-
摘要: 根据雾气浓度的视觉特征,提出一种雾气浓度估计模型.在此基础上,结合大气散射模型,提出一种新的图像去雾算法.首先,基于雾气浓度估计模型计算出雾气浓度量化图,利用模糊聚类算法在量化图中识别出雾气最浓区域并估计出全球光; 然后,对量化图中的“非雾气最浓”区域再次进行聚类处理,根据文中所提最优透射率评价指标估计出每个聚类单元的透射率,将全球光与透射图以及有雾图像导入散射模型,便可达到去雾的目的; 最后,针对去雾后图像较实际场景偏暗,提出一种基于小波域的多尺度锐化算法进行增强处理,以改善其主观视觉质量.实验结果表明,本文算法与现有主流算法相比,具有更好的去雾效果,并且其计算速度也相对较快.Abstract: This paper proposes a haze thickness estimation model based on visual characteristics of haze thickness, and combines this model with atmosphere scattering model to present an innovative image dehazing algorithm. First, a haze thickness quantitative map is calculated via the haze thickness estimation model, from which the thickest area is identified by the fuzzy clustering algorithm and global atmospheric light is estimated. After that, the algorithm carries on clustering processing towards the non-thickest area in the quantitative map, and estimates the transmission of each cluster unit according to the optimized transmission evaluation index mentioned in this paper. The haze-free image can be restored from scattering model with global light, refined transmission map and original hazy image. At last, we propose a multi-scale sharpening algorithm based on wavelet domain to make up for the defect that the haze-free image is dark-look so as to improve the visual effect. Several numerical experiments demonstrate that the proposed method outperforms the mainstream dehazing algorithms in daze removal effect at a much lower implementation cost.
-
图 8 对比分析((a)基于指标Ψ与邻域估计法得到的透射图; (b)基于指标Ψ与聚类估计法得到的透射图; (c)透射图(a)对应的去雾效果; (d)透射图(b)对应的去雾效果; (e)基于黑色通道先验与邻域估计法得到的透射图; (f)基于黑色通道先验与聚类估计法得到的透射图; (g)透射图(e)对应的去雾效果; (h)透射图(f)对应的去雾效果)
Fig. 8 Comparative analysis ((a) Transmission map with neighborhood and index Ψ; (b) Transmission map with cluster unit and index Ψ; (c) Dehazed image using (a); (d) Dehazed image using (b); (e) Transmission map with neighborhood and dark channel prior; (f) Transmission map with cluster unit and dark channel prior; (g) Dehazed image using (e); (h) Dehazed image using (f))
表 1 图像质量评价指标
Table 1 Image quality evaluation parameters
实验对象 Ancuti算法 Zhu算法 Tarel算法 Tan算法 He算法 Meng算法 本文算法 评价指标 r S H r S H r S H r S H r S H r S H r S H Dolls 1.22 0.73 0.04 2.11 0.81 0.00 2.63 0.66 0.32 3.60 0.28 0.34 2.67 0.65 0.00 2.76 0.68 0.01 2.98 0.72 0.08 Manhattan 1.52 0.80 0.33 1.32 0.82 0.01 1.98 0.60 0.03 3.17 0.73 0.00 1.76 0.84 0.01 2.12 0.78 0.28 1.97 0.81 0.04 Cityscape 1.81 0.73 0.27 2.81 0.70 0.08 4.70 0.38 0.20 6.25 0.29 0.72 4.43 0.52 0.08 3.25 0.64 0.03 4.47 0.62 0.01 Mountain 1.08 0.67 0.01 1.25 0.78 0.02 1.80 0.80 0.01 2.06 0.74 0.05 1.13 0.82 0.00 1.66 0.83 0.17 1.46 0.84 0.01 Town 1.37 0.88 0.14 1.43 0.82 0.05 2.59 0.73 0.14 2.81 0.48 0.32 1.77 0.78 0.01 2.06 0.77 0.67 2.10 0.92 0.03 Pizza 1.54 0.90 0.17 1.07 0.92 0.01 1.71 0.79 0.03 3.00 0.32 0.72 1.30 0.96 0.02 1.37 0.90 0.15 1.18 0.93 0.00 -
[1] Yadav G, Maheshwari S, Agarwal A. Contrast limited adaptive histogram equalization based enhancement for real time video system. In: Proceedings of the 2014 International Conference on Advances in Computing, Communications, and Informatics. New Delhi, India: IEEE, 2014. 2392-2397 [2] Ancuti C O, Ancuti C. Single image dehazing by multi-scale fusion. IEEE Transactions on Image Processing, 2013, 22(8): 3271-3282 doi: 10.1109/TIP.2013.2262284 [3] Farid H. Blind inverse gamma correction. IEEE Transactions on Image Processing, 2001, 10(10): 1428-1433 doi: 10.1109/83.951529 [4] 禹晶, 李大鹏, 廖庆敏.基于颜色恒常性的低照度图像视见度增强.自动化学报, 2011, 37(8): 923-931 http://www.aas.net.cn/CN/abstract/abstract17511.shtmlYu Jing, Li Da-Peng, Liao Qing-Min. Color constancy-based visibility enhancement of color images in low-light conditions. Acta Automatica Sinica, 2011, 37(8): 923-931 http://www.aas.net.cn/CN/abstract/abstract17511.shtml [5] 李建奇, 阳春华, 朱红求, 曹斌芳, 刘金平.