Single-Image Dehazing Using Extreme Reflectance Channel Prior

Yutong Zhang,Junwei Wang,Zizheng Hua,Qiong Wu,Xiaodian Zhang,Kun Gao,Hong Wang

doi:10.1109/access.2021.3090202

Yutong Zhang, Junwei Wang + Show 5 more

Open Access

https://doi.org/10.1109/access.2021.3090202

Copy DOI

Abstract

Image dehazing algorithms based on dark channel prior principle have achieved good results for most scenes. However, the popular dark channel prior tends to underestimate transmissions of bright areas or objects, such as the skies, white areas and self-luminous bodies, which may cause color distortions during dehazing. A complementary prior called the extreme reflectance channel prior (ERC), which combines the dark channel prior with the bright channel prior, is proposed to estimate the transmission map. The extreme reflectance channel is the union of dark and bright channel's pixels which satisfy the corresponding channel. Based on the scattering analysis results that the intensities of pixels in ERC are often close to 0 or 1 for the natural haze-free images or close to global atmospheric light if hazes occur in the air, the pixels in a hazy image