Interactive image dehazing network based on dynamic frequency-domain modulation

doi:10.3785/j.issn.1008-973X.2026.06.009

Journal of ZheJiang University (Engineering Science)

2026, Vol. 60

Issue (6): 1221-1230 DOI: 10.3785/j.issn.1008-973X.2026.06.009

Interactive image dehazing network based on dynamic frequency-domain modulation

Yan YANG(

),Xinyu SONG

School of Electronic and Information Engineering, Lanzhou Jiao Tong University, Lanzhou 730070, China

Download:

HTML

PDF(7783KB) HTML
Export: BibTeX | EndNote (RIS)

Abstract

An interactive dual-branch image dehazing network based on dynamic frequency-domain modulation was proposed in order to address the limitation of existing image dehazing method in diverse and complex haze scene. A parallel encoder composed of a global semantic modeling branch and a residual detail modeling branch was constructed to capture global semantic information and local texture feature, respectively. An adaptive cross fusion module was introduced to enable dynamic interaction between cross-branch feature and enhance cross-feature collaboration capability. A dynamic frequency-domain enhancement module was designed to strengthen the response of the model to high-frequency detail and complex haze region. An edge-guided auxiliary supervision mechanism was introduced in the decoder, which formed complementary constraint with frequency-domain enhancement to guide the network to focus on the image contour. Then detail restoration and visual clarity were improved. The experimental results on the RESIDE, NH-HAZE, O-HAZE and I-HAZE datasets demonstrate that the proposed method achieves stronger structure restoration capability and better visual consistency. PSNR and SSIM reached 24.93 dB and 0.812 6 on the I-HAZE dataset, respectively, which were improved by 2.02 dB and 0.0496 compared with the second-best method.

Key words： image dehazing dual-branch architecture feature interaction frequency-domain modulation edge guidance

Received: 29 July 2025 Published: 06 May 2026

CLC:

TP 391

Fund: 国家自然科学基金资助项目（62063014）；甘肃省高等学校产业支撑计划资助项目（2021CYZC-04）；甘肃省优秀研究生“创新之星”项目（2025CXZX-681）.

	Service
	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Yan YANG
	Xinyu SONG

Cite this article:

Yan YANG,Xinyu SONG. Interactive image dehazing network based on dynamic frequency-domain modulation. Journal of ZheJiang University (Engineering Science), 2026, 60(6): 1221-1230.

URL:

https://www.zjujournals.com/eng/10.3785/j.issn.1008-973X.2026.06.009 OR https://www.zjujournals.com/eng/Y2026/V60/I6/1221

基于动态频域调制的交互式图像去雾网络

针对现有图像去雾方法在多样性复杂雾气场景中的不足，提出基于动态频域调制的交互式双分支图像去雾网络. 构建由全局语义建模分支与残差细节建模分支组成的并行编码器，分别捕捉图像的全局语义信息与局部纹理特征，提出自适应交叉融合模块，实现跨分支特征的动态交互，提升特征协同能力. 设计动态频域增强模块，强化模型对高频细节与复杂雾气区域的响应能力. 在解码器中引入边缘辅助监督，与频域增强形成互补约束，引导网络关注图像轮廓，提升细节恢复能力及视觉清晰度. 在RESIDE、NH-HAZE、O-HAZE和I-HAZE数据集上的实验结果表明，所提方法具备更强的结构还原能力与视觉一致性，在I-HAZE数据集上，PSNR和SSIM分别达到24.93 dB和0.812 6，较次优方法分别提升了2.02 dB和0.049 6.

