A Lightweight Image Super-resolution Feature Enhancement Method Based on Channel Attention Fusion

DOI: https://doi.org/10.62517/jike.202504412

Author(s)

Cheng Xu

Affiliation(s)

School of Nanjing Tech University, Nanjing, China *Corresponding Author

Abstract

This paper focuses on the core contradiction between model lightweighting and feature enhancement in the field of image super-resolution and proposes a lightweight feature enhancement method based on channel attention fusion. By constructing a dynamic weight distribution mechanism and a multi-scale feature interaction framework, high-frequency detail reconstruction is achieved while maintaining the lightweight of the model. Theoretical analysis shows that this method effectively solves the problems of feature redundancy and detail loss in traditional lightweight models through inter-channel correlation modeling and local feature enhancement. The application scenarios cover fields such as mobile visual enhancement, medical image processing, and remote sensing image analysis, providing theoretical support and practical paths for real-time super-resolution reconstruction.

Keywords

Image Super-Resolution; Lightweight Feature Enhancement; Channel Attention Fusion; Application Scenarios

References

[1] Zhang, M., Sun, X., Zhu, Q., & Zheng, G. (2021, July). A survey of hyperspectral image super-resolution technology. In 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS (pp. 4476-4479). IEEE. [2] Zuo, X., Liu, H., Liu, M., Li, D., Wang, K., Li, P., & Wang, R. (2025). Super-Resolution Image Reconstruction based on Random-coupled Neural Network and EDSR. Signal, Image and Video Processing, 19(10), 1-9. [3] Ledig, C., Theis, L., Huszár, F., Caballero, J., Cunningham, A., Acosta, A., ... & Shi, W. (2017). Photo-realistic single image super-resolution using a generative adversarial network. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 4681-4690). [4] Dong, C., Loy, C. C., He, K., & Tang, X. (2015). Image super-resolution using deep convolutional networks. IEEE transactions on pattern analysis and machine intelligence, 38(2), 295-307. [5] Shi, C., Cen, Y., & Cen, G. (2025). Feature-Separated Lightweight Model for Road Extraction in Wild Environments. IEEE Access. [6] Jin, X., Xie, Y., Wei, X. S., Zhao, B. R., Chen, Z. M., & Tan, X. (2022). Delving deep into spatial pooling for squeeze-and-excitation networks. Pattern Recognition, 121, 108159. [7] Wang, W., Ding, J., Shu, X., Xu, W., & Wu, Y. (2025). FFAE-UNet: an efficient pear leaf disease segmentation network based on U-shaped architecture. Sensors, 25(6), 1751. [8] Luo, G., Zhou, Y., Ji, R., Sun, X., Su, J., Lin, C. W., & Tian, Q. (2020, October). Cascade grouped attention network for referring expression segmentation. In Proceedings of the 28th ACM international conference on multimedia (pp. 1274-1282).