Wavelet -assisted efficient Swin Transformer network for image dehazing
Rahul Vishnoirahulv.engineering@tmu.ac.inDepartment of Electronics & Communication (EC) EngineeringCollege of EngineeringTeerthanker Mahaveer UniversityMoradabad, Uttar Pradesh, 244001, IndiaView full profile → , *Alka VermaCorresponding authordralka.engineering@tmu.ac.inDepartment of Electronics & Communication (EC) EngineeringCollege of EngineeringTeerthanker Mahaveer UniversityMoradabad, Uttar Pradesh, 244001, IndiaView full profile → , Vibhor Kumar Bhardwajvibhor.engineering@tmu.ac.inDepartment of Electronics & Communication (EC) EngineeringCollege of EngineeringTeerthanker Mahaveer UniversityMoradabad, Uttar Pradesh, 244001, IndiaView full profile →
* Corresponding author · click or hover a name for details
- Received:
- 01 Dec 2025
- Published Online:
- 31 Jul 2026
- Article type:
- Research Article
- Language:
- EN
- Article no.:
- JIOS-2406
- Pages:
- 2837–2845
Abstract
Image dehazing increases the clarity of the visual aspect of a driver using a vehicle; when surveilling; and during consumer photographic use. Dehazing using CNN and the physics-based method has no method of understanding the entire global image scene; and both dehazing methods fail to perform in non-homogeneous haze setup, and finally transformer-based methods require either very large scale datasets or lots of processing power to accomplish their tasks. This research presents a unified framework for single-image dehazing named Physically Guided Wavelet Swin Network (PhyWave-Swin). PhyWave-Swin combines the necessary components of physical interpretability, multifrequency analysis and attention driven global modelling to advance this area of work. PhyWave-Swin was developed using the RESIDE dataset and underwent evaluation for both synthetic and real-world test data. The resulting performance metrics for PhyWave-Swin are 32.84/0.946 PSNR/SSIM respectively for SOTS-Indoor and 30.29/0.939 PSNR/SSIM respectively for SOTS-Outdoor test data; and PhyWave-Swin produces considerably better perceptual image quality versus other dehazing work that produces NIQE=3.65 and BRISQUE=33.9 in realworld scenarios. Therefore; a combination of physics, wavelet frequency representation and transformer-based global modelling provide the PhyWave-Swin with a very effective solution to single-image dehazing.
Keywords
Subject Classifications
References
[1] W. Wang and X. Yuan, “Recent advances in image dehazing,” IEEE/CAA Journal of Automatica Sinica, vol. 4, no. 3, pp. 410–436 (Jul. 2017).
[2] Y. Song, J. Li, X. Wang, and X. Chen, “Single image dehazing using ranking convolutional neural network,” IEEE Transactions on Multimedia, vol. 20, no. 6, pp. 1548–1560 (Jun. 2018).
[3] J. Dong and J. Pan, “Physics-based feature dehazing networks,” in Computer Vision – ECCV 2020 Workshops, Cham, Switzerland: Springer, pp. 188–204 (2020).
[4] Y. Wang, X. Yan, M. Wei, and J. Xiong, “USCFormer: Unified transformer with semantically contrastive learning for image dehazing,” IEEE Transactions on Intelligent Transportation Systems, vol. 24, no. 10, pp. 11321–11333 (Oct. 2023).
[5] Y. Song, X. Du, H. Qian, and Z. He, “Vision transformers for single image dehazing,” IEEE Transactions on Image Processing, vol. 32, pp. 1927–1941 (2023).
[6] Y. Yeh, L.-W. Kang, and C.-Y. Lin, “Haze effect removal from image via haze density estimation in optical model,” Optics Express, vol. 21, no. 22, pp. 27127–27141 (2013).
[7] K. He, J. Sun, and X. Tang, “Single image haze removal using dark channel prior,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 33, no. 12, pp. 2341–2353 (Dec. 2011).
[8] K. Ding and T. P. Chen, “Study on damage detection of bridge based on wavelet multiscale analysis,” Advanced Materials Research, vol. 639–640, pp. 1010–1014 (2013).
[9] T. Tao, X. Guan, H. Zhou, and H. Xu, “MCADNet: A multi-scale cross-attention network for remote sensing image dehazing,” Mathematics, vol. 12, no. 23, Art. no. 3650 (Nov. 2024).
[10] Z. Liu, Y. Lin, Y. Cao, H. Hu, Y. Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” arXiv preprint arXiv:2103.14030 (Mar. 2021).
[11] I. Mitov, B. Depaire, K. Vanhoof, and K. Ivanova, “Classifier PGN: Classification with high confidence rules,” Serdica Journal of Computing, vol. 7, no. 2, pp. 143–164 (2013).
[12] G. Chen, Y. Jia, Y. Yin, S. Fu, D. Liu, and T. Wang, “Remote sensing image dehazing using a wavelet-based generative adversarial network,” Scientific Reports, vol. 15, no. 1 (2025).
[13] S. Thakral and P. Manhas, “Image processing using different types of discrete wavelet transform,” in Proceedings of the International Conference on Advanced Computing, Communication and Informatics, Singapore: Springer, pp. 499–507 (2018).
[14] S. Zhang, L. Shen, W. Ren, S. Wan, and X. Zhang, “Wavelet-based physically guided normalization network for real-time traffic dehazing,” Pattern Recognition, vol. 172, Art. no. 112451 (Apr. 2026).
[15] L. Qi, H. Yang, X. Geng, and Y. Shi, “NormAUG: Normalization-guided augmentation for domain generalization,” IEEE Transactions on Image Processing, vol. 33, pp. 1419–1431 (2024).
[16] C. Ancuti, C. O. Ancuti, R. Timofte, and C. De Vleeschouwer, “I-HAZE: A dehazing benchmark with real hazy and haze-free indoor images,” in Proc. 19th Int. Conf. Advanced Concepts for Intelligent Vision Systems (ACIVS), Lecture Notes in Computer Science, vol. 11182, Cham, Switzerland: Springer, pp. 620–631 (2018), doi: 10.1007/978-3-030-01449-0_52.
[17] C. Ancuti, C. O. Ancuti, and C. De Vleeschouwer, “D-HAZY: A dataset to evaluate quantitatively dehazing algorithms,” in Proc. IEEE International Conference on Image Processing (ICIP), Phoenix, AZ, USA, pp. 2226–2230 (Sep. 2016).
[18] C. O. Ancuti, R. Timofte, and C. Ancuti, “NH-HAZE: An image dehazing benchmark with non-homogeneous hazy and haze-free images,” in Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pp. 444–453 (Jun. 2020).
[19] Z. Wang, X. Cun, J. Bao, and J. Liu, “Uformer: A general U-shaped transformer for image restoration,” in Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 17662–17672 (Jun. 2022).