关键词： 图像去雾, 双分支结构, 特征交互, 频域调制, 边缘引导

Fig.1 Overall architecture of interactive image dehazing network based on dynamic frequency-domain modulation

Fig.2 Dynamic frequency domain enhancement module

Fig.3 Multi-scale feature enhanced encoding block

Fig.4 Adaptive cross fusion module

Fig.5 Multi-scale channel attention decoder block

Fig.6 Defogging result of different method on SOTS-outdoor dataset

Fig.7 Defogging result of different method on NH-HAZE dataset

Fig.8 Defogging result of different method on O-HAZE dataset

Fig.9 Defogging result of different method on I-HAZE dataset

Fig.10 Defogging result of different method on real image

Tab.1 PSNR and SSIM result of different dehazing algorithm on different dataset

Tab.2 Objective index comparison of ablation experiment

Fig.11 Subjective visualization comparison of ablation experiment on O-HAZE dataset


[1]	孟小哲, 冯钰新, 苏卓, 等基于不变学习的真实雾霾去除方法[J]. 浙江大学学报: 工学版, 2024, 58 (2): 268- 278 MENG Xiaozhe, FENG Yuxin, SU Zhuo, et al Real-world dehazing method with invariant learning[J]. Journal of Zhejiang University: Engineering Science, 2024, 58 (2): 268- 278

[2]	LI Y, LING Q, AN Y, et al DHC-net: a remote sensing object detection under haze and class imbalance[J]. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 4703016

[3]	THANGARASU G, KANNAN K N, RAMAMOORTHY M, et al. Artificial intelligence-driven image dehazing using deep convolutional neural networks for enhanced satellite imagery perception [C]//Proceedings of the IEEE 15th Symposium on Computer Applications and Industrial Electronics. Penang: IEEE, 2025: 1–6.

[4]	SI Y, CHEN L, ZHENG Z, et al A novel algorithm of haze identification based on FY3D/MERSI-II remote sensing data[J]. Remote Sensing, 2023, 15 (2): 438 doi: 10.3390/rs15020438

[5]	HOWARD J N Scattering phenomena: optics of the atmosphere. scattering by molecules and particles[J]. Science, 1977, 196 (4294): 1084- 1085

[6]	HE K, SUN J, TANG X Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33 (12): 2341- 2353 doi: 10.1109/TPAMI.2010.168

[7]	NANDINI B M, KAULGUD N. Wavelet-based method for enhancing the visibility of hazy images using color attenuation prior [C]//International Conference on Recent Trends in Electronics and Communication. Mysore: IEEE, 2023: 1–6.

[8]	CAI B, XU X, JIA K, et al DehazeNet: an end-to-end system for single image haze removal[J]. IEEE Transactions on Image Processing, 2016, 25 (11): 5187- 5198 doi: 10.1109/TIP.2016.2598681

[9]	REN W, LIU S, ZHANG H, et al. Single image dehazing via multi-scale convolutional neural networks [C]// European Conference on Computer Vision. Amsterdam: Springer, 2016: 154–169.

[10]	LI B, PENG X, WANG Z, et al. Aod-net: all-in-one dehazing network [C]// IEEE International Conference on Computer Vision. Venice: IEEE, 2017: 4770–4778.

[11]	LIU X, MA Y, SHI Z, et al. Griddehazenet: attention-based multi-scale network for image dehazing [C]// IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 7314–7323.

[12]	HU J, SHEN L, SUN G. Squeeze-and-excitation networks [C]// IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 7132–7141.

[13]	QIN X, WANG Z, BAI Y, et al. FFA-Net: feature fusion attention network for single image dehazing [C]// AAAI Conference on Artificial Intelligence. New York: AAAI Press, 2020: 11908–11915.

[14]	SONG Y, HE Z, QIAN H, et al Vision transformers for single image dehazing[J]. IEEE Transactions on Image Processing, 2023, 32: 1927- 1941 doi: 10.1109/TIP.2023.3256763

[15]	MOU C, WANG Q, ZHANG J. Deep generalized unfolding networks for image restoration [C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 17378–17389.

[16]	FU M, LIU H, YU Y, et al. Dw-gan: a discrete wavelet transform gan for nonhomogeneous dehazing [C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. [S. l.]: IEEE, 2021: 203–212.

[17]	WANG Z, CUN X, BAO J, et al. Uformer: a general U-shaped transformer for image restoration [C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 17662–17672.

[18]	XU J, CHEN Z X, LUO H, et al An efficient dehazing algorithm based on the fusion of transformer and convolutional neural network[J]. Sensors, 2022, 23 (1): 43 doi: 10.3390/s23010043

[19]	LIU Z, LIN Y, CAO Y, et al. Swin transformer: hierarchical vision transformer using shifted windows [C]//IEEE/CVF International Conference on Computer Vision. Montreal: IEEE, 2021: 9992–10002.

[20]	CHEN L C, PAPANDREOU G, KOKKINOS I, et al DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 40 (4): 834- 848

[21]	ZAMIR S W, ARORA A, KHAN S, et al. Restormer: efficient transformer for high-resolution image restoration [C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 5728–5739.

[22]	LAI W S, HUANG J B, AHUJA N, et al. Deep Laplacian pyramid networks for fast and accurate super-resolution [C]//IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 5835–5843.

[23]	ZAMIR S W, ARORA A, KHAN S, et al. Multi-stage progressive image restoration [C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 14821–14831.

[24]	PASZKE A, GROSS S, MASSA F, et al. Pytorch: an imperative style, high-performance deep learning library [C]// Advances in Neural Information Processing Systems. Vancouver: Curran Associates, 2019: 8024–8035.

[25]	LI B, REN W, FU D, et al Benchmarking single-image dehazing and beyond[J]. IEEE Transactions on Image Processing, 2018, 28 (1): 492- 505

[26]	ANCUTI C O, ANCUTI C, TIMOFTE R. NH-HAZE: an image dehazing benchmark with non-homogeneous hazy and haze-free images [C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. Seattle: IEEE, 2020: 444–445.

[27]	ANCUTI C O, ANCUTI C, TIMOFTE R, et al. O-haze: a dehazing benchmark with real hazy and haze-free outdoor images [C]// IEEE Conference on Computer Vision and Pattern Recognition Workshops. Salt Lake: IEEE, 2018: 754–762.

[28]	ANCUTI C, ANCUTI C O, TIMOFTE R, et al. I-HAZE: a dehazing benchmark with real hazy and haze-free indoor images [C]// International Conference on Advanced Concepts for Intelligent Vision Systems. Poitiers: Springer, 2018: 620–631.

[29]	CHEN D, HE M, FAN Q, et al. Gated context aggregation network for image dehazing and deraining [C]// IEEE Winter Conference on Applications of Computer Vision. Waikoloa: IEEE, 2019: 1375–1383.

[30]	ZHENG Y, ZHAN J, HE S, et al. Curricular contrastive regularization for physics-aware single image dehazing [C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver: IEEE, 2023: 5785–5794.

[31]	CHEN Z, HE Z, LU Z M DEA-Net: single image dehazing based on detail-enhanced convolution and content-guided attention[J]. IEEE Transactions on Image Processing, 2024, 33: 1002- 1015 doi: 10.1109/TIP.2024.3354108

[1]	Yinqing HUANG,Li ZENG. Pixel-labeling-based parameter-adaptive dehazing algorithm for aerial image[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(3): 556-564.

[2]	Yishan LIN,Jing ZUO,Shuhua LU. Multimodal sentiment analysis based on multi-head self-attention mechanism and MLP-Interactor[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(8): 1653-1661.

[3]	Yan YANG,Lipeng CHAO. A two-branch feature joint dehazing network based on multidimensional collaborative attention[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(6): 1119-1129.

[4]	Dejun ZHANG,Yanzi BAI,Feng CAO,Yiqi WU,Zhanya XU. Point cloud Transformer adapter for dense prediction task[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(5): 920-928.

[5]	Liming LIANG,Pengwei LONG,Jiaxin JIN,Renjie LI,Lu ZENG. Steel surface defect detection algorithm based on improved YOLOv8s[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(3): 512-522.

[6]	Yan YANG,Cunpeng JIA. An efficient image dehazing algorithm with Agent Attention for domain feature interaction[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(12): 2527-2538.

[7]	Xiaozhe MENG,Yuxin FENG,Zhuo SU,Fan ZHOU. Real-world dehazing method with invariant learning[J]. Journal of ZheJiang University (Engineering Science), 2024, 58(2): 268-278.

Viewed

Full text

Abstract

Cited

Shared

Discussed