基于改进型Retinex算法的雾天图像增强技术

张驰; 谭南林; 李响; 李国正; 苏树强

doi:10.13700/j.bh.1001-5965.2018.0316

基于改进型Retinex算法的雾天图像增强技术

doi: 10.13700/j.bh.1001-5965.2018.0316

1.
北京交通大学机械与电子控制工程学院, 北京 100044
2.
华东交通大学交通运输与物流学院, 南昌 330013

基金项目:

国家自然科学基金 61527812

详细信息

作者简介:
张驰  男, 博士研究生。主要研究方向:图像处理、列车控制

谭南林  男, 博士, 教授, 博士生导师。主要研究方向:控制理论、模式识别

李响  男, 博士, 讲师。主要研究方向:信号处理、在线检测

李国正  男, 博士, 讲师。主要研究方向:传感器设计、模式识别

苏树强  男, 博士, 讲师。主要研究方向:磁性液体、传感器设计

通讯作者:
李国正, E-mail: liguozheng@bjtu.edu.cn

中图分类号: TP751.1
计量
- 文章访问数: 893
- HTML全文浏览量: 177
- PDF下载量: 418
- 被引次数: 0
出版历程
- 收稿日期: 2018-06-01
- 录用日期: 2018-08-24
- 网络出版日期: 2019-02-20

Foggy image enhancement technology based on improved Retinex algorithm

1.
School of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing 100044, China
2.
School of Transportation and Logistics, East China Jiaotong University, Nanchang 330013, China

Funds:

National Natural Science Foundation of China 61527812

More Information

Corresponding author: LI Guozheng, E-mail: liguozheng@bjtu.edu.cn

摘要

摘要:
为增强雾天图像的对比度及颜色和亮度恒常性，提出了一种改进型Retinex算法雾天图像增强算法。使用改进的双边滤波器作为滤波函数，在保持边缘信息的同时去除噪声的干扰；并使用S型函数曲线对Retinex算法中对数域相减去除入射光分量的图像进行颜色恢复处理，增强整幅图像的对比度和感知特性，还原图像的色彩信息。实验结果表明，所提的改进算法能有效提高雾天图像的清晰度和对比度，相较原雾天图像清晰度提升约200%，标准差提升约110%，信息熵提升约10%。同时，可保持更加真实鲜艳的图像颜色，计算复杂度较低，满足实时性要求。
- 图像增强 /
- Retinex /
- 双边滤波 /
- 边缘信息 /
- 颜色恢复
Abstract:
In order to enhance the contrast, color and brightness constancy of foggy images, an improved Retinex algorithm of foggy image enhancement is proposed. The algorithm uses an improved bilateral filter as a filter function to remove the noise interference while maintaining the edge information. It uses the S-shape function curve to restore image color by subtracting the incident light component from the logarithmic domain in Retinex algorithm, enhances the contrast and the perceptual characteristics of the whole image, and restores the color information of the image. The experimental results show that the improved algorithm proposed in this paper can effectively improve the clarity and contrast of the foggy image. Compared with the original foggy image, the image clarity is raised by about 200%, the standard deviation is raised by about 110%, and the information entropy is raised by about 10%. At the same time, it can maintain more realistic color information, and the computational complexity is low, which meets the real-time requirements.
- image enhancement /
- Retinex /
- bilateral filtering /
- edge information /
- color restore

