Image Dehazing in Disproportionate Haze Distributions

Shih-Chia Huang,Da-Wei Jaw,Wenli Li,Zhihui Lu,Sy-Yen Kuo,Benjamin C. M. Fung,Bo-Hao Chen,Thanisa Numnonda

DOI: https://doi.org/10.1109/access.2021.3065968

IF: 3.9

2021-01-01

IEEE Access

Abstract:Haze removal techniques employed to increase the visibility level of an image play an important role in many vision-based systems. Several traditional dark channel prior-based methods have been proposed to remove haze formation and thereby enhance the robustness of these systems. However, when the captured images contain disproportionate haze distributions, these methods usually fail to attain effective restoration in the restored image. Specifically, disproportionate haze distribution in an image means that the background region possesses heavy haze density and the foreground region possesses little haze density. This phenomenon usually occurs in a hazy image with a deep depth of field. In response, a novel hybrid transmission map-based haze removal method that specifically targets this situation is proposed in this work to achieve clear visibility restoration and effective information maintenance. Experimental results via both qualitative and quantitative evaluations demonstrate that the proposed method is capable of performing with higher efficacy when compared with other state-of-the-art methods, in respect to both background regions and foreground regions of restored test images captured in real-world environments.

computer science, information systems,telecommunications,engineering, electrical & electronic

What problem does this paper attempt to address?

The paper attempts to address the problem of removing haze from images with disproportionate haze distributions to improve visibility and information retention. Specifically, when the background area of an image has heavy haze while the foreground area has little to no haze, existing methods based on the dark channel prior often fail to effectively restore the image. This phenomenon is particularly common in haze images with a large depth of field. Therefore, this paper proposes a new haze removal method using a hybrid transmission map, specifically targeting this situation, aiming to achieve clear visibility restoration and effective information retention. Experimental results show that this method demonstrates higher performance in both the background and foreground areas of test images taken in real-world environments, outperforming other state-of-the-art methods.