结合局部细节补充和全局信息增强的图像去雾算法

doi:10.3778/j.issn.1002-8331.2401-0091

摘要/Abstract

摘要： 针对现有的去雾方法在去雾过程中因局部细节信息丢失导致去雾后的图像纹理模糊、因全局信息缺乏导致去雾图像的亮度和对比度失调，从而使得去雾后的图像不能满足人类视觉感知的问题，提出了LGDEFormer模型，具体来说在DehazeFormer-T模型基础上加入局部细节补充模块（local detail supplement module，LDSM）来补充在去雾过程中丢失的局部信息以还原图像的细微结构和纹理，提出全局信息增强模块（global information enhancement module，GIEM）来增强在去雾过程中所需要的全局信息以减轻亮度和对比度失调。在全局信息增强模块中，提出一种选择性分支特征融合方法在不增加额外参数的情况下自适应地融合不同感受野分支的特征以实现全局信息的充分利用。通过实验表明，LGDEFormer模型比DehazeFormer-T模型增加少量参数和计算量的情况下在RESIDE的ITS数据集上的PSNR和SSIM指标分别提升了1.06 dB和0.002，在OTS数据集上的PSNR指标提升了0.45 dB。

关键词: 图像去雾, DehazeFormer-T, LGDEFormer模型, 局部细节补充, 全局信息增强

Abstract: To address the issues of texture blurring due to loss of local detail information, and brightness and contrast imbalance caused by a lack of global information in existing dehazing methods, this study introduces the LGDEFormer model. Specifically, it builds upon the DehazeFormer-T model by incorporating a local detail supplement module (LDSM) to replenish the local information lost during dehazing, thereby restoring the fine structure and texture of the image. Furthermore, a global information enhancement module (GIEM) is proposed to augment the global information essential for dehazing, aiming to alleviate issues with brightness and contrast. Within the GIEM, a selective branch feature fusion method is introduced, which adaptively merges features from different receptive field branches without adding extra parameters, thus effectively utilizing global information. Experimental results demonstrate that the LGDEFormer model, with only a marginal increase in parameters and computational overhead compared to the DehazeFormer-T model, achieves an enhancement of 1.06 dB in PSNR and 0.002 in SSIM on the RESIDE ITS dataset. Additionally, on the OTS dataset, there is an improvement of 0.45 dB in PSNR.

Key words: image dehazing, DehazeFormer-T, LGDEFormer model, local detail supplement, global information enhancement

李凯玲, 刘帆, 李毅. 结合局部细节补充和全局信息增强的图像去雾算法[J]. 计算机工程与应用, 2025, 61(9): 288-295.

LI Kailing, LIU Fan, LI Yi. Image Dehazing Algorithm Integrating Local Detail Supplement and Global Information Enhancement[J]. Computer Engineering and Applications, 2025, 61(9): 288-295.

参考文献

[1] HE K, SUN J, TANG X. Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 33(12): 2341-2353.
[2] ZHU Q, MAI J, SHAO L. A fast single image haze removal algorithm using color attenuation prior[J]. IEEE Transactions on Image Processing, 2015, 24(11): 3522-3533.
[3] FATTAL R. Dehazing using color-lines[J]. ACM Transactions on Graphics (TOG), 2014, 34(1): 1-14.
[4] NAYAR S K, NARASIMHAN S G. Vision in bad weather[C]//Proceedings of the Seventh IEEE International Conference on Computer Vision, 1999: 820-827.
[5] REN W, PAN J, ZHANG H, et al. Single image dehazing via multi-scale convolutional neural networks with holistic edges[J]. International Journal of Computer Vision, 2020, 128: 240-259.
[6] 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.
[7] LI B, PENG X, WANG Z, et al. AOD-Net: all-in-one dehazing network[C]//Proceedings of the IEEE International Conference on Computer Vision, 2017: 4770-4778.
[8] DONG H, PAN J, XIANG L, et al. Multi-scale boosted dehazing network with dense feature fusion[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 2157-2167.
[9] QIN X, WANG Z, BAI Y, et al. FFA-Net: feature fusion attention network for single image dehazing[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2020: 11908-11915.
[10] FAN G, GAN M, FAN B, et al. Multiscale cross-connected dehazing network with scene depth fusion[J]. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35(2): 1598-1612.
[11] SHEN H, ZHAO Z Q, ZHANG Y, et al. Mutual information-driven triple interaction network for efficient image dehazing[C]//Proceedings of the 31st ACM International Conference on Multimedia, 2023: 7-16.
[12] 吴峻江, 储珺, 卢昂, 等. 基于感知监督和多层次特征融合的去雾算法[J]. 计算机工程与应用, 2023, 59(21):201-213.
WU J J,CHU J,LU A,et al Perceptual supervision-guided and multi-hierarchical feature fusion for lmage dehazing[J]. Computer Engineering and Applications, 2023, 59(21): 204-213.
[13] DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An image is worth 16x16 words: Transformers for image recognition at scale[J]. arXiv:2010.11929, 2020.
[14] LIU Z, LIN Y, CAO Y, et al. Swin Transformer: hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 10012-10022.
[15] SONG Y, HE Z, QIAN H, et al. Vision transformers for single image dehazing[J]. IEEE Transactions on Image Processing, 2023, 32: 1927-1941.
[16] RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]// Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention(MICCAI 2015), Munich, Germany, October 5-9, 2015: 234-241.
[17] LI X, WANG W, HU X, et al. Selective kernel networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 510-519.
[18] HENDRYCKS D, GIMPEL K. Gaussian error linear units (GELUs)[J]. arXiv:1606. 08415, 2016.
[19] NAIR V, HINTON G E. Rectified linear units improve restricted boltzmann machines[C]//Proceedings of the 27th International Conference on Machine Learning (ICML-10), 2010: 807-814.
[20] MAAS A L, HANNUN A Y, NG A Y. Rectifier nonlinearities improve neural network acoustic models[C]//Proceedings of the International Conference on Machine Learning (ICML) ,2013.
[21] LIANG J, CAO J, SUN G, et al. SwinIR: image restoration using swin transformer[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 1833-1844.
[22] WANG Z, CUN X, BAO J, et al. Uformer: a general u-shaped transformer for image restoration[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022: 17683-17693.
[23] PARK N, KIM S. How do vision transformers work?[J]. arXiv:2202.06709, 2022.
[24] DAI J, QI H, XIONG Y, et al. Deformable convolutional networks[C]//Proceedings of the IEEE International Conference on Computer Vision, 2017: 764-773.
[25] LIU J J, HOU Q, CHENG M M, et al. Improving convolutional networks with self-calibrated convolutions[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 10096-10105.
[26] LIN M, CHEN Q, YAN S. Network in network[J]. arXiv:1312.4400, 2013.
[27] WANG Z, JI S. Smoothed dilated convolutions for improved dense prediction[C]//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2018: 2486-2495.