EUVIMEDOpdag EUVIMED
Open access · CC BY 4.0

Lightweight Asymmetric Convolutional Residual Network for Efficient Motion Image Deblurring

Hongyin Li, Yang Yue, Jiahao Li, Zhongning Guo iD, Ming Wu iD

Entropy · 2026-08-10 · 28 (8) · 896

DOI: 10.3390/e28080896 ↗

At a glance

Journal article in Entropy, published online on 2026-08-10. Written by Hongyin Li and 4 co-authors. Institution: School of AI Engineering, Guangzhou College of Technology and Business, Guangzhou 510006, China. 48 references. The full text is freely available (open access, CC BY 4.0).

AbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Motion image deblurring remains challenging because many existing models rely on complex architectures, leading to high computational cost and parameter redundancy, particularly under non-uniform blur in real-world scenes. To mitigate these limitations, we propose a Lightweight Asymmetric Convolutional Residual Network (LACR) for efficient motion image deblurring. LACR introduces an asymmetric convolutional residual module that combines local spatial embedding with horizontal and vertical asymmetric refinement, enabling direction-sensitive blur modeling with reduced spatial redundancy. A shallow deep feature fusion mechanism is further designed to integrate low-level convolutional cues with deep restoration representations, thereby complementing low-frequency structural information with high-frequency texture details. Experiments on four benchmark datasets show that LACR improves the reconstruction of edges, textures, and structural details while maintaining a lightweight design. Compared with representative lightweight deblurring methods under consistent evaluation settings, LACR achieves an average PSNR gain of 0.38 dB and reduces computational cost by more than 20%. Quantitative and qualitative results demonstrate that LACR achieves a favorable balance between restoration quality and computational efficiency.

Subjects

Author keywords

asymmetric convolutionlightweight architecturemotion image deblurring

Authors and institutions

  • School of AI Engineering, Guangzhou College of Technology and Business, Guangzhou 510006, China.
  • School of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, China.
  • College of Mechanical and Electrical Engineering, Guangdong University of Technology, Guangzhou 510006, China.
  • Department of Computer Science, KU Leuven, 3000 Leuven, Belgium.
  • Department of Mechanical Engineering, KU Leuven, 3000 Leuven, Belgium.

ORCID profiles: Zhongning Guo, Ming Wu

Cite

APA

Li, H., Yue, Y., Li, J., Guo, Z., & Wu, M. (2026). Lightweight Asymmetric Convolutional Residual Network for Efficient Motion Image Deblurring. Entropy, 28(8), 896. https://doi.org/10.3390/e28080896

Vancouver

Li H, Yue Y, Li J, Guo Z, Wu M. Lightweight Asymmetric Convolutional Residual Network for Efficient Motion Image Deblurring. Entropy. 2026;28(8):896. doi:10.3390/e28080896

BibTeX
@article{Li2026,
  author = {Li, Hongyin and Yue, Yang and Li, Jiahao and Guo, Zhongning and Wu, Ming},
  title = {Lightweight Asymmetric Convolutional Residual Network for Efficient Motion Image Deblurring},
  journal = {Entropy},
  year = {2026},
  volume = {28},
  number = {8},
  pages = {896},
  doi = {10.3390/e28080896}
}
Start for free →