基于均值不等關(guān)系優(yōu)化的自適應圖像去霧算法
doi: 10.11999/JEIT190368 cstr: 32379.14.JEIT190368
-
蘭州交通大學電子與信息工程學院 蘭州 730070
Adaptive Image Dehazing Algorithm Based on Mean Unequal Relation Optimization
-
School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China
-
摘要:
針對暗通道先驗去霧算法的不足，如天空區(qū)域透射率估計過小和在景深突變處易發(fā)生光暈效應，該文提出一種新穎且高效的去霧算法。首先通過幾何分析建立霧圖對應無霧圖像暗通道圖的平面扇形模型，然后設定一種新型的高斯均值函數(shù)，對其標準差進行自適應處理，用以估計扇形模型的上下邊界值，通過引入均值不等關(guān)系對兩側(cè)邊界進行逼近，擬合出最優(yōu)無霧圖像暗通道圖，進一步求得最佳透射率，同時也改進局部大氣光的探索方法并復原出最終結(jié)果。實驗表明，與其它一些經(jīng)典算法相比較，所提算法能廣泛適用于各類圖像，去霧程度徹底且效果清晰自然，具有較低的時間復雜度，有利于實時處理。
Abstract:In view of shortcomings of dark channel prior dehazing methods, such as transmission in sky areas is small and halo effects are prone to occur in the edges, this paper proposes a novel and efficient dehazing algorithm. Firstly, the fan-shaped model with dark channel map of haze-free image is established by geometric analysis. Then a new Gaussian mean function is set to estimate the boundary values of the model and its standard deviation is adaptive processing. Mean-value unequal relationship is also introduced to approximate the two-sided boundary, which is used to fit the most excellent dark channel map of haze-free, further obtains the best transmission. At the same time the local atmospheric light is improved to recover the final result. Experimental results show that the proposed method can be widely applied to all kinds of images compared with other classical algorithms. The degree of dehazing is thorough, final result is clear and natural. More importantly, it is favorable for real-time processing that has low time complexity.
-
Key words:
- Image restoration /
- Dehazing /
- Dark Channel Prior (DCP) /
- Transmission /
- Atmospheric scattering model
-
表 1 改進的大氣光探索方法
輸入：有霧圖像${{I}^c}(x)$； 步驟 1 找出有霧圖像的3顏色通道的最大值${A}_{\max }^c(x) = \mathop {\max }\limits_{c \in \{\rm r,g,b\} } {{I}^c}(x)$ 步驟 2 進行形態(tài)學閉操作，濾波核尺寸分別為${r_1} = \min [w,h]/5$, ${r_2} = \min [w,h]/20$，得到兩次閉操作結(jié)果${s_1}$和${s_2}$； 步驟 3 求取兩次閉操作的平均值，$s = ({s_1} + {s_2})/2$ ； 步驟 4 進行交叉濾波平滑處理，得到最后的結(jié)果${{A}^c}$。 
下載: 導出CSV
表 2 各個算法的
$e$ 和$r$ 指標對比圖像 He[9]算法 Meng[11]算法 Ren[13]算法 Cai[12]算法 Sun[16]算法 本文算法 e r e r e r e r e r e r 1 4.50 1.28 5.82 1.79 7.55 1.47 2.76 1.08 6.44 1.22 9.01 1.41 2 8.44 1.69 5.36 2.48 20.71 1.52 17.87 1.56 15.74 1.49 18.68 1.81 3 13.89 1.70 22.56 2.59 10.82 1.97 9.11 1.47 11.22 2.01 21.83 2.01 4 10.83 1.48 24.93 3.77 27.00 3.01 9.87 1.36 12.74 1.99 22.63 2.22 5 6.87 1.28 12.12 1.69 15.61 1.76 11.10 1.28 17.25 2.06 17.18 1.64 6 26.23 1.73 31.11 1.90 31.36 2.60 18.85 1.30 22.75 1.94 30.04 2.38 7 15.51 1.85 38.03 4.12 20.35 2.55 14.53 1.63 24.74 2.98 18.47 2.95 8 3.69 1.41 3.12 1.58 8.94 1.79 2.49 1.13 6.33 1.74 8.56 1.42 均值 11.24 1.55 17.88 2.49 17.79 2.08 11.82 1.35 14.65 1.93 18.30 1.98
下載: 導出CSV
表 3 各個算法的
$\theta $ 和$T(s)$ 指標對比圖像 He[9]算法 Meng[11]算法 Ren[13]算法 Cai[12]算法 Sun[16]算法 本文算法 $\theta $ T $\theta $ T $\theta $ T $\theta $ T $\theta $ T $\theta $ T 1 0.00018 2.51 0.00651 3.80 0 4.27 0.00931 3.01 0.00347 2.47 0.00001 2.65 2 0.00022 2.56 0.00355 3.16 0 3.05 0 2.87 0.00019 2.67 0.00001 2.04 3 0.00031 2.38 0.00066 3.08 0 3.78 0.00197 2.94 0.00162 2.01 0 2.06 4 0 2.61 0.00003 4.54 0 4.60 0.00126 2.98 0.00276 2.39 0 2.07 5 0.00036 2.46 0.00004 3.50 0.00013 2.67 0 4.01 0 2.00 0 2.07 6 0.00161 2.80 0 4.40 0 3.36 0.00118 3.68 0.00019 2.17 0 2.09 7 0.00009 3.02 0.00014 5.10 0 3.22 0 3.31 0 2.57 0 2.43 8 0.00294 3.94 0.00079 6.55 0.00018 3.34 0.00169 7.34 0.00024 2.77 0.00016 2.55 均值 0.00071 2.78 0.00146 4.27 0.00003 3.53 0.00192 3.77 0.00105 2.38 0.00002 2.25
下載: 導出CSV
-
