光学学报, 2018, 38 (10): 1010006, 网络出版: 2019-05-09
基于低通滤波和多特征联合优化的夜间图像去雾 下载: 1644次封面文章
Nighttime Image Dehazing Based on Low-Pass Filtering and Joint Optimization of Multi-Feature
图 & 表
图 2. 不同方法估计环境光与去雾效果。(a)输入图像;(b)文献[ 7]算法环境光;(c)文献[ 7]算法去雾;(d)文献[ 14]算法环境光;(e)文献[ 14]算法去雾;(f)本文算法低通滤波;(g)本文算法环境光;(h)本文算法去雾结果
Fig. 2. Atmospheric light and dehazed results based on different methods. (a) Input image; (b) atmospheric light of Ref. [7] method; (c) dehazed result of Ref. [7] method; (d) atmospheric light of Ref. [14] method; (e) dehazed result of Ref. [14] method; (f) low-pass filtering of proposed method; (g) atmospheric light of proposed method; (h) dehazed result of proposed method
图 4. 不同透射率时图像去雾效果。(a)输入图像;(b) t=0.1;(c) t=0.5;(d) t=0.9
Fig. 4. Dehazed results based on different transmittances. (a) Input image; (b) t=0.1; (c) t=0.5; (d) t=0.9
图 5. 透射率估计与优化。(a)原始图像;(b)初始透射率;(c)优化透射率;(d)去雾后图像
Fig. 5. Estimation and optimization of transmittance. (a) Original image; (b) initial transmittance; (c) optimized transmittance; (d) dehazed image
图 6. 不同方法对夜间图像颜色校正结果。(a1)(a2)原始图像;(b1)(b2) Shade of Gray[20]算法;(c1)(c2)白平衡[21]算法;(d1)(d2)本文算法校正后图像
Fig. 6. Color correction of nighttime image by different methods. (a1)(a2) Original images; (b1)(b2) corrected images by Shade of Gray[20] method; (c1)(c2) corrected images by White balance[21] method; (d1)(d2)corrected images by proposed method
图 7. 不同方法的去雾结果。(a1)(a2)输入夜间图像;(b1)(b2) Retinex算法;(c1)(c2)直方图均衡化;(d1)(d2)本文算法
Fig. 7. Dehazed results of different methods. (a1)(a2) Input nighttime images; (b1)(b2) results of Retinex method; (c1)(c2) results of histogram equalization method; (d1)(d2) results of proposed method
图 8. 不同方法对Pavilion去雾结果。(a) Pavilion图;(b)文献[ 7]算法;(c)文献[ 12]算法;(d)文献[ 14]算法;(e)本文算法
Fig. 8. Dehazed results of different methods for Pavilion. (a) Pavilion image; (b) result of Ref. [7] method; (c) result of Ref. [12] method; (d) result of Ref. [14] method; (e) result of proposed method
图 9. 不同方法对Train去雾结果。(a) Train图;(b)文献[ 7]算法;(c)文献[ 12]算法;(d)文献[ 14]算法;(e)本文算法
Fig. 9. Dehazed resultsof different methods for Train. (a) Train image; (b) result of Ref. [7] method; (c) result of Ref. [12] method; (d) result of Ref. [14] method; (e) result of proposed method
图 10. 不同方法对Building去雾结果。(a) Building图;(b)文献[ 7]算法;(c)文献[ 12]算法;(d)文献[ 14]算法;(e)本文算法
Fig. 10. Dehazed results of different methods for Building. (a) Building image; (b) result of Ref. [7] method; (c) result of Ref. [12] method; (d) result of Ref. [14] method; (e) result of proposed method
表 1各方法相关指标对比
Table1. Comparison of K/C of different methods
|
杨爱萍, 赵美琪, 王海新, 鲁立宇. 基于低通滤波和多特征联合优化的夜间图像去雾[J]. 光学学报, 2018, 38(10): 1010006. Aiping Yang, Meiqi Zhao, Haixin Wang, Liyu Lu. Nighttime Image Dehazing Based on Low-Pass Filtering and Joint Optimization of Multi-Feature[J]. Acta Optica Sinica, 2018, 38(10): 1010006.