基于改进方向波变换的泡沫图像增强新方法.中南大学学报(自然科学版), 2013, 44(3): 1030-1036 http://www.cnki.com.cn/Article/CJFDTOTAL-ZNGD201303028.htmLi Jian-Qi, Yang Chun-Hua, Zhu Hong-Qiu, Cao Bin-Fang, Liu Jin-Ping. A new bubble image enhancement algorithm based on improved directionlet transform. Journal of Central South UniversityScience and Technology), 2013, 44(3): 1030-1036 http://www.cnki.com.cn/Article/CJFDTOTAL-ZNGD201303028.htm [6] Tan R T. Visibility in bad weather from a single image. In: Proceedings of the 2008 IEEE Conference on Computer Vision and Pattern Recognition. Anchorage, USA: IEEE, 2008. 1-8 http://www.oalib.com/references/16885065 [7] Nishino K, Kratz L, Lombardi S. Bayesian defogging. International Journal of Computer Vision, 2012, 98(3): 263-278 doi: 10.1007/s11263-011-0508-1 [8] Fattal R. Single image dehazing. ACM Transactions on Graphics, 2008, 27(3): Article No. 72 http://cn.bing.com/academic/profile?id=2028990532&encoded=0&v=paper_preview&mkt=zh-cn [9] Tarel J P, Hautiére N. Fast visibility restoration from a single color or gray level image. In: Proceedings of the 12th IEEE International Conference on Computer Vision. Kyoto, Japan: IEEE, 2009. 2201-2208 [10] He K, Sun J, Tang X. Single image haze removal using dark channel prior. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(12): 2341-2353 doi: 10.1109/TPAMI.2010.168 [11] Wang J B, He N, Zhang L L, Lu K. Single image dehazing with a physical model and dark channel prior. Neurocomputing, 2015, 149: 718-728 doi: 10.1016/j.neucom.2014.08.005 [12] 陈书贞, 任占广, 练秋生.基于改进暗通道和导向滤波的单幅图像去雾算法.自动化学报, 2016, 42(3): 455-465 http://www.aas.net.cn/CN/abstract/abstract18833.shtmlChen Shu-Zhen, Ren Zhan-Guang, Lian Qiu-Sheng. Single image dehazing algorithm based on improved dark channel prior and guided filter. Acta Automatica Sinica, 2016, 42(3): 455-465 http://www.aas.net.cn/CN/abstract/abstract18833.shtml [13] 张小刚, 唐美玲, 陈华, 汤红忠.一种结合双区域滤波和图像融合的单幅图像去雾算法.自动化学报, 2014, 40(8): 1733-1739 http://www.aas.net.cn/CN/abstract/abstract18440.shtmlZhang Xiao-Gang, Tang Mei-Ling, Chen Hua, Tang Hong-Zhong. A dehazing method in single image based on double-area filter and image fusion. Acta Automatica Sinica, 2014, 40(8): 1733-1739 http://www.aas.net.cn/CN/abstract/abstract18440.shtml [14] 蒋建国, 侯天峰, 齐美彬.改进的基于暗原色先验的图像去雾算法.电路与系统学报, 2011, 16(2): 7-12 http://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201502002.htmJiang Jian-Guo, Hou Tian-Feng, Qi Mei-Bin. Improved algorithm on image haze removal using dark channel prior. Journal of Circuits and Systems, 2011, 16(2): 7-12 http://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201502002.htm [15] 李加元, 胡庆武, 艾明耀, 严俊.结合天空识别和暗通道原理的图像去雾.中国图象图形学报, 2015, 20(4): 514-519 http://www.cnki.com.cn/Article/CJFDTOTAL-ZGTB201504008.htmLi Jia-Yuan, Hu Qing-Wu, Ai Ming-Yao, Yan Jun. Image haze removal based on sky region detection and dark channel prior. Journal of Image and Graphics, 2015, 20(4): 514-519 http://www.cnki.com.cn/Article/CJFDTOTAL-ZGTB201504008.htm [16] Koschmeider H. Theorie der horizontalen sichtweite. Beiträge zur Physik der Freien Atmosphäre, 1924, 12: 33-53 [17] He K M, Sun J, Tang X O. Guided image filtering. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397-1409 doi: 10.1109/TPAMI.2012.213 [18] Namer E, Schechner Y Y. Advanced visibility improvement based on polarization filtered images. In: Proceedings of the 2005 Polarization Science and Remote Sensing II. San Diego, USA: SPIE, 2005. 