Research Progress in Image De Fogging Technology and Comparison of SOTA Methods (2025-2026)

Xue Zhang; Bin Wu; YangJie Qin

doi:10.54691/j52jas56

Authors

Xue Zhang School of Big Data and Artificial Intelligence, Anhui Xinhua University, Hefei 230088, China
Bin Wu School of Big Data and Artificial Intelligence, Anhui Xinhua University, Hefei 230088, China
YangJie Qin School of Big Data and Artificial Intelligence, Anhui Xinhua University, Hefei 230088, China

DOI:

https://doi.org/10.54691/j52jas56

Keywords:

Image Dehazing; Deep Learning; Complex Scenarios; Core Techniques.

Abstract

Through in-depth research on the complex optical characteristics of foggy atmospheres and their effects on images, an image dehazing technology for suppressing and restoring foggy atmospheric scattering in images has been proposed, achieving better restoration results for foggy images. This improves the authenticity of images in fields such as autonomous driving, remote sensing imaging, and surveillance security, providing reliable image data support for these areas. Based on the explosive development of deep learning in recent years, image dehazing technology has gradually evolved, shifting from traditional physics-based model methods to hybrid architectures that combine Transformers, diffusion models, and physical priors, with significantly improved dehazing performance in various complex scenarios. By systematically reviewing the latest research results in the field of image dehazing for 2025-2026 and conducting an in-depth analysis of the current five state-of-the-art core techniques, including PDFormer, DiffDehaze, HazeGen, DehazeXL, HistoFusionNet, and multi-module collaborative U-Net, this study covers both the theoretical aspects and practical comparisons of these representative image dehazing methods. It further reveals the development trends of image dehazing techniques and their profound significance for the future development of image dehazing.

Downloads

Download data is not yet available.

References

[1] Jiang, J., Han, C., Zhao, W. X., & Wang, J. (2023). PDFormer: Propagation Delay-Aware Dynamic Long-Range Transformer for traffic flow prediction. Proceedings of the AAAI Conference on Artificial Intelligence, 37(4), 4365–4373.

[2] Li, J., Niu, J., Pang, Y., et al. (2025). Diff-HazeNet: enhancing real-world image dehazing via diffusion models in complex conditions. The Visual Computer, 41, 12009–12024.

[3] Wang, R., et al. (2025). Learning hazing to dehazing: Towards realistic haze generation for real-world image dehazing. In 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 23091–23100). Nashville, TN, USA.

[4] Chen, J., Yan, X., Xu, Q., et al. (2025). Tokenize image patches: Global context fusion for effective haze removal in large images.In 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2258-2268.

[5] Heydari, M. S., Dong, W., Shirani, S., Chen, J., & Zhou, H. (2026). HistoFusionNet: Histogram-guided fusion and frequency-adaptive refinement for nighttime image dehazing. ArXiv abs/2604.03800 (2026):1914-1924.

[6] Falk, T., Mai, D., Bensch, R., et al. (2019). U-Net: deep learning for cell counting, detection, and morphometry. Nature Methods, 16, 67-70.

[7] Wang, C.-Y., Bochkovskiy, A., & Liao, H.-Y. M. (2023). YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7464–7475). IEEE.

[8] Yang, C., Shao, L., Deng, Y., et al. (2025). SwinCNet leveraging Swin Transformer V2 and CNN for precise color correction and detail enhancement in underwater image restoration. Frontiers in Marine Science, 12(12), Article 1523729.

[9] Liu, X., Liu, J., Tang, J., & Wu, G. (2025). CATANet: Efficient content-aware token aggregation for lightweight image super-resolution. In 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 17902–17912). IEEE.