EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Dual-Branch Multi-Perspective Modulation Network for Efficient Infrared Image Super-Resolution

Zepeng Liu iD, Duanyang Zhang, Ruimin Qi, Guodong Zhang, Yizong Wang iD, Lizhi Liu

Sensors · 04.08.2026 · 26 (15) · 4928

DOI: 10.3390/s26154928 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 04.08.2026. Verfasst von Zepeng Liu und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Information Engineering, Xinjiang Institute of Technology, Akesu 843100, China. Die Arbeit nennt 66 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: "Tianchi Talents Attraction Project" of Xinjiang Uygur Autonomous Region, Heilongjiang Post-doctoral Foundation, “Tianchi Talents Attraction Project” of Xinjiang Uygur Autonomous Region.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Transformer-based super-resolution methods achieve competitive performance by capturing non-local information through self-attention. Nevertheless, the computation of the self-attention introduces heavy computational overhead, and its inherent low-pass characteristic restricts the learning of local details. To address these problems, we propose an effective dual-branch multi-perspective modulation network (DMMN) for efficient infrared image super-resolution. Specifically, we design a multi-scale feature modulation enhancement unit (MFMEU) to capture cross-scale spatial features and a frequency-domain cross-correlation patch modulation unit (FCPMU) to explore global feature representations. We further develop an efficient bidirectional cross modulation unit (BCMU) to promote feature interaction between outputs of MFMEU and FCPMU. Extensive experimental results verify that DMMN achieves a competitive trade-off between reconstruction accuracy and computational efficiency. For instance, compared with the ×4 SRFormer-light, the proposed DMMN obtains an average gain of 0.08 dB in PSNR across five public test datasets, runs 2.7× faster, and uses only about 24% of the FLOPs.

Schlagwörter

Stichwörter der Autor:innen

cross modulationinfrared image super-resolutionmulti-perspectiveself-attention

Autor:innen und Institutionen

  • School of Information Engineering, Xinjiang Institute of Technology, Akesu 843100, China.
  • School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu 611730, China.
  • School of Electrical Engineering, Liaoning University of Technology, Jinzhou 121001, China.

ORCID-Profile: Zepeng Liu, Yizong Wang

Förderung

  • "Tianchi Talents Attraction Project" of Xinjiang Uygur Autonomous Region
  • Heilongjiang Post-doctoral Foundation
  • “Tianchi Talents Attraction Project” of Xinjiang Uygur Autonomous Region

Zitieren

APA

Liu, Z., Zhang, D., Qi, R., Zhang, G., Wang, Y., & Liu, L. (2026). Dual-Branch Multi-Perspective Modulation Network for Efficient Infrared Image Super-Resolution. Sensors, 26(15), 4928. https://doi.org/10.3390/s26154928

Vancouver

Liu Z, Zhang D, Qi R, Zhang G, Wang Y, Liu L. Dual-Branch Multi-Perspective Modulation Network for Efficient Infrared Image Super-Resolution. Sensors. 2026;26(15):4928. doi:10.3390/s26154928

BibTeX
@article{Liu2026,
  author = {Liu, Zepeng and Zhang, Duanyang and Qi, Ruimin and Zhang, Guodong and Wang, Yizong and Liu, Lizhi},
  title = {Dual-Branch Multi-Perspective Modulation Network for Efficient Infrared Image Super-Resolution},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {15},
  pages = {4928},
  doi = {10.3390/s26154928}
}
Kostenlos starten →