您的位置：山东大学 -> 科技期刊社 -> 《山东大学学报（理学版）》

《山东大学学报(理学版)》 ›› 2020, Vol. 55 ›› Issue (9): 72-80.doi: 10.6040/j.issn.1671-9352.0.2020.202

•   • 上一篇    下一篇

基于变分结构引导滤波的低照度图像增强算法

林明星2,*()   

  1. 1. 山东交通职业学院机电工程学院, 山东 潍坊 261206
    2. 山东大学机械工程学院, 山东 济南 250061
  • 收稿日期:2020-05-08 出版日期:2020-09-20 发布日期:2020-09-17
  • 通讯作者: 林明星 E-mail:mxLin@sdu.edu.cn
  • 作者简介:董雪(1986—)，女，硕士，讲师,研究方向为视觉检测、图像处理. E-mail：dongxue966@126.com
  • 基金资助:
    山东省重点研发计划(重大科技创新工程)项目(2019JZZY020703);山东交通职业学院青年骨干教师国内访问学者项目

Low-light image enhancement algorithm based on variational structure

Ming-xing LIN2,*()   

  1. 1. Techanical and Electrical Engineering Department, Shandong Transport Vocational College, Weifang 261206, Shandong, China
    2. School of Mechanical Engineering, Shandong University, Jinan 250061, Shandong, China
  • Received:2020-05-08 Online:2020-09-20 Published:2020-09-17
  • Contact: Ming-xing LIN E-mail:mxLin@sdu.edu.cn

RichHTML

73

PDF (PC)

239

摘要：

针对低光照条件下获取的图像存在亮度低、可见性差等缺陷，提出了基于变分结构引导滤波的低照度图像增强算法。首先在输入图像的每个像素点处计算R、G、B三通道的最大值,并使用最大值滤波获取亮通道图。其次,创新性地构建基于变分模型的引导滤波器对亮通道图进行精炼从而估计出照射分量,并根据Retinex理论,去除照射分量得到反射分量。最后,采用同态滤波和线性拉伸进一步提高反射分量的对比度以输出清晰的图像。综合实验表明,该算法能快速有效地增强低照度图像的亮度和对比度,且能较好地保持图像细节。

关键词: 低照度图像, Retinex, 变分模型, 引导滤波, 同态滤波

Abstract:

Aiming at the problem that the images captured under low illumination present some defects, such as low brightness and poor visibility, a low-light image enhancement algorithm based on variational structure guided filtering is proposed. First, the maximum of R, G and B channels are searched at each pixel of the input image, and the bright channel map is obtained by using the maximum filtering. Secondly, a novel guided filtering based on variational model is constructed to refine the bright channel image to estimate the illumination component. Subsequently, the reflection component is obtained by removing the illumination component according to the Retinex model. Finally, homomorphic filtering and linear stretching are employed to further improve the contrast of the reflection component to output an enhanced image with clear visibility. Comprehensive experimental results show that this algorithm can effectively and quickly enhance the illumination and contrast of low-light images and retain details.

Key words: low-light image, Retinex, variational model, guided filtering, homomorphic filtering

中图分类号: 

  • TP391.41

图1

Retinex模型"

图2

算法示意图"

图3

光晕效应"

图4

估计的照射分量"

图5

4种图像增强算法效果对比"

图6

4种增强算法的局部放大图"

表1

不同算法平均指标比较"

