Removal of Haze and Noise from a Single Image

被引:17
作者
Matlin, Erik [1 ]
Milanfar, Peyman [1 ]
机构
[1] Univ Calif Santa Cruz, Dept Elect Engn, Santa Cruz, CA 95064 USA
来源
COMPUTATIONAL IMAGING X | 2012年 / 8296卷
关键词
dehazing; denoising; dark channel prior; BM3D; atmospheric light; transmission map; single image; DENOISING ALGORITHMS;
D O I
10.1117/12.906773
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Images of outdoor scenes often contain degradation due to haze, resulting in contrast reduction and color fading. For many reasons one may need to remove these effects. Unfortunately, haze removal is a difficult problem due the inherent ambiguity between the haze and the underlying scene. Furthermore, all images contain some noise due to sensor (measurement) error that can be amplified in the haze removal process if ignored. A number of methods have been proposed for haze removal from images. Existing literature that has also addressed the issue of noise has relied on multiple images either for denoising prior to dehazing(1) or in the dehazing process itself. 2, 3 However, multiple images are not always available. Recent single image approaches, one of the most successful being the "dark channel prior", 4 have not yet considered the issue of noise. Accordingly, in this paper we propose two methods for removing both haze and noise from a single image. The first approach is to denoise the image prior to dehazing. This serial approach essentially treats haze and noise separately, and so a second approach is proposed to simultaneously denoise and dehaze using an iterative, adaptive, non-parametric regression method. Experimental results for both methods are then compared. Our findings show that when the noise level is precisely known a priori, simply denoising prior to dehazing performs well. When the noise level is not given, latent errors from either "under"- denoising or "over"-denoising can be amplified, and in this situation, the iterative approach can yield superior results.
引用
收藏
页数:12
相关论文
共 50 条
[41]   SINGLE IMAGE HAZE REMOVAL WITH WLS-BASED EDGE-PRESERVING SMOOTHING FILTER [J].
Park, Dubok ;
Han, David K. ;
Ko, Hanseok .
2013 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2013, :2469-2473
[42]   Single Image Haze Removal Based on Dark Channel Prior Applied on Air Duct Robot [J].
Yang, Pengfei ;
Sun, Wei ;
Liu, Shengnan ;
OuYang, Minghua .
ADVANCES IN MANUFACTURING TECHNOLOGY, PTS 1-4, 2012, 220-223 :1307-1310
[43]   Saliency-driven single image haze removal method based on reliable airlight and transmission [J].
Zhang, Libao ;
Wang, Xiaohan ;
She, Chen ;
Wang, Shiyi ;
Zhang, Zhe .
JOURNAL OF ELECTRONIC IMAGING, 2018, 27 (02)
[44]   An Adaptive Haze Removal Method for Single Remotely Sensed Image Considering the Spatial and Spectral Varieties [J].
Qi Q. ;
Zhang C. ;
Yuan Q. ;
Li H. ;
Shen H. ;
Cheng Q. .
Wuhan Daxue Xuebao (Xinxi Kexue Ban)/Geomatics and Information Science of Wuhan University, 2019, 44 (09) :1369-1376
[45]   Joint rain and atmospheric veil removal from single image [J].
Mi, Zetian ;
Wang, Yafei ;
Zhao, Congcong ;
Du, Fengming ;
Fu, Xianping .
IET IMAGE PROCESSING, 2020, 14 (06) :1150-1156
[46]   Image Haze Removal Using Dark Channel Prior [J].
Liu, ShaSha ;
Shen, Xianghui .
PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON COMPUTER, NETWORKS AND COMMUNICATION ENGINEERING (ICCNCE 2013), 2013, 30 :269-271
[47]   Automatic Image Haze Removal Based on Luminance Component [J].
Guo, Fan ;
Cai, Zixing ;
Xie, Bin ;
Tang, Jin .
2010 6TH INTERNATIONAL CONFERENCE ON WIRELESS COMMUNICATIONS NETWORKING AND MOBILE COMPUTING (WICOM), 2010,
[48]   A Simple Haze Removal Algorithm to Remove Haze from Videos [J].
Karthika, P. B. ;
Sindhu, R. .
2017 INTERNATIONAL CONFERENCE ON COMPUTING METHODOLOGIES AND COMMUNICATION (ICCMC), 2017, :461-464
[49]   Different Haze Image Conditions for Single Image Dehazing Method [J].
Husain, Noor Asma ;
Rahim, Mohd Shafry Mohd ;
Chaudhry, Huma .
2021 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (IEEE SSCI 2021), 2021,
[50]   Robust Single-Image Haze Removal Using Optimal Transmission Map and Adaptive Atmospheric Light [J].
Ngo, Dat ;
Lee, Seungmin ;
Kang, Bongsoon .
REMOTE SENSING, 2020, 12 (14)