HTML全文

图 1 改进型Retinex算法流程图

Figure 1. Flowchart of improved Retinex algorithm

下载: 全尺寸图片幻灯片

图 2 雾天原始图像和入射光图像

Figure 2. Original foggy image and incident light image

下载: 全尺寸图片幻灯片

图 3 颜色还原直方图曲线示例

Figure 3. Example of color reduction histogram curve

下载: 全尺寸图片幻灯片

图 4 MSRCR算法和本文算法处理结果

Figure 4. Proccessing result of MSRCR algorithm and proposed algorithm

下载: 全尺寸图片幻灯片

图 5 雾天图像1增强效果比较

Figure 5. Comparison of enhancement effects for foggy image 1

下载: 全尺寸图片幻灯片

图 6 雾天图像2增强效果比较

Figure 6. Comparison of enhancement effects for foggy image 2

下载: 全尺寸图片幻灯片

图 7 雾天图像3增强效果比较

Figure 7. Comparison of enhancement effects for foggy image 3

下载: 全尺寸图片幻灯片

图 8 雾天图像4增强效果比较

Figure 8. Comparison of enhancement effects for foggy image 4

下载: 全尺寸图片幻灯片

表 1 雾天图像1客观评价结果

Table 1. Objective evaluation result of foggy image 1

原图及算法	亮度	标准差	信息熵/bit	清晰度	处理时间/ms
原图	151.5	22.9	6.5	6.0
暗通道算法	98.3	36.7	7.1	10.2	2 185.0
HE算法	129.1	68.9	6.4	10.6	4.1
MSRCR算法	136.4	45.4	7.4	13.9	68.7
本文算法	128.2	52.8	7.5	17.6	39.2

下载: 导出CSV

表 2 雾天图像2客观评价结果

Table 2. Objective evaluation result of foggy image 2

原图及算法	亮度	标准差	信息熵/bit	清晰度	处理时间/ms
原图	145.5	24.4	6.4	1.4
暗通道算法	94.8	38.3	6.7	2.1	1 845.5
HE算法	129.4	72.9	6.3	2.6	3.9
MSRCR算法	133.8	46.6	6.5	6.2	56.5
本文算法	129.3	61.0	6.9	6.4	33.4

下载: 导出CSV

表 3 雾天图像3客观评价结果

Table 3. Objective evaluation result of foggy image 3

原图及算法	亮度	标准差	信息熵/bit	清晰度	处理时间/ms
原图	139.6	57.7	7.2	2.5
暗通道算法	120.9	69.4	7.2	3.2	2 010.1
HE算法	128.6	74.1	7.1	2.0	3.0
MSRCR算法	90.8	67.4	6.3	4.6	58.1
本文算法	128.0	75.2	7.4	5.1	33.8

下载: 导出CSV

表 4 雾天图像4客观评价结果

Table 4. Objective evaluation result of foggy image 4

原图及算法	亮度	标准差	信息熵/bit	清晰度	处理时间/ms
原图	116.8	7.4	4.8	1.3
暗通道算法	99.8	11.6	5.4	2.1	1 564.2
HE	132.8	70.0	4.8	6.3	2.0
MSRCR	118.4	17.7	6.0	4.2	58.2
本文算法	102.7	52.6	5.6	12.8	35.3

下载: 导出CSV

参考文献(18)