测试图像 评价指标 原图 MSRCR算法 ALSM算法 NPEA算法 本文算法
图a 对比度 509.95 513.34 466.77 616.71 531.88
信息熵 3.58 4.45 5.06 4.93 5.22
LOE — 266.40 396.55 214.57 188.79
运行时间(s) — 1.01 28.85 28.59 1.53
图b 对比度 210.25 217.46 696.55 1 057.02 440.52
信息熵 4.02 3.89 4.66 5.17 4.81
LOE — 194.76 294.01 679.54 232.50
运行时间(s) — 1.28 55.74 58.87 2.76
图c 对比度 8.79 337.74 79.97 384.38 446.81
信息熵 1.03 2.79 1.14 1.91 3.26
LOE — 354.97 256.44 301.35 305.08
运行时间(s) — 1.30 32.59 46.47 1.83
图d 对比度 45.75 1 530.11 1 300.21 1 242.26 2 267.35
信息熵 3.02 5.53 5.39 5.76 6.27
LOE — 184.42 207.60 467.38 129.01
运行时间(s) — 1.07 16.97 21.14 1.09
图e 对比度 121.87 122.05 227.60 342.48 346.26
信息熵 2.88 3.08 4.21 3.78 4.09
LOE — 333.30 256.60 283.84 287.54
运行时间(s) — 4.52 178.43 133.72 11.01
图f 对比度 166.80 327.72 439.99 334.66 441.50
信息熵 3.92 3.49 3.55 3.35 3.78
LOE — 1 796.78 389.27 703.72 291.33
运行时间(s) — 1.95 91.84 55.95 4.33
图g 对比度 385.02 427.07 2 120.26 700.07 882.83
信息熵 4.35 4.62 5.62 5.22 5.37
LOE — 1 622.83 238.81 271.75 257.89
运行时间(s) — 1.23 40.07 31.89 1.87
图h 对比度 361.12 346.66 523.40 471.64 553.95
信息熵 4.42 4.67 5.25 5.00 5.18
LOE — 168.73 364.20 345.64 192.04
运行时间(s) — 1.16 16.15 24.08 1.19
平均值 对比度 226.19 477.77 731.84 643.65 737.89
信息熵 3.40 4.07 4.36 4.39 4.75
LOE — 615.27 300.43 408.47 235.52
运行时间(s) — 1.69 57.58 50.09 3.20
1 赵丽.基于直方图均衡的夜间图像增强算法研究[D].兰州:兰州大学, 2018.
ZHAO Li. Nighttime image enhancement algorithm research based on histogram equalization[D]. Lanzhou: Lanzhou University, 2018.
2 赵春丽, 董静薇, 徐博, 等. 融合直方图均衡化与同态滤波的雾天图像增强算法研究[J]. 哈尔滨理工大学学报, 2019, 24 (06): 93- 97.
ZHAO Chunli , DONG Jingwei , XU Bo , et al. Research on finger-image enhancement algorithm based on fusion histogram equalization and homomorphic filtering[J]. Journal of Harbin University of Science and Technology, 2019, 24 (06): 93- 97.
3 ZHOU Z , SANG N , HU X . Global brightness and local contrast adaptive enhancement for low illumination color image[J]. Optik-International Journal for Light and Electron Optics, 2014, 125 (6): 1795- 1799.
doi: 10.1016/j.ijleo.2013.09.051
4 SAYED M S , DELVA J . Low complexity contrast enhancement algorithm for nighttime visual surveillance[J]. Intelligent Systems Design and Applications, 2010, 15 (1): 835- 838.
5 LI Lin, WANG Ronggang, WANG Wenmin, et al. A low-light image enhancement method for both denoising and contrast enlarging[C]//Proceedings of IEEE International Conference on Image Processing. Quebec: IEEE, 2015.
6 董浩伟, 陈洁. 一种基于大气散射模型和Retinex的红外图像去雾算法[J]. 红外技术, 2019, 41 (4): 347- 356.
DONG Haowei , CHEN Jie . Infrared image defogging algorithm based on atmospheric scattering model and Retinex[J]. Infrared Technology, 2019, 41 (4): 347- 356.
7 余春艳, 徐小丹, 林晖翔, 等. 应用雾天退化模型的低照度图像增强[J]. 中国图象图形学报, 2017, 22 (9): 1194- 1205.
YU Chunyan , XU Xiaodan , LIN Huixiang , et al. Low-illumination image enhancement method based on a fog-degraded model[J]. Journal of Image and Graphics, 2017, 22 (9): 1194- 1205.
8 WANG Y F , LIU H M , FU Z W . Low-light image enhancement via the absorption light scattering model[J]. IEEE Transactions on Image Processing, 2019, 28 (11): 5679- 5690.
doi: 10.1109/TIP.2019.2922106
