Image clarification algorithm for underground dust and mist based on enhanced grid network
-
摘要: 针对目前井下尘雾图像清晰化算法存在的图像偏暗、细节丢失和过度增强等问题，提出一种基于增强网格网络的井下尘雾图像清晰化算法。该算法由前处理模块、主干模块和输出模块3个部分组成。前处理模块通过特征提取模块IRDB生成一组特征图，作为主干模块的输入，IRDB融合了Inception架构和密集残差连接模块（RDB）的优势，可在网络资源有限的情况下增加网络的深度和宽度，从而增强网络的表征能力、泛化能力及其对不同尺度尘雾的处理能力；主干模块采用网格网络进一步提取图像不同尺度的特征，并通过上采样和下采样实现特征图不同尺度的变换，为更好地捕捉图像中的细节信息，在网格网络中引入通道注意力机制。实验结果表明：IRDB数量为5时，网络模型的峰值信噪比（PSNR）、结构相似度指数（SSIM）和自然图像质量评价指标（NIQE）最好；从视觉效果上看，用所提算法清晰化处理后的图像细节信息更加丰富，色彩更加自然，具有良好的清晰度和对比度；在井下数据集上用所提算法处理后的图像PSNR、SSIM和NIQE分别为23.69，
0.8401 ，8.95，图像处理速度处于中等水平，整体性能优于DCP，AOD−Net等同类算法。Abstract: To address the issues of dark images, detail loss, and over-enhancement in existing underground dust and mist image clarification algorithms, an image clarification algorithm based on enhanced grid networks was proposed. This algorithm consisted of three parts: a preprocessing module, a backbone module, and an output module. The preprocessing module generated a set of feature maps using the feature extraction module IRDB, which served as the input for the backbone module. The IRDB integrated the advantages of the Inception architecture and the Residual Dense Block (RDB), increasing the depth and width of the network under limited resources, thereby enhancing the network's representational ability, generalization capability, and handling of dust and mist at different scales. The backbone module employed a grid network to further extract features at various scales of the image and implemented transformations of feature maps at different scales through upsampling and downsampling. To better capture detailed information in the images, a channel attention mechanism was introduced within the grid network. Experimental results indicated that with 5 IRDB modules, the network model achieved the best Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Naturalness Image Quality Evaluator (NIQE) metrics. Visually, the images processed using the proposed algorithm exhibited richer detail information, more natural colors, and improved clarity and contrast. The PSNR, SSIM, and NIQE values for the images processed by the proposed algorithm on the underground dataset were 23.69,0.8401 , and 8.95, respectively, with a moderate image processing speed, and the overall performance surpassed similar algorithms such as DCP and AOD-Net. -
图 1 井下尘雾图像清晰化算法整体结构
Figure 1. Overall structure of the algorithm for clarifying underground dust and mist images
图 11 不同算法对单幅图像的处理时间
Figure 11. Processing time of different algorithms for a single image
表 1 消融实验结果
Table 1. Results of ablation experiments
网络模型 PSNR SSIM NIQE w/o IRDB 21.57 0.7418 9.78 w/o SE 21.38 0.7604 9.88 w/o IRDB+SE 20.30 0.7353 10.24 完整网络 23.69 0.8401 8.95 
下载: 导出CSV
表 2 不同网络配置下的实验结果
Table 2. Experimental results under different network configurations
r c IRDB数量 PSNR SSIM NIQE 1 2 1 15.21 0.6603 12.56 2 4 3 18.91 0.7451 10.84 3 6 5 23.69 0.8401 8.95
下载: 导出CSV
表 3 在合成数据集上的定量评价指标
Table 3. Quantitative evaluation indicators on synthetic datasets
算法 PSNR SSIM NIQE DCP 16.61 0.8546 7.52 AOD−Net 20.51 0.8162 9.73 DehazeNet 19.82 0.8209 5.94 GridDehazeNet 24.72 0.8642 6.94 GFN 24.91 0.9186 9.13 MSCNN 19.84 0.8327 5.79 本文算法 31.42 0.9743 4.83
下载: 导出CSV
表 4 在井下数据集上的定量评价指标
Table 4. Quantitative evaluation indicators on underground datasets
算法 PSNR SSIM NIQE DCP 22.35 0.8494 9.09 AOD−Net 19.67 0.5315 9.53 DehazeNet 11.11 0.3910 11.03 GridDehazeNet 20.70 0.7791 9.79 GFN 20.75 0.6792 10.25 MSCNN 17.25 0.4923 9.51 本文算法 23.69 0.8401 8.95
下载: 导出CSV
-
[1] 张立. 矿山数字化转型与智能化管理[J]. 世界有色金属,2023(12):232-234. doi: 10.3969/j.issn.1002-5065.2023.12.075ZHANG Li. Digital transformation and intelligent management of mines[J]. World Nonferrous Metals,2023(12):232-234. doi: 10.3969/j.issn.1002-5065.2023.12.075 [2] GU Yanan,GAO Yiming,LIU Hairong,et al. Multi-directional rain streak removal based on infimal convolution of oscillation TGV[J]. Neurocomputing,2022,486:61-76. doi: 10.1016/j.neucom.2022.02.059 [3] 王道累,张天宇. 图像去雾算法的综述及分析[J]. 图学学报,2020,41(6):861-870.WANG Daolei,ZHANG Tianyu. Review and analysis of image defogging algorithm[J]. Journal of Graphics,2020,41(6):861-870. [4] 郑凤仙,王夏黎,何丹丹,等. 单幅图像去雾算法研究综述[J]. 计算机工程与应用,2022,58(3):1-14. doi: 10.3778/j.issn.1002-8331.2106-0134ZHENG Fengxian,WANG Xiali,HE Dandan,et al. Survey of single image defogging algorithm[J]. Computer Engineering and Applications,2022,58(3):1-14. doi: 10.3778/j.issn.1002-8331.2106-0134 [5] 涂毅晗,汪普庆. 基于多尺度局部直方图均衡化的矿井图像增强方法[J]. 工矿自动化,2023,49(8):94-99.TU Yihan,WANG Puqing. Mine image enhancement method based on multi-scale local histogram equalization[J]. Journal of Mine Automation,2023,49(8):94-99. [6] 胡明宇,陈小桥,谢银波. 宽波段微型光谱仪的小波奇异值差分去噪[J]. 武汉大学学报(工学版),2021,54(3):269-276.HU Mingyu,CHEN Xiaoqiao,XIE Yinbo. Wavelet singular value difference de-noising for broadband micro spectrometer[J]. Engineering Journal of Wuhan University,2021,54(3):269-276. [7] 徐勤功,郭杜杜. 基于Retinex理论的暗光图像增强算法[J]. 中国科技论文,2023,18(11):1267-1274. doi: 10.3969/j.issn.2095-2783.2023.11.015XU Qingong,GUO Dudu. Low light image enhancement algorithm based on Retinex theory[J]. China Sciencepaper,2023,18(11):1267-1274. doi: 10.3969/j.issn.2095-2783.2023.11.015 [8] 谢伟,余瑾,涂志刚,等. 消除光晕效应和保持细节信息的图像快速去雾算法[J]. 计算机应用研究,2019,36(4):1228-1231.XIE Wei,YU Jin,TU Zhigang,et al. Fast algorithm for image defogging by eliminating halo effect and preserving details[J]. Application Research of Computers,2019,36(4):1228-1231. [9] 闫新宇. 基于物理模型的图像去雾算法研究[D]. 北京:北京交通大学,2021:8-14.YAN Xinyu. Research of image dehazing algorithm based on physical model[D]. Beijing:Beijing Jiaotong University,2021:8-14. [10] 王媛彬,韦思雄,段誉,等. 基于自适应双通道先验的煤矿井下图像去雾算法[J]. 工矿自动化,2022,48(5):46-51,84.WANG Yuanbin,WEI Sixiong,DUAN Yu,et al. Defogging algorithm of underground coal mine image based on adaptive dual-channel prior[J]. Journal of Mine Automation,2022,48(5):46-51,84. [11] 黄鹤,胡凯益,郭璐,等. 改进的海雾图像去除方法[J]. 哈尔滨工业大学学报,2021,53(8):81-91. doi: 10.11918/202008105HUANG He,HU Kaiyi,GUO Lu,et al. Improved defogging algorithm for sea fog[J]. Journal of Harbin Institute of Technology,2021,53(8):81-91. doi: 10.11918/202008105 [12] HE Kaiming,SUN Jian,TANG Xiao'ou. 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 [13] LIU Jianlei,YU Hao,ZHANG Zhongzheng,et al. Deep multi-scale network for single image dehazing with self-guided maps[J]. Signal,Image and Video Processing,2023,17(6):2867-2875. doi: 10.1007/s11760-023-02505-2 [14] CHEN Zixuan,HE Zewei,LU Zheming. 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 [15] YANG Yizhong,HOU Ce,HUANG Haixia,et al. Cascaded deep residual learning network for single image dehazing[J]. Multimedia Systems,2023,29(4):2037-2048. doi: 10.1007/s00530-023-01087-w [16] WANG Zhendong,CUN Xiaodong,BAO Jianmin,et al. Uformer:a general U-shaped transformer for image restoration[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition,New Orleans,2022:17662-17672. [17] 王满利,张航,李佳悦,等. 基于深度神经网络的煤矿井下低光照图像增强算法[J]. 煤炭科学技术,2023,51(9):231-241. doi: 10.12438/cst.2022-1626WANG Manli,ZHANG Hang,LI Jiayue,et al. Deep neural network-based image enhancement algorithm for low-illumination images underground coal mines[J]. Coal Science and Technology,2023,51(9):231-241. doi: 10.12438/cst.2022-1626 [18] FOURURE D,EMONET R,FROMONT E,et al. Residual conv-deconv grid network for semantic segmentation[EB/OL]. [2024-06-10]. https://arxiv.org/abs/1707.07958v2. [19] HU Jie,SHEN Li,SUN Gang. Squeeze-and- excitation networks[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:7132-7141. [20] LI Boyi,PENG Xiulian,WANG Zhangyang,et al. AOD-net:all-in-one dehazing network[C]. IEEE International Conference on Computer Vision,Venice,2017:4780-4788. [21] CAI Bolun,XU Xiangmin,JIA Kui,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 [22] LIU Xiaohong,MA Yongrui,SHI Zhihao,et al. GridDehazeNet:attention-based multi-scale network for image dehazing[C]. IEEE/CVF International Conference on Computer Vision,Seoul,2019:7313-7322. [23] REN Wenqi,MA Lin,ZHANG Jiawei,et al. Gated fusion network for single image dehazing[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:3253-3261. [24] REN Wenqi,LIU Si,ZHANG Hua,et al. Single Image Dehazing via Multi-scale Convolutional Neural Networks[C]. European Conference on Computer Vision,Munich,2016:154-169. -
点击查看大图
图(11) / 表(4)
计量
- 文章访问数: 68
- HTML全文浏览量: 26
- PDF下载量: 8
- 被引次数: 0
