Abstract
Many real-life tasks such as military reconnaissance and traffic monitoring require high-quality images. However, images acquired in foggy or hazy weather pose obstacles to the implementation of these real-life tasks; consequently, image dehazing is an important research problem. To meet the requirements of practical applications, a single image dehazing algorithm has to be able to effectively process real-world hazy images with high computational efficiency. In this article, we present a fast and robust semi-supervised dehazing algorithm named SADnet for practical applications. SADnet utilizes both synthetic datasets and natural hazy images for training, so it has good generalizability for real-world hazy images. Furthermore, considering the uneven distribution of haze in the atmospheric environment, a Channel-Spatial Self-Attention (CSSA) mechanism is presented to enhance the representational power of the proposed SADnet. Extensive experimental results demonstrate that the presented approach achieves good dehazing performances and competitive running times compared with other state-of-the-art image dehazing algorithms.
- [1] . 2010. A fast semi-inverse approach to detect and remove the haze from a single image. In Proceedings of the 10th Asian Conference on Computer Vision (ACCV’10). 501–514.
DOI: DOI: DOI: https://doi.org/10.1007/978-3-642-19309-5_39 Google ScholarCross Ref
- [2] . 2016. Supervised, unsupervised, and semi-supervised feature selection: A review on gene selection. IEEE/ACM Trans. Comput. Biol. Bioinform. 13, 5 (2016), 971–989.
DOI: DOI: DOI: https://doi.org/10.1109/TCBB.2015.2478454 Google ScholarCross Ref
- [3] . 2016. Non-local image dehazing. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’16). 1674–1682.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2016.185Google ScholarCross Ref
- [4] . 2016. DehazeNet: An end-to-end system for single image haze removal. IEEE Trans. Image Process. 25, 11 (2016), 5187–5198.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2016.2598681 Google ScholarDigital Library
- [5] . 2015. An advanced visibility restoration algorithm for single hazy images. ACM Trans. Multim. Comput. Commun. Applic. 11 (2015).
DOI: DOI: DOI: https://doi.org/10.1145/2726947 Google ScholarCross Ref
- [6] . 2019. Gated context aggregation network for image dehazing and deraining. In Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV’19). 1375–1383.
DOI: DOI: DOI: https://doi.org/10.1109/WACV.2019.00151Google ScholarCross Ref
- [7] . 2017. SCA-CNN: Spatial and channel-wise attention in convolutional networks for image captioning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’17). 6298–6306.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2017.667Google ScholarCross Ref
- [8] . 2020. Discrete haze level dehazing network. In Proceedings of the 28th ACM International Conference on Multimedia (MM’20). New York, NY, 1828–1836.
DOI: DOI: DOI: https://doi.org/10.1145/3394171.3413876 Google ScholarCross Ref
- [9] . 2009. ImageNet: A large-scale hierarchical image database. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’09). 248–255.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2009.5206848Google ScholarCross Ref
- [10] . 2018. BERT: Pre-training of deep bidirectional transformers for language understanding. arXiv (2018).
DOI: DOI: DOI: https://doi.org/arXiv:1810.04805Google Scholar - [11] . 2018. Cycle-dehaze: Enhanced cycleGAN for single image dehazing. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW’18). 938–9388.
DOI: DOI: DOI: https://doi.org/10.1109/CVPRW.2018.00127Google ScholarCross Ref
- [12] . 2008. Single image dehazing. ACM Trans. Graph. 27, 3 (2008).
DOI: DOI: DOI: https://doi.org/10.1145/1360612.1360671Google ScholarCross Ref
- [13] . 2015. Dehazing using color-lines. ACM Trans. Graph. 34, 1 (2015).