9 方志军, 刘心韵, 伍世虔, 等. 基于子带加权融合的多尺度Retinex图像增强算法[J]. 山东大学学报(理学版), 2013, 48 (11): 93- 98.
FANG Zhijun , LIU Xinyun , WU Shiqian , et al. The multi-scale Retinex algorithm for image enhancement based on sub-band weighting fusion[J]. Journal of Shandong University(Natural Science), 2013, 48 (11): 93- 98.
10 李宏宇, 朱一峰, 黄怡. 基于Retinex改进的夜间图像增强算法[J]. 长春理工大学学报(自然科学版), 2018, 41 (6): 104- 108.
LI Hongyu , ZHU Yifeng , HUANG Yi . Improved night image enhancement algorithm based on Retinex[J]. Journal of Changchun University of Science and Technology (Natural Science Edition), 2018, 41 (6): 104- 108.
11 张杰, 周浦城, 薛模根. 基于方向性全变分Retinex的低照度图像增强[J]. 计算机辅助设计与图形学学报, 2018, 30 (10): 1943- 1953.
ZHANG Jie , ZHOU Pucheng , XUE Mogen . Low-light image enhancement based on directional total variation Retinex[J]. Journal of Computer-Aided Design & Computer Graphics, 2018, 30 (10): 1943- 1953.
12 李赓飞, 李桂菊, 韩广良, 等. 亮通道先验Retinex对低照度图像的光照补偿[J]. 光学精密工程, 2018, 26 (5): 1191- 1200.
LI Gengfei , LI Guiju , HAN Guangliang , et al. Illumination compensation using Retinex model based on bright channel prior[J]. Optics and Precision Engineering, 2018, 26 (5): 1191- 1200.
13 RAHMAN Z, JOBSON D J, WOODELL G A. Multiscale Retinex for color image enhancement[C]//Proceedings of the International Conference Image Processing, Lausanne. Switzerland: IEEE, 1996.
14 GUO X , LI Y , LING H . LIME: Low-light image enhancement via illumination map estimation[J]. IEEE Transactions on Image Processing, 2017, 26 (2): 982- 993.
15 WANG S H , ZHENG J , HU H M , et al. Naturalness preserved enhancement algorithm for nonuniform illumination images[J]. IEEE Transactions on Image Processing, 2013, 22 (9): 3538- 3548.
doi: 10.1109/TIP.2013.2261309
16 黄伟国, 张永萍, 毕威, 等. 梯度稀疏和最小平方约束下的低照度图像分解及细节增强[J]. 电子学报, 2018, 46 (2): 424- 432.
HUANG Weiguo , ZHANG Yongping , BI Wei , et al. Low illumination image decomposition and details enhancement under gradient sparse and least square constraint[J]. Acta Electronica Sinica, 2018, 46 (2): 424- 432.
17 涂清华.低照度图像增强与去噪研究[D].泉州:华侨大学, 2016. TU Qinghua. The research on low illumination imageenhancement and denoising[D]. Quanzhou: Huaqiao University, 2016.
[1] 潘振宽,魏伟波,张海涛 . 基于梯度和拉普拉斯算子的图像扩散变分模型[J]. J4, 2008, 43(11): 11-16 .
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
[1] 杨莹,江龙*,索新丽. 容度空间上保费泛函的Choquet积分表示及相关性质[J]. J4, 2013, 48(1): 78 -82 .
[2] 李永明1, 丁立旺2. PA误差下半参数回归模型估计的r-阶矩相合[J]. J4, 2013, 48(1): 83 -88 .
[3] 董丽红1,2,郭双建1. Yetter-Drinfeld模范畴上的弱Hopf模基本定理[J]. J4, 2013, 48(2): 20 -22 .
[4] 周伟娜,左连翠*. 几类图的笛卡尔积图的(d,1)-全标号[J]. 山东大学学报（理学版）, 2014, 49(04): 24 -28 .
[5] 霍玉洪,季全宝. 一类生物细胞系统钙离子振荡行为的同步研究[J]. J4, 2010, 45(6): 105 -110 .
[6] 张雪凤1,刘鹏1,2. 基于类间差异最大化的加权距离改进K-means算法[J]. J4, 2010, 45(7): 28 -33 .
[7] 程李晴1,2, 石巧连2. 一种新的混合共轭梯度算法[J]. J4, 2010, 45(6): 81 -85 .
[8] 宋玉丹,王士同*. 基于特征缺省的最小类内方差支持向量机[J]. J4, 2010, 45(7): 102 -107 .
[9] 史艳华1,石东洋2*. 伪双曲方程类Wilson非协调元逼近[J]. J4, 2013, 48(4): 77 -84 .
[10] 谢云龙,杜英玲 . 函数S-粗集与规律积分度量[J]. J4, 2007, 42(10): 118 -122 .
版权所有 © 2018 《山东大学学报(理学版)》编辑部　
地址: 山东省济南市山大南路27号山东大学中心校区（250100）　电话: +86-0531-88366917　E-mail: xblxb@sdu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发