Zur Kurzanzeige

dc.contributor.author
Qin, Feiwei
dc.contributor.author
Yan, Kang
dc.contributor.author
Wang, Changmiao
dc.contributor.author
Ge, Ruiquan
dc.contributor.author
Peng, Yong
dc.contributor.author
Zhang, Kai
dc.date.accessioned
2024-08-21T07:57:52Z
dc.date.available
2024-02-19T06:49:27Z
dc.date.available
2024-02-19T17:46:18Z
dc.date.available
2024-08-21T07:57:52Z
dc.date.issued
2024-08
dc.identifier.issn
1380-7501
dc.identifier.issn
1573-7721
dc.identifier.other
10.1007/s11042-024-18409-3
en_US
dc.identifier.uri
http://hdl.handle.net/20.500.11850/660103
dc.description.abstract
Given the broad application of infrared technology across diverse fields, there is an increasing emphasis on investigating super-resolution techniques for infrared images within the realm of deep learning. Despite the impressive results of current Transformer-based methods in image super-resolution tasks, their reliance on the self-attention mechanism intrinsic to the Transformer architecture results in images being treated as one-dimensional sequences, thereby neglecting their inherent two-dimensional structure. Moreover, infrared images exhibit a uniform pixel distribution and a limited gradient range, posing challenges for the model to capture effective feature information. Consequently, we suggest a potent Transformer model, termed Large Kernel Transformer (LKFormer), to address this issue. Specifically, we have designed a Large Kernel Residual Depth-wise Convolutional Attention (LKRDA) module with linear complexity. This mainly employs depth-wise convolution with large kernels to execute non-local feature modeling, thereby substituting the standard self-attention layer. Additionally, we have devised a novel feed-forward network structure called Gated-Pixel Feed-Forward Network (GPFN) to augment the LKFormer's capacity to manage the information flow within the network. Comprehensive experimental results reveal that our method surpasses the most advanced techniques available, using fewer parameters and yielding considerably superior performance.
en_US
dc.language.iso
en
en_US
dc.publisher
Springer
en_US
dc.subject
Infrared image
en_US
dc.subject
Super-resolution
en_US
dc.subject
Deep learning
en_US
dc.subject
Large kernel convolution
en_US
dc.title
LKFormer: large kernel transformer for infrared image super-resolution
en_US
dc.type
Journal Article
dc.date.published
2024-02-06
ethz.journal.title
Multimedia Tools and Applications
ethz.journal.volume
83
en_US
ethz.journal.issue
28
en_US
ethz.journal.abbreviated
Multimed Tools Appl
ethz.pages.start
72063
en_US
ethz.pages.end
72077
en_US
ethz.identifier.wos
ethz.identifier.scopus
ethz.publication.status
published
en_US
ethz.date.deposited
2024-02-19T06:49:34Z
ethz.source
WOS
ethz.eth
yes
en_US
ethz.availability
Metadata only
en_US
ethz.rosetta.installDate
2024-08-21T07:57:53Z
ethz.rosetta.lastUpdated
2024-08-21T07:57:53Z
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=LKFormer:%20large%20kernel%20transformer%20for%20infrared%20image%20super-resolution&rft.jtitle=Multimedia%20Tools%20and%20Applications&rft.date=2024-08&rft.volume=83&rft.issue=28&rft.spage=72063&rft.epage=72077&rft.issn=1380-7501&1573-7721&rft.au=Qin,%20Feiwei&Yan,%20Kang&Wang,%20Changmiao&Ge,%20Ruiquan&Peng,%20Yong&rft.genre=article&rft_id=info:doi/10.1007/s11042-024-18409-3&
 Printexemplar via ETH-Bibliothek suchen

Dateien zu diesem Eintrag

DateienGrößeFormatIm Viewer öffnen

Zu diesem Eintrag gibt es keine Dateien.

Publikationstyp

Zur Kurzanzeige