地物分类任务[8],卷积神经网络(Convolutional Neural Network,CNN)[9]的应用减轻了对手工设计特征的依赖,提高了在复杂场景中的地物分类性能。尽管该方法在学习Pol-SAR图像的空间和极化特征表示方面非常有效,仍受限于相位信息和感受野的丧失。为了解决上述问题并考虑到PolSAR地物分类属于像素级别的分类任务,全卷积神经网络(Fully Convolutional Network,FCN)[10]被引入到PolSAR地物分类领域,Adugna Mullissa等人[11]提出的模型PolSARNet,降低了计算复杂度的同时获得了令人满意的分类结果。尽管FCN可以实现对PolSAR图像的端到端像素级分类,但现有的方法中没有考虑到不同特征图之间存在的潜在关系。其变体U-Net[12]通过连接编码器和解码器部分的特征图,保留更多的低级语义信息并减少空间信息的丧失,而这对于Pol-SAR地物分类任务来说是非常重要的,但其中的卷积操作依旧具有局部性,容易受到感受野大小的限制。而Transformer[13]模型中的自注意力机制能够有效捕捉输入序列中各个位置的全局依赖关系,这意味着Transformer可以更好地保留远离中心像素的信息,从而提高对全局图像结构的理解能力,Dosovitskiy等[14]提出了视觉转换器(ViT)模型,凭借自注意力机制获取全局特征,有效弥补了CNN的局部性,可以很好地学习图像全局像素的依赖关系。陈辉等[15]提出了一种基于Swin Trans-former(采用平移窗口的分层视觉转换器)的多尺度混合光谱注意力网络(SMSaNet),该网络实现了不同特征区域之间对全局信息建立长距离依赖关系,有效提升了图像的分类准确率,但上述方法无法充分利用PolSAR数据的相位信息。Zhang等人[16]首次将PolSAR数据的复数域信息作为输入数据提出了复值CNN(Complex-valued CNN,CV-CNN)取得了较为满意的分类精度。Yu等人[17]在CV-CNN网络模型的基础上提出了CVFCNN网络模型,该网络模型具有比CV-CNN网络模型更好的分类性能,同时也验证了PolSAR数据的复数域中包含的相位信息的重要性。Xie等人[18]最近提出的TF-CVFCN(复值深度全卷积神经网络)在CV-FCN的基础上引入迁移学习取得了令人满意的分类效果,进一步说明了PolSAR地物分类在深度学习领域广阔的发展前景。
基于前述分析并受以前的工作启发,本文提出了一种PolSAR地物分类模型CT-Unet,该模型在复数域中将Transformer模型与Unet模型进行结合,从PolSAR图像中有效捕获全局、局部和多尺度的特征,从而提高分类性能。实验采用西安地区数据集和德国ESAR数据集在多个不同的模型上进行验证,所提出的模型为PolSAR地物分类任务提供了一种可行的方法。
本文提出的网络模型如图1所示。为了充分利用PolSAR数据的相位信息,将复数域信息作为网络模型的输入数据,在通道维度上将输入数据分割成实部与虚部,接下来整体网络模型也将对输入数据的实部与虚部分别进行处理。首先,对其进行多次3D卷积和2D卷积,将提取的特征输入到不同尺度的Transformer中,融合Transformer得到的特征图并重新转换作为解码器的输入,解码器包括特征融合块、上采样卷积块以及用于整合实部和虚部的模块。最后,通过重塑操作将输出的图像尺寸调整为与输入特征图的尺寸相同。
在PolSAR地物分类任务当中,网络模型利用相位信息能够提升其分类性能[19],复数域中则包含PolSAR数据的相位信息,因此本文将在复数域上完成网络模型的搭建,将卷积、池化及激活函数从实数域扩展到复数域中。首先将输入的数据与复数卷积核进行卷积,以下公式所示:
其中,w为复数卷积核,w=a+jbh为输入数据,h=x+jyR(w*h)表示卷积结果的实部,I(w*h)表示卷积结果的虚部。通过提取并结合输入数据和复数卷积核的实部与虚部进行卷积操作,再结合得到复数卷积的结果。池化操作能够有效减小训练参数,增加模型泛化能力[20],扩展到复数域上的池化操作以平均池化为例,其池化过程如下式所示:
AvePooling(h)=AvePooling(x)+jAvePooling(y)=R(h)+jI(h)
其中,h为输入数据,其定义为h=x+jyR(h)表示平均池化结果的实部,I(h)表示平均池化结果的虚部。通过对输入数据的实部和虚部分别进行平均池化操作,将结果进行组合得到复数平均池化的结果。激活函数使得神经网络能够学习和表示更为复杂的函数[21],本文将ReLU激活函数扩展到复数域,具体操作如下式所示:
ReLU(h)=ReLU(x)+jReLU(y)=R(h)+jI(h)
其中,R(h)表示激活函数结果的实部,I(h)表示激活函数结果的虚部。将输入数据的实部和虚部分别进行激活函数处理,将所得结果进行结合得到复数域激活函数的结果。
编码器部分位于图1的左侧,该部分包括卷积模块和Transformer模块,主要实现特征提取的功能,所提出的网络首先会将输入数据在通道维度分割成实部与虚部,此步骤旨在充分利用PolSAR数据中的相位信息。然后为了获取更加细致和精确的局部特征和全局特征,在Transformer模型之前利用卷积核大小不同的四层3D卷积和四层2D卷积提取输入数据的局部特征,该提取特征的方法能够实现在控制参数量的前提下,充分提取数据的空间信息及通道信息[22],然后将提取到的特征输入到多尺度的Transformer模型以提取全局特征,提取全局特征的过程则依赖于多头自注意力机制[23],该机制的过程如图2所示。
实现过程如式(4)~式(6)所示,其中注意力函数计算公式如下
其中,QKV分别是查询、键和值的矩阵,dk是键的维度。每个注意力头的计算公式如下
其中,是参数矩阵。多头自注意力计算公式如下
Multihead(Q,K,V)=Concat(h1,…,hh)WO
其中,WO是参数矩阵。多尺度Transformer部分一共由两个Transformer模块组成,这里的多尺度指的是输入Transfomrer模块的特征尺寸不同,第一个Transformer模块的输入来自第三层的2D卷积,第二个Transformer模块的输入来自第四层的2D卷积,其多头自注意力机制模块堆叠的次数不同,其中的自注意力头数也不同。第一个Trans-former模块中堆叠了八个多头自注意力机制模块,自注意力头数为6,第二个Transformer模块中堆叠了6个多头自注意力机制模块,自注意力头数为4,这样设计的编码器可在连续卷积的过程中尽可能地提取局部与全局信息。最终,经过编码器获得的特征图进行融合与重构后输入到解码器中。
解码器部分位于图1的右侧,该部分包括特征融合模块、上采样卷积模块、实部与虚部融合模块,主要实现上采样并得到最后的分类结果图。在解码器中,首先将输入的图像进行上采样,将其结果与编码器部分经跳跃连接传递的特征进行拼接,再进行卷积后得到的特征图继续上采样。该步骤通过获得编码阶段的浅层特征从而帮助模型更好地得到多尺度特征信息。重复上述步骤四次后将实部与虚部两个支路得到的特征图进行融合,并通过重构将其恢复到原始图像大小,该解码器部分的设计在保留图像语义信息的同时有效地恢复了空间细节。
本文采用西安数据集和德国数据集验证所提出模型的有效性。西安数据集是2010年由RADARSAT-2(星载合成孔径雷达)卫星系统[24]获取的中国西安西部渭河地区场景,图像大小为8 m×8 m,选取其分辨率为512×512的子图像块进行实验,该数据集共包括草地、城市和河流三种地形类别。德国数据集是由德国宇航中心ESAR(电子扫描阵列雷达)系统获取的关于Oberpfaffenhofen(奧伯法芬霍芬)地区的全极化雷达图像,图像大小为3 m×3 m,分辨率为1 300×1 200,该数据集共包括建筑区、开放区和林区三种地形类别。
将滑动窗口和步幅设置为128×128和6对西安地区图像进行处理,训练集和测试集的样本数量分别为125和516,模型训练时的批量大小设置为16,并使用1E-3作为学习率的Adam(自适应矩估计)优化器来优化目标函数,迭代次数设置为300次。所有实验均在配备TensorFlow(机器学习服务平台)2.10.0版本和NVIDIA GTX 4070ti GPU(英伟达显卡)的PC上运行。为了说明本文所提出模型的有效性,在相同的条件下分别测试了FCN(全卷积网络),Unet(语义分割网络)和TF-CVFCN模型,并进行了消融实验,其中模型1为Unet与单尺度的Transformer进行结合,模型2是在模型1的基础上加入了3D卷积,以及CT-Unet。在本文涉及的所有实验中,均使用整体准确率(OA)、KAPPA(卡帕)系数和平均交并比(MIoU)作为其分类效果的评价指标,对比实验的可视化结果如图3所示。每个类别具体的精度以及各个模型的参数量和运算量如表1所示。图3展示了不同模型在西安数据集上的分类结果图,绿色区域为草地,红色区域为城市,蓝色区域为河流。从图中可以看出,FCN及Unet取得的分类结果图比较连贯平滑,但还是存在较多的错误分类点,在一些细节方面分类情况不是很理想。TF-CVFCN的错误分类点较少,整体效果更加接近本文提出的模型。从模型1开始加入Transformer后,能够较为有效地提取全局特征,加深了网络模型对于特征的提取能力,所得的分类结果也有所改善。而多尺度Transformer的采用更进一步加深了网络模型对于PolSAR数据的理解能力,特别是黄色圆圈圈出的区域图像更加清晰且错分现象大大减少,说明了通过多头自注意力机制能够更好地捕捉PolSAR的全局特征,更全面地理解地物的分布,从而提高地物分类的精度。表1中罗列了不同模型在西安数据集上的分类准确度。通过表1可以看到,提出的模型在OA值、MIoU和Kappa系数分别为98.85%、97.89%和97.24%,相较于其他的模型都获得了最高的分数,其中OA值分别比FCN、Unet、TF-CVFCN、模型1及模型2提高了2.64%、2%、0.74%、1.77%、0.83%。MIoU分别提高了5.91%、4.03%、1.59%、3.19%、1.6%。Kappa分别提高了4.79%、3.23%、1.25%、2.51%、1.26%。在具体类别上包括草地、城市及河流也都取得了最高的准确度,所得到的结果图更加接近于真实标注图,同时由于Trans-former模块的加入也伴随着运算量的升高。