DOI: DOI: DOI: https://doi.org/10.1145/2651362 Google ScholarCross Ref
- [14] . 2019. “Double-DIP”: Unsupervised image decomposition via coupled deep-image-priors. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’19). 11018–11027.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2019.01128Google ScholarCross Ref
- [15] . 2017. Semi-supervised sparse representation based classification for face recognition with insufficient labeled samples. IEEE Trans. Image Process. 26, 5 (2017), 2545–2560.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2017.2675341 Google ScholarCross Ref
- [16] . 2012. An investigation of dehazing effects on image and video coding. IEEE Trans. Image Process. 21, 2 (2012), 662–673.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2011.2166968 Google ScholarCross Ref
- [17] . 2020. Unsupervised single image dehazing using dark channel prior loss. IEEE Trans. Image Process. 29 (2020), 2692–2701.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2019.2952032Google ScholarCross Ref
- [18] . 2011. Single image haze removal using dark channel prior. IEEE Trans. Pattern Anal. Mach. Intell. 33, 12 (2011), 2341–2353.
DOI: DOI: DOI: https://doi.org/10.1109/TPAMI.2010.168Google ScholarDigital Library
- [19] . 2013. Guided image filtering. IEEE Trans. Pattern Mach. Intell. 35, 6 (2013), 1397–1409.
DOI: DOI: DOI: https://doi.org/10.1109/TPAMI.2012.213 Google ScholarCross Ref
- [20] . 2020. Squeeze-and-excitation networks. IEEE Trans. Pattern Anal. Mach. Intell. 42, 8 (2020), 2011–2023.
DOI: DOI: DOI: https://doi.org/10.1109/TPAMI.2019.2913372Google ScholarDigital Library
- [21] . 2013. Fine-grained semi-supervised labeling of large shape collections. ACM Trans. Graph. 32, 6 (2013).
DOI: DOI: DOI: https://doi.org/10.1145/2508363.2508364 Google ScholarCross Ref
- [22] . 1998. A model of saliency-based visual attention for rapid scene analysis. IEEE Trans. Pattern Anal. Mach. Intell. 20, 11 (1998), 1254–1259.
DOI: DOI: DOI: https://doi.org/10.1109/34.730558 Google ScholarDigital Library
- [23] . 2014. Adam: A method for stochastic optimization. arXiv (2014).
DOI: DOI: DOI: https://doi.org/arXiv:1412.6980Google Scholar - [24] . 2017. Semi-supervised deep learning for monocular depth map prediction. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’17). 2215–2223.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2017.238Google ScholarCross Ref
- [25] . 2020. Zero-shot image dehazing. IEEE Trans. Image Process. 29 (2020), 8457–8466.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2020.3016134Google ScholarCross Ref
- [26] . 2017. AOD-Net: All-in-one dehazing network. In Proceedings of the IEEE International Conference on Computer Vision (ICCV’17). 4780–4788.
DOI: DOI: DOI: https://doi.org/10.1109/ICCV.2017.511Google ScholarCross Ref
- [27] . 2019. Benchmarking single-image dehazing and beyond. IEEE Trans. Image Process. 28, 1 (2019), 492–505.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2018.2867951 Google ScholarCross Ref
- [28] . 2020. Semi-supervised image dehazing. IEEE Trans. Image Process. 29 (2020), 2766–2779.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2019.2952690Google ScholarCross Ref
- [29] . 2018. Single image dehazing via conditional generative adversarial network. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’18). 8202–8211.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2018.00856Google ScholarCross Ref
- [30] . 2018. Transformation networks for target-oriented sentiment classification. arXiv (2018).
DOI: DOI: DOI: https://doi.org/arXiv:1805.01086Google Scholar - [31] . 2020. Semi-supervised online multi-task metric learning for visual recognition and retrieval. In Proceedings of the 28th ACM International Conference on Multimedia (MM’20). 3377–3385.
DOI: DOI: DOI: https://doi.org/10.1145/3394171.3413948 Google ScholarCross Ref
- [32] . 2019. GridDehazeNet: Attention-based multi-scale network for image dehazing. In Proceedings of the IEEE International Conference on Computer Vision (ICCV’19). 7313–7322.
DOI: DOI: DOI: https://doi.org/10.1109/ICCV.2019.00741Google ScholarCross Ref
- [33] . 2019. Learning deep priors for image dehazing. In Proceedings of the IEEE International Conference on Computer Vision (ICCV’19). 2492–2500.