周妍, 李慶武, 霍冠英. 基于非下采樣Contourlet變換系數(shù)直方圖匹配的自適應圖像增強[J]. 光學 精密工程, 2014, 22(8): 2214–2222. doi: 10.3788/OPE.20142208.2214ZHOU Yan, LI Qingwu, and HUO Guanying. Adaptive image enhancement based on NSCT coefficient histogram matching[J]. Optics and Precision Engineering, 2014, 22(8): 2214–2222. doi: 10.3788/OPE.20142208.2214 CHEN Yang, LI Dan, and ZHANG Jianqiu. Complementary color wavelet: A novel tool for the color image/video analysis and processing[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2019, 29(1): 12–27. doi: 10.1109/TCSVT.2017.2776239 劉海波, 楊杰, 吳正平, 等. 基于暗通道先驗和Retinex理論的快速單幅圖像去霧方法[J]. 自動化學報, 2015, 41(7): 1264–1273. doi: 10.16383/j.aas.2015.c140748LIU Haibo, YANG Jie, WU Zhengping, et al. A fast single image dehazing method based on dark channel prior and Retinex theory[J]. Acta Automatica Sinica, 2015, 41(7): 1264–1273. doi: 10.16383/j.aas.2015.c140748 SCHECHNER Y Y, NARASIMHAN S G, and NAYAR S K. Polarization-based vision through haze[J]. Applied Optics, 2003, 42(3): 511–525. doi: 10.1364/AO.42.000511 NARASIMHAN S G and NAYAR S K. Interactive (de) weathering of an image using physical models[C]. 2003 IEEE Workshop on Color and Photometric Methods in Computer Vision, Nice, France, 2003: 1–8. TAN R T. Visibility in bad weather from a single image[C]. 2008 IEEE Conference on Computer Vision and Pattern Recognition, Anchorage, USA, 2008: 1–8. doi: 10.1109/CVPR.2008.4587643. FATTAL R. Single image dehazing[J]. ACM Transactions on Graphics, 2008, 27(3): 72. doi: 10.1145/1360612.1360671 TAREL J P and HAUTIèRE N. Fast visibility restoration from a single color or gray level image[C]. The 12th IEEE International Conference on Computer Vision, Kyoto, Japan, 2009: 2201–2208. HE Kaiming, SUN Jian, and TANG Xiaoou. 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 ZHU Qingsong, MAI Jiaming, and SHAO Ling. A fast single image haze removal algorithm using color attenuation prior[J]. IEEE Transactions on Image Processing, 2015, 24(11): 3522–3533. doi: 10.1109/TIP.2015.2446191 MENG Gaofeng, WANG Ying, DUAN Jiangyong, et al. Efficient image dehazing with boundary constraint and contextual regularization[C]. 2013 IEEE International Conference on Computer Vision, Sydney, Australia, 2013: 617–624. 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 REN Wenqi, LIU Si, ZHANG Hua, et al. Single image dehazing via multi-scale convolutional neural networks[C]. The 14th European Conference on Computer Vision, Amsterdam, The Netherlands, 2016: 154–169. 江巨浪, 孫偉, 王振東, 等. 基于透射率權(quán)值因子的霧天圖像融合增強算法[J]. 電子與信息學報, 2018, 40(10): 2388–2394. doi: 10.11999/JEIT171032JIANG Julang, SUN Wei, WANG Zhendong, et al. Integrated enhancement algorithm for hazy image using transmittance as weighting factor[J]. Journal of Electronics &Information Technology, 2018, 40(10): 2388–2394. doi: 10.11999/JEIT171032 HE Kaiming, SUN Jian, and TANG Xiaoou. Guided image filtering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397–1409. doi: 10.1109/TPAMI.2012.213 SUN Wei, WANG Hao, SUN Changhao, et al. Fast single image haze removal via local atmospheric light veil estimation[J]. Computers & Electrical Engineering, 2015, 46: 371–383. doi: 10.1016/j.compeleceng.2015.02.009 MIN Xiongkuo, ZHAI Guangtao, GU Ke, et al. Objective quality evaluation of dehazed images[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 20(8): 2879–2892. doi: 10.1109/TITS.2018.2868771 楊愛萍, 王南, 龐彥偉, 等. 人工光源條件下夜間霧天圖像建模及去霧[J]. 電子與信息學報, 2018, 40(6): 1330–1337. doi: 10.11999/JEIT170704YANG Aiping, WANG Nan, PANG Yanwei, et al. Nighttime haze removal based on new imaging model with artificial light sources[J]. Journal of Electronics &Information Technology, 2018, 40(6): 1330–1337. doi: 10.11999/JEIT170704 -
計量
- 文章訪問數(shù): 2363
- HTML全文瀏覽量: 921
- PDF下載量: 115
- 被引次數(shù): 0
下載:
