Accelerated haze removal for a single image by dark channel prior

Haze scatters light transmitted in the air and reduces the visibility of images. Dealing with haze is still a challenge for image processing applications nowadays. For the purpose of haze removal, we propose an accelerated dehazing method based on single pixels. Unlike other methods based on regions...

Full description

Saved in:
Bibliographic Details
Published inFrontiers of information technology & electronic engineering Vol. 20; no. 8; pp. 1109 - 1118
Main Authors Yue, Bo-xuan, Liu, Kang-ling, Wang, Zi-yang, Liang, Jun
Format Journal Article
LanguageEnglish
Published Hangzhou Zhejiang University Press 01.08.2019
Springer Nature B.V
State Key Laboratory of Industrial Control Technology,Zhejiang University,Hangzhou 310027,China
Subjects
Acceleration
Communications Engineering
Computer Hardware
Computer Science
Computer Systems Organization and Communication Networks
Electrical Engineering
Electronics and Microelectronics
Entropy (Information theory)
Haze
Image contrast
Image processing
Image restoration
Instrumentation
Light
Methods
Networks
Pixels
Signal to noise ratio
Hazy image model
TP751
Dark channel prior
Bilateral filtering
Haze removal
Online AccessGet full text

Cover

More Information
Summary:Haze scatters light transmitted in the air and reduces the visibility of images. Dealing with haze is still a challenge for image processing applications nowadays. For the purpose of haze removal, we propose an accelerated dehazing method based on single pixels. Unlike other methods based on regions, our method estimates the transmission map and atmospheric light for each pixel independently, so that all parameters can be evaluated in one traverse, which is a key to acceleration. Then, the transmission map is bilaterally filtered to restore the relationship between pixels. After restoration via the linear hazy model, the restored images are tuned to improve the contrast, value, and saturation, in particular to offset the intensity errors in different channels caused by the corresponding wavelengths. The experimental results demonstrate that the proposed dehazing method outperforms the state-of-the-art dehazing methods in terms of processing speed. Comparisons with other dehazing methods and quantitative criteria (peak signal-to-noise ratio, detectable marginal rate, and information entropy difference) are introduced to verify its performance.
ISSN:2095-9184
2095-9230
DOI:10.1631/FITEE.1700148