DOI: DOI: DOI: https://doi.org/10.1109/ICCV.2019.00258Google ScholarCross Ref
- [34] . 2013. Efficient image dehazing with boundary constraint and contextual regularization. In Proceedings of the IEEE International Conference on Computer Vision (ICCV’13). 617–624.
DOI: DOI: DOI: https://doi.org/10.1109/ICCV.2013.82 Google ScholarCross Ref
- [35] . 2002. Vision and the atmosphere. Int. J. Comput. Vis. 48, 3 (2002), 233–254.
DOI: DOI: DOI: https://doi.org/10.1023/A:1016328200723 Google ScholarCross Ref
- [36] . 2016. Blind image deblurring using dark channel prior. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’16). 1628–1636.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2016.180Google ScholarCross Ref
- [37] 2017. Automatic differentiation in Pytorch. In Proceedings of the 31st International Conference on Neural Information Processing Systems Workshops (NIPSW’17). 1–4.Google Scholar
- [38] . 2020. FFA-Net: Feature fusion attention network for single image dehazing. Proc. 34th AAAI Conf. Artif. Intell. 34, 7 (2020), 11908–11915.Google Scholar
Cross Ref
- [39] . 2016. Single image dehazing via multi-scale convolutional neural networks. In Proceedings of the 14th European Conference on Computer Vision (ECCV’16). 154–169.
DOI: DOI: DOI: https://doi.org/10.1007/978-3-319-46475-6_10Google ScholarCross Ref
- [40] . 2018. Gated fusion network for single image dehazing. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’18). 3253–3261.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2018.00343Google ScholarCross Ref
- [41] . 2020. Single image dehazing via multi-scale convolutional neural networks with holistic edges. Int. J. Comput. Vis. 128 (2020), 240259.
DOI: DOI: DOI: https://doi.org/10.1007/s11263-019-01235-8Google ScholarCross Ref
- [42] . 2015. U-net: Convolutional networks for biomedical image segmentation. In Proceedings of the 18th International Conference on Medical Image Computing and Computer-assisted Intervention (MICCAI’15). 234–241.
DOI: DOI: DOI: https://doi.org/10.1007/978-3-319-24574-4_28Google Scholar - [43] . 2014. Very deep convolutional networks for large-scale image recognition. arXiv (2014).
DOI: DOI: DOI: https://doi.org/arXiv:1409.1556Google Scholar - [44] . 2017. Semi supervised semantic segmentation using generative adversarial network. In Proceedings of the IEEE International Conference on Computer Vision (ICCV’17). 5689–5697.
DOI: DOI: DOI: https://doi.org/10.1109/ICCV.2017.606Google ScholarCross Ref
- [45] . 2021. ICycleGAN: Single image dehazing based on iterative dehazing model and CycleGAN. Comput. Vis. Image Underst. 203 (2021), 103133.
DOI: DOI: DOI: https://doi.org/10.1016/j.cviu.2020.103133Google ScholarCross Ref
- [46] . 2008. Visibility in bad weather from a single image. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’08). 1–8.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2008.4587643Google ScholarCross Ref
- [47] . 2019. Adaptive hypergraph embedded semi-supervised multi-label image annotation. IEEE Trans. Multim. 21, 11 (2019), 2837–2849.
DOI: DOI: DOI: https://doi.org/10.1109/TMM.2019.2909860Google ScholarCross Ref
- [48] . 2017. Attention is all you need. In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS’17). 6000–6010.
DOI: DOI: DOI: https://doi.org/10.5555/3295222.3295349 Google ScholarDigital Library
- [49] . 2017. Residual attention network for image classification. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’17). 6450–6458.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2017.683Google ScholarCross Ref
- [50] . 2019. A novel dehazing method for color fidelity and contrast enhancement on mobile Devices. IEEE Trans. Consum. Electron. 65, 1 (2019), 47–56.
DOI: DOI: DOI: https://doi.org/10.1109/TCE.2018.2884794 Google ScholarCross Ref
- [51] . 2018. Smoothed dilated convolutions for improved dense prediction. In Proceedings of the 24th ACM International Conference on Knowledge Discovery and Data Mining (KDD’18). 2486–2495.
