针对现有算法处理森林雾天图像时普遍存在的去雾不彻底、暗区细节丢失及色彩失真等问题，提出一种基于特征融合注意力与对比学习的自适应森林图像去雾算法。具体而言，首先设计了一种多尺度特征融合注意力机制，通过联合通道与空间注意力动态调节特征响应，增强重要特征的表达能力；其次构建了局部对比正则模块，提升模型对暗区与远景区域雾浓度变化的判别能力；此外，引入自适应色彩校正模块，有效缓解色彩失真问题。在合成森林雾图与真实森林雾图数据集上的实验结果表明，所提算法在多项评价指标上均优于现有方法，峰值信噪比与结构相似度显著提升，自然图像质量评估值降低，展现出良好的鲁棒性和泛化能力。
Current algorithms often struggle with incomplete dehazing, loss of details in dark areas, and color distortion in forest foggy images. To address these issues, this paper proposed an adaptive forest image dehazing algorithm based on feature fusion attention and contrastive learning. A multi-scale feature fusion attention mechanism was designed, which dynamically adjusted feature responses by combining channel and spatial attention, thereby enhancing the representation capability of important features. A local contrast regularization module was constructed to enhance the ability of the model to discriminate variations in fog concentration in dark and distant areas. Furthermore, an adaptive color correction module was introduced to mitigate color distortion. Experimental results on both synthetic and real-world forest foggy image datasets demonstrate that the proposed algorithm outperforms existing methods, achieving significant improvements in peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) and a reduction in natural image quality evaluator (NIQE), and exhibiting strong robustness and generalization ability.
[1] Zhu R, Wang L J. Improved wavelet transform algorithm for single image dehazing [J]. Optik, 2014, 125(13): 3064-3066.
[2] 陆松浩. 单幅图像去雾算法研究[D]. 南京: 南京邮电大学, 2023.
[3] 李波, 胡红萍, 杨正民. 基于超像素分割的暗通道先验图像去雾算法[J]. 测试技术学报, 2025, 39(4): 415-423. Li B, Hu H P, Yang Z M. Dark channel prior image dehazing algorithm based on superpixel segmentation [J]. Journal of Test and Measurement Technology, 2025, 39(4): 415-423. (in Chinese)
[4] Li P Y, Tian J D, Tang Y D, et al. Deep retinex network for single image dehazing [J]. IEEE Transactions on Image Processing, 2020, 30: 1100-1115.
[5] Galdran A, Alvarez-Gila A, Bria A, et al. On the duality between retinex and image dehazing [C]//IEEE Conference on Computer Vision and Pattern Recognition, 2018: 8212-8221.
[6] 蒋华伟， 杨震， 张鑫， 等. 图像去雾算法研究进展[J]. 吉林大学学报（工学版）, 2021, 51(4): 1169-1181. Jiang H W, Yang Z, Zhang X, et al. Research progress of image dehazing algorithms [J]. Journal of Jilin University (Engineering and Technology Edition), 2021, 51(4): 1169-1181. (in Chinese)
[7] 陈海秀, 黄仔洁, 陆康, 等. 基于特征增强的双重注意力去雾网络[J]. 电光与控制, 2025, 32(1): 15-20, 67. Cheng H X, Huang Z J, Lu K, et al. Dual-attention dehazing network based on feature enhancement [J]. Electronics Optics & Control, 2025, 32(1): 15-20, 67. (in Chinese)
[8] Qin X, Wang Z L, Bai Y C, et al. FFA-Net: feature fusion attention network for single image dehazing [DB/OL]. (2019-12-05) [2025-08-05]. https://arxiv.org/abs/1911.07559.
[9] Li B Y, Peng X L, Wang Z Y, et al. AOD-Net: all-in-one dehazing network [C]//16th IEEE International Conference on Computer Vision (ICCV), 2017: 4780-4788.
[10] Xiong J M, Yan X F, Wang Y Z, et al. RSHazeDiff: a unified Fourier-aware diffusion model for remote sensing image dehazing [J]. IEEE Transactions on Intelligent Transportation Systems, 2025, 26(1): 1055-1070.
[11] Wang Y Z, Yan X F, Wang F L, et al. UCL-Dehaze: toward real-world image dehazing via unsupervised contrastive learning [DB/OL]. (2022-05-04) [2025-08-05]. https://arxiv.org/abs/2205.01871.
[12] Liu X H, Ma Y R, Shi Z H, et al. Grid Dehaze Net: attention-based multi-scale network for image dehazing [C]//IEEE/CVF International Conference on Computer Vision, 2019: 7314- 7323.
[13] 禹晶, 李大鹏, 廖庆敏. 基于物理模型的快速单幅图像去雾方法[J]. 自动化学报, 2011, 37(2): 143-149. Yu J, Li D P, Liao Q M. Physics-based fast single image fog removal [J]. Acta Automatica Sinica, 2011, 37(2): 143-149. (in Chinese)
[14] 陈科圻, 朱志亮, 邓小明, 等. 多尺度目标检测的深度学习研究综述[J]. 软件学报, 2021, 32(4): 1201-1227. Chen K Q, Zhu Z L, Deng X M, et al. Deep learning for multi-scale object detection: a survey [J]. Journal of Software, 2021, 32(4): 1201-1227. (in Chinese)
[15] Li R D, Pan J S, Li Z C, et al. Single image dehazing via conditional generative adversarial network [C]//IEEE Conference on Computer Vision and Pattern Recognition, 2018: 8202-8211.
[16] Reinhard E, Ashikhmin M, Gooch B, et al. Color transfer between images [J]. IEEE Computer Graphics and Applications, 2001, 21(5): 34-41.
[17] Pesme S, Flammarion N. Online robust regression via SGD on the ‘1 loss [C]//34th Conference on Neural Information Processing Systems (NeurIPS), 2020: 2540-2552.
[18] Li B Y, Ren W Q, Fu D P, et al. Benchmarking single-image dehazing and beyond [DB/OL]. (2019-04-22) [2025-08-05]. https://arxiv.org/abs/1712.04143.
[19] 张丹丹, 赵迎会. 自然图像质量评价方法综述[J]. 电脑知识与技术, 2020, 16(9): 209-211. Zhang D D, Zhao Y H. Survey of natural images-quality assessment method [J]. Computer Knowledge and Technology, 2020, 16(9): 209-211. (in Chinese)
[20] Mittal A, Soundararajan R, Bovik A C. Making a “completely blind” image quality analyzer [J]. IEEE Signal Processing Letters, 2013, 20(3): 209-212.
[21] He K M, Sun J, Tang X O. Single image haze removal using dark channel prior [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 33(12): 2341-2353.