[1]	李海波, 曹云峰, 丁萌, 等.火星沙尘环境光学图像增强方法[J].北京航空航天大学学报, 2018, 44(3):444-453. http://bhxb.buaa.edu.cn/CN/abstract/abstract14430.shtml LI H B, CAO Y F, DING M, et al.Optical image enhancement method in dust environment on Mars[J].Journal of Beijing University of Aeronautics and Astronautics, 2018, 44(3):444-453(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract14430.shtml
[2]	刘春辉, 齐越, 丁文锐.基于大气光鲁棒估计的无人机图像去雾方法[J].北京航空航天大学学报, 2017, 43(6):1105-1111. http://bhxb.buaa.edu.cn/CN/abstract/abstract14088.shtml LIU C H, QI Y, DING W R.A haze removal method for unmanned aerial vehicle images based on robust estimation of atmospheric light[J].Journal of Beijing University of Aeronautics and Astronautics, 2017, 43(6):1105-1111(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract14088.shtml
[3]	王浩, 张叶, 沈宏海, 等.图像增强算法综述[J].中国光学, 2017, 10(4):438-448. http://d.old.wanfangdata.com.cn/Periodical/ccgxjmjxxyxb201603013 WANG H, ZHANG Y, SHEN H H, et al.Review of image enhancement algorithms[J].Chinese Optics, 2017, 10(4):438-448(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/ccgxjmjxxyxb201603013
[4]	HE K, SUN J, TANG X.Single image haze removal using dark channel prior[C]//IEEE Conference on Computer Vision and Pattern Recognition, 2009.Piscataway, NJ: IEEE Press, 2009: 1956-1963.
[5]	赵春丽, 董静薇.基于暗通道及多尺度Retinex的雾霾天气图像增强算法[J].激光杂志, 2018, 39(1):104-109. http://d.old.wanfangdata.com.cn/Periodical/jgzz201801023 ZHAO C L, DONG J W.Image enhancement algorithm of haze weather based on dark channel and multi-scale Retinex[J].Laser Journal, 2018, 39(1):104-109(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/jgzz201801023
[6]	刘海波, 杨杰, 吴正平, 等.基于暗通道先验和Retinex理论的快速单幅图像去雾方法[J].自动化学报, 2015, 41(7):1264-1273. http://www.cnki.com.cn/Article/CJFDTOTAL-XXDL201804016.htm LIU H B, YANG J, WU Z P, 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(in Chinese). http://www.cnki.com.cn/Article/CJFDTOTAL-XXDL201804016.htm
[7]	ZHANG W, LIANG J, JU H, et al.Study of visibility enhancement of hazy images based on dark channel prior in polarimetric imaging[J].Optik-International Journal for Light and Electron Optics, 2017, 130:123-130. doi: 10.1016/j.ijleo.2016.11.047
[8]	JI W, QIAN Z, XU B, et al.A nighttime image enhancement method based on Retinex and guided filter for object recognition of apple harvesting robot[J].International Journal of Advanced Robotic Systems, 2018, 15(1):1-12.
[9]	WAN M, GU G, QIAN W, et al.Infrared small target enhancement:Grey level mapping based on improved sigmoid transformation and saliency histogram[J].Journal of Modern Optics, 2018, 65(10):1161-1179. doi: 10.1080/09500340.2018.1426796
[10]	黄宇晴, 丁文锐, 李红光.基于图像增强的无人机侦察图像去雾方法[J].北京航空航天大学学报, 2017, 43(3):592-601. http://bhxb.buaa.edu.cn/CN/abstract/abstract13929.shtml HUANG Y Q, DING W R, LI H G.Haze removal method for UAV reconnaissance images based on image enhancement[J].Journal of Beijing University of Aeronautics and Astronautics, 2017, 43(3):592-601(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract13929.shtml
[11]	高明, 秦世引.基于非线性直方图变换的对比度畸变图像校正[J].北京航空航天大学学报, 2016, 42(3):514-521. http://bhxb.buaa.edu.cn/CN/abstract/abstract13338.shtml GAO M, QIN S Y.Correction of contrast distortion image based on nonlinear transform of histogram[J].Journal of Beijing University of Aeronautics and Astronautics, 2016, 42(3):514-521(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract13338.shtml
[12]	谢娜.基于图像增强的图像去雾算法研究[J].机械设计与制造工程, 2017, 46(12):31-33. doi: 10.3969/j.issn.2095-509X.2017.12.007 XIE N.Research on the image defogging algorithm based on image enhancement[J].Machine Design and Manufacturing Engineering, 2017, 46(12):31-33(in Chinese). doi: 10.3969/j.issn.2095-509X.2017.12.007
[13]	PARK S, YU S, MOON B, et al.Low-light image enhancement using variational optimization-based Retinex model[J].IEEE Transactions on Consumer Electronics, 2017, 63(2):178-184. doi: 10.1109/TCE.2017.014847
[14]	XU K, JUNG C.Retinex-based perceptual contrast enhancement in images using luminance adaptation[C]//IEEE International Conference on Acoustics, Speech and Signal Processing.Piscataway, NJ: IEEE Press, 2017: 1363-1367.
[15]	刘晓阳, 乔通, 乔智.基于双边滤波和Retinex算法的矿井图像增强方法[J].工矿自动化, 2017, 43(2):49-54. http://d.old.wanfangdata.com.cn/Periodical/mkzdh201702011 LIU X Y, QIAO T, QIAO Z.Image enhancement method of mine based on bilateral filtering and Retinex algorithm[J].Industry and Mine Automation, 2017, 43(2):49-54(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/mkzdh201702011
[16]	LIU C, CHENG I, ZHANG Y, et al.Enhancement of low visibility aerial images using histogram truncation and an explicit Retinex representation for balancing contrast and color consistency[J].ISPRS Journal of Photogrammetry & Remote Sensing, 2017, 128:16-26. http://adsabs.harvard.edu/abs/2017JPRS..128...16L
[17]	MA J, FAN X, NI J, et al.Multi-scale Retinex with color restoration image enhancement based on Gaussian filtering and guided filtering[J].International Journal of Modern Physics B, 2017, 31(16-19):1744077. doi: 10.1142/S0217979217440775
[18]	YU T H, MENG X, ZHU M, et al.An improved multi-scale Retinex FOG and haze image enhancement method[C]//International Conference on Information System and Artificial Intelligence.Piscataway, NJ: IEEE Press, 2017: 557-560.