DOI: DOI: DOI: https://doi.org/10.1007/s10618-021-00765-5 Google ScholarCross Ref
- [52] . 2018. CBAM: Convolutional block attention module. In Proceedings of the 15th European Conference on Computer Vision (ECCV’18). 286–301.
DOI: DOI: DOI: https://doi.org/10.1007/978-3-030-01234-2_1Google ScholarCross Ref
- [53] . 2018. Semi-supervised deep learning using pseudo labels for hyperspectral image classification. IEEE Trans. Image Process. 27, 3 (2018), 1259–1270.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2017.2772836 Google ScholarCross Ref
- [54] . 2020. Learning interleaved cascade of shrinkage fields for joint image dehazing and denoising. IEEE Trans. Image Process. 29 (2020), 1788–1801.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2019.2942504Google ScholarDigital Library
- [55] . 2018. Towards perceptual image dehazing by physics-based disentanglement and adversarial training. In Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI’18). 7485–7492. Google Scholar
Digital Library
- [56] . 2018. Self-attention generative adversarial networks. arXiv (2018).
DOI: DOI: DOI: https://doi.org/arXiv:1805.08318Google Scholar - [57] . 2018. Densely connected pyramid dehazing network. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’18). 3194–3203.
DOI: DOI: DOI: https://doi.org/10.1109/CVPR.2018.00337Google ScholarCross Ref
- [58] . 2017. HazeRD: An outdoor scene dataset and benchmark for single image dehazing. In Proceedings of the IEEE International Conference on Image Processing (ICIP’17). 3205–3209.
DOI: DOI: DOI: https://doi.org/10.1109/ICIP.2017.8296874Google ScholarCross Ref
- [59] . 2018. Image super-resolution using very deep residual channel attention networks. In Proceedings of the 15th European Conference on Computer Vision (ECCV’18). 286–301.
DOI: DOI: DOI: https://doi.org/10.1007/978-3-030-01234-2_18Google ScholarCross Ref
- [60] . 2020. Single image numerical iterative dehazing method based on local physical features. IEEE Trans. Circ. Syst. Vid. Technol. 30, 10 (2020), 3544–3557.
DOI: DOI: DOI: https://doi.org/10.1109/TCSVT.2019.2939853Google ScholarCross Ref
- [61] . 2015. A fast single image haze removal algorithm using color attenuation prior. IEEE Trans. Image Process. 24, 11 (2015), 3522–3533.
DOI: DOI: DOI: https://doi.org/10.1109/TIP.2015.2446191Google ScholarDigital Library
- [62] . 2009. Introduction to Semi-supervised Learning. Morgan and Claypool, San Mateo, CA.Google Scholar
Digital Library
Index Terms
SADnet: Semi-supervised Single Image Dehazing Method Based on an Attention Mechanism
Recommendations
DehazeFlow: Multi-scale Conditional Flow Network for Single Image Dehazing
MM '21: Proceedings of the 29th ACM International Conference on MultimediaSingle image dehazing is a crucial and preliminary task for many computer vision applications, making progress with deep learning. The dehazing task is an ill-posed problem since the haze in the image leads to the loss of information. Thus, there are ...
A content-adaptive method for single image dehazing
PCM'10: Proceedings of the Advances in multimedia information processing, and 11th Pacific Rim conference on Multimedia: Part IIA content adaptive method for single image dehazing is proposed in this work. Since the degradation level affected by haze is related to the depth of the scene and pixels in each specific part of the image (such as trees, buildings or other objects) ...
Non-homogeneous haze data synthesis based real-world image dehazing with enhancement-and-restoration fused CNNs
AbstractSingle image dehazing is a challenging ill-posed task, we address it from the aspects of dehazing dataset and network architecture. As for the dehazing dataset, there are such problems as unnatural hazy images, unqualified ground ...
Graphical abstractDisplay Omitted
Highlights- We devise a haze data synthesis method to create a dehazing dataset named FiveK-Haze.






Comments