36-45 [19] Kim J H, Jang W D, Sim J Y, Kim C S. Optimized contrast enhancement for real-time image and video dehazing. Journal of Visual Communication and Image Representation, 2013, 24(3): 410-425 doi: 10.1016/j.jvcir.2013.02.004 [20] Uchiyama T, Arbib M A. Color image segmentation using competitive learning. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1994, 16(12): 1197-1206 doi: 10.1109/34.387488 [21] 常发亮, 刘静, 乔谊正.基于自组织神经网络的彩色图像自适应聚类分割.控制与决策, 2006, 21(4): 449-452 http://www.cnki.com.cn/Article/CJFDTOTAL-KZYC200604018.htmChang Fa-Liang, Liu Jing, Qiao Yi-Zheng. Color image self-adapting clustering segmentation based on self-organizing feature map network. Control and Decision, 2006, 21(4): 449-452 http://www.cnki.com.cn/Article/CJFDTOTAL-KZYC200604018.htm [22] Duan Ming-Xiu. Research and Application of Hierarchical Clustering Algorithm [Master dissertation], Central South University, China, 2009段明秀.层次聚类算法的研究及应用[硕士学位论文], 中南大学, 中国, 2009. [23] Hartigan J A, Wong M A. A K-means clustering algorithm. Applied Statistics, 1979, 28(1): 100-108 doi: 10.2307/2346830 [24] Bezdek J C, Ehrlich R, Full W. FCM: the fuzzy c-means clustering algorithm. Computers and Geosciences, 1984, 10(2-3): 191-203 doi: 10.1016/0098-3004(84)90020-7 [25] Hautiére N, Tarel J P, Aubert D, Dumont E. Blind contrast enhancement assessment by gradient ratioing at visible edges. Image Analysis and Stereology, 2008, 27(2): 87-95 doi: 10.5566/ias.v27.p87-95 [26] Xiao C X, Gan J J. Fast image dehazing using guided joint bilateral filter. The Visual Computer, 2012, 28(6-8): 713-721 doi: 10.1007/s00371-012-0679-y [27] Li J F, Zhang H, Yuan D, Sun M G. Single image dehazing using the change of detail prior. Neurocomputing, 2015, 156: 1-11 doi: 10.1016/j.neucom.2015.01.026 [28] Burrus C S, Gopinath R A, Guo H T. Introduction to Wavelets and Wavelet Transforms. New Jersey: Prentice Hall, 1998. [29] Tian Y, Xia D, Xu Y P. Single foggy image restoration based on spatial correlation analysis of dark channel prior. Journal of Systems Engineering and Electronics, 2014, 25(4): 688-696 doi: 10.1109/JSEE.2014.00079 [30] 张登银, 鞠铭烨, 王雪梅.一种基于暗通道先验的快速图像去雾算法.电子学报, 2015, 43(7): 1437-1443 http://www.cnki.com.cn/Article/CJFDTOTAL-DZXU201507029.htmZhang Deng-Yin, Ju Ming-Ye, Wang Xue-Mei. A fast image daze removal algorithm using dark channel prior. Acta Electronica Sinica, 2015, 43(7): 1437-1443 http://www.cnki.com.cn/Article/CJFDTOTAL-DZXU201507029.htm [31] Zhu Q, Mai J, Shao L. A fast single image haze removal algorithm using color attenuation prior. IEEE Transactions on Image Processing, 2015, 24(11): 3522-3533 doi: 10.1109/TIP.2015.2446191 [32] Meng G, Wang Y, Duan J, Xiang S, Pan C. Efficient image dehazing with boundary constraint and contextual regularization. In: Proceedings of the 2013 IEEE International Conference on Computer Vision. Sydney, Australia: IEEE, 2013. 617-624 [33] Wang Z, Bovik A C, Sheikh H R, Simoncelli E P. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing, 2004, 13(4): 600-612 doi: 10.1109/TIP.2003.819861 [34] Jobson D J, Rahman Z U, Woodell G A. Statistics of visual representation. In: Proceedings of the 2002 Visual Information Processing. Orlando, FL: SPIE, 2002. 25-35 [35] Ling Z G, Li S T, Wang Y N, Shen H, Lu X. Adaptive transmission compensation via human visual system for efficient single image dehazing. Visual Computer, 2016, 32(5): 653-662 doi: 10.1007/s00371-015-1081-3 [36] 刘海波, 杨杰, 吴正平, 张庆年, 邓勇.基于暗通道先验和Retinex理论的快速单幅图像去雾方法.自动化学报, 2015, 41(7): 1264-1273 http://www.aas.net.cn/CN/abstract/abstract18700.shtmlLiu Hai-Bo, Yang Jie, Wu Zheng-Ping, Zhang Qing-Nian, Deng Yong. A fast single image dehazing method based on dark channel prior and Retinex theory. Acta Automatica Sinica, 2015, 41(7): 1264-1273 http://www.aas.net.cn/CN/abstract/abstract18700.shtml -