为了更好地验证CT-Unet网络模型的分类性能,使用德国ESAR数据集进行测试,将滑动窗口和步幅设置为128×128和31对德国ESAR图像进行处理,训练集和测试集的样本数量分别为296和5 727,模型训练时的批量大小设置为110,并使用1E-3作为学习率的Adam优化器来优化目标函数,迭代次数设置为200次。同样,分别测试了FCN,Unet和TF-CVFCN模型,并进行了消融实验,各个模型与西安地区地物分类实验相同。对比实验的可视化结果如图4所示,每个类别具体的精度及采用德国数据集后各个模型的运算量及参数量如表2所示。图4展示了不同模型在德国ESAR数据集上的分类结果图,建筑区为红色区域,林区为绿色区域,黄色为开放区区域[25]。从图中可以看到FCN及Unet取得的分类结果图相较于真实标注图存在较多的错误分类点,有部分将建筑区错分为林区的现象,TF-CVFCN对于建筑区的分类较为准确,但开放区错误分类的现象较多,综合分类效果要优于模型1和模型2。模型1加入了Trans-former后分类结果图准确性较FCN和Unet有所提升,开放区的错分现象大幅减少,也是所有本文提到的模型中在开放区类别上表现最好的模型。模型2在林区上的表现相较于模型1在准确率上进一步得到提升,绿色的林区区域错分现象较少。最后,本文提出的模型综合表现优于其他模型,各个类别的准确率从结果图来看均达到较好的水平。特别是图像中圈出的建筑区区域更加平滑且几乎没有错分现象,再次验证了多尺度Trans-former能够有效捕捉PolSAR图像的全局特征。
表2中罗列了不同模型在德国数据集上的分类准确度,通过表2可以看到所提出的模型在OA值、MIoU和Kappa系数分别为98.42%、96.01%和96.93%,相较于其他的模型都获得了最高的分数,其中OA值分别比FCN、Unet、TF-CVFCN、模型1及模型2提高了2.93%、1.46%、0.44%、1.08%和0.57%。MIoU分别提高了6.83%、3.33%、0.64%、2.73%和0.92%。Kappa分别提高了5.53%、2.65%、0.50%、2.15%和0.72%。而随着数据集的变化,运算量也会随之变化,在使用德国数据集后,各个模型的运算量均有增长。表2中的数据再次证实了所提出的模型能够提升PolSAR地物分类的精度。
本文提出了一种PolSAR地物分类模型CT-Unet,该模型实现了在复数域上将Transformer与Unet相结合进行PolSAR地物分类任务。为了充分提取PolSAR图像中的有效信息,将Transformer用于编码器部分来提取特征信息,可以在不丢失局部信息的同时,更好地整合全局信息。Unet则通过其编码器-解码器结构提供多尺度的特征,通过将二者结合有效捕捉到PolSAR图像局部、全局、多尺度的特征,从而提高PolSAR地物分类的精度。从实验结果也可以看出:所提出的模型分类效果比其他方法更加平滑和准确。但由于多头自注意力机制带来的计算量,本文提出的模型出现计算复杂及参数量膨胀的问题,未来考虑将在不影响分类精度的情况下实现模型的轻量化。
  • 国家自然科学基金(61901365)
  • 国家自然科学基金(62071379)
  • 陕西省自然科学基金(2019JQ-377)
  • 陕西省教育厅专项科研计划(19JK0805)
  • 西安邮电大学西邮新星团队项目(xyt2016-01)
  • 陕西高校青年创新团队
参考文献 引证文献
排序方式:
[1]
叶晖, 李博, 王金华, 等. 光学遥感卫星地面图像数据测试的局限及解决方法初探[J]. 遥测遥控, 2015, 36(5):62-67.
YE Hui, LI Bo, WANG Jinhua, et al. Preliminary exploration of a new method for image data testing of optical sat-ellites[J]. Journal of Telemetry, Tracking and Command, 2015, 36(5): 62-67.
[2]
肖雨彤, 张继贤, 黄国满, 等. FCN与CRF结合的Pol-SAR影像建筑区域提取[J]. 遥感信息, 2020, 35(3):44-49.
XIAO Yutong, ZHANG Jixian, HUANG Guoman, et al. Building area extraction of PolSAR images based on FCN and CRF[J]. Remote Sensing Information, 2020, 35(3): 44-49.
[3]
郭交, 尉鹏亮, 周正舒, 等. 基于时变特征的多时相Pol-SAR农作物分类方法[J]. 农业机械学报, 2017, 48(12):174-182.
GUO Jiao, WEI Pengliang, ZHOU Zhengshu, et al. Crop classification method with differential characteristics based on multi-temporal PolSAR images[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48(12): 174-182.
[4]
OH Y, HONG S Y, KIM Y, et al. Polarimetric backscattering coefficients of flooded rice fields at L-and C-bands:Measurements, modeling, and data analysis[J]. IEEE Transactions on Geoscience and Remote Sensing, 2009, 47(8): 2714-2721.
[5]
SHARIF M, ATTARCHI S. Estimation of hyrcanian forests coverage using ALOS PALSAR and MODIS satellite data (2007—2017)[J]. Advances in Space Research, 2024, 73(1): 426-439.
[6]
LIANG D, TANG T, BAN Y, et al. Unsupervised polari-metric SAR urban area classification based on model-based decomposition with cross scattering[J]. ISPRS Jour-nal of Photogrammetry and Remote Sensing, 2016, 116:86-100.
[7]
刘振. 小样本PolSAR图像地物分类技术研究[D]. 西安: 西安电子科技大学, 2018.
[8]
PARIKH H, PATEL S, PATEL V. Classification of SAR and PolSAR images using deep learning: A review[J]. In-ternational Journal of Image and Data Fusion, 2020, 11(1): 1-32.
[9]
CUI Y, LIU F, JIAO L, et al. Polarimetric multipath con-volutional neural network for PolSAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 1-18.
[10]
XIE W, JIAO L, HUA W. Complex-valued multi-scale fully convolutional network with stacked-dilated convo-lution for PolSAR image classification[J]. Remote Sensing, 2022, 14(15): 3737.
[11]
MULLISSA A G, PERSELLO C, STEIN A. PolSAR-Net: A deep fully convolutional network for polarimetric SAR image classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(12): 5300-5309.
[12]
HUANG H, LIN L, TONG R, et al. Unet 3+: A full-scale connected unet for medical image segmentation[C]//ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing(ICASSP). IEEE, 2020: 1055-1059.
[13]
WANG W, CHEN W, QIU Q, et al. Crossformer++: A versatile vision transformer hinging on cross-scale attention[J]. arXiv preprint arXiv:2303.06908, 2023.
[14]
DOSOVITSKIY A, BEYER L, KOLESNIKOV A,et al. An image is worth 16x16 words: Transformers for image recognition at scale[C]//International Conference on Learning Representations, 2021. arXiv:
[15]
陈辉, 张甜, 陈润斌. 基于轻量级卷积Transformer的图像分类方法及在遥感图像分类中的应用[J]. 电子与信息学报, 2022, 44: 1-9.
CHEN Hui, ZHANG Tian, CHEN Runbin. Image classi-fication method based on lightweight convolutional transformer and its application in remote sensing[J]. Journal of Electronics & Information Technology, 2022, 44: 1-9.
[16]
ZHANG Z, WANG H, XU F, et al. Complex-valued con-volutional neural network and its application in polari-metric SAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(12): 7177-7188.
[17]
YU L, HU Y, XIE X, et al. Complex-valued full convo-lutional neural network for SAR target classification[J]. IEEE Geoscience and Remote Sensing Letters, 2019(99): 1-5.
[18]
XIE W, SUN H, ZHANG Y, et al. Polsar image classifi-cation via transfer learning and fully convolutional net-work[C]//IGARSS 2023-2023 IEEE International Geo-science and Remote Sensing Symposium. IEEE, 2023:8026-8029.
[19]
秦先祥, 余旺盛, 王鹏, 等. 基于复值卷积神经网络样本精选的极化SAR图像弱监督分类方法[J]. 雷达学报, 2020, 9(3): 525-538.
QIN Xianxiang, YU Wangsheng, WANG Peng, et al. Weakly supervised classification of PolSAR images based on sample refinement with complex-valued convo-lutional neural network[J]. Journal of Radars, 2020, 9(3): 525-538.
[20]
江泽涛, 秦嘉奇, 张少钦. 参数池化卷积神经网络图像分类方法[J]. 电子学报, 2020, 48(9): 1729-1734.
JIANG Zetao, QIN Jiaqi, ZHANG Shaoqin. Parame-terized pooling convolution neural network for image classification[J]. Acta Electronica Sinica, 2020, 48(9):1729-1734.
[21]
杜圣杰, 贾晓芬, 黄友锐, 等. 用于CNN模型图像分类任务的高效激活函数[J]. 红外与激光工程, 2022, 51(3): 483-491.
DU Shengjie, JIA Xiaofen, HUANG Yourui, et al. High efficient activation function design for CNN model image classification task[J]. Infrared and Laser Engi-neering, 2022, 51(3): 483-491.
[22]
张腊梅, 陈泽茜, 邹斌. 基于3D卷积神经网络的Pol-SAR图像精细分类[J]. 红外与激光工程, 2018, 47(7):7-14.
ZHANG Lamei, CHEN Zexi, ZOU Bin. Fine classification of polarimetric SAR images based on 3D convolutional neural network[J]. Infrared and Laser Engineering, 2018, 47(7): 7-14.
[23]
HAN K, WANG Y, CHEN H, et al. A survey on vision transformer[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(1): 87-110.
[24]
LOU Y, KIM Y, ZYL J V, et al. The NASA/JPL airborne synthetic aperture radar's 1996 pacRim deployment[C]//International Geoscience and Remote Sensing Sympo-sium. IEEE, 1997.
[25]
VAN ZYL J, CARANDE R, LOU Y, et al. The NASA/JPL three-frequency polarimetric AIRSAR system[C]//IGARSS'92 International Geoscience and Remote Sensing Symposium. IEEE, 1992, 1: 649-651.
2024年第45卷第3期
PDF下载
91
42
引用本文
BibTeX
文章信息
doi: 10.12347/j.ycyk.20240116002
  • 接收时间:2024-01-16
  • 首发时间:2026-03-18
  • 出版时间:2024-05-15
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2024-01-16
  • 修回日期:2024-03-20
基金
国家自然科学基金(61901365)
国家自然科学基金(62071379)
陕西省自然科学基金(2019JQ-377)
陕西省教育厅专项科研计划(19JK0805)
西安邮电大学西邮新星团队项目(xyt2016-01)
陕西高校青年创新团队
作者信息
    西安邮电大学通信与信息工程学院 西安 710121
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/ycyk/CN/10.12347/j.ycyk.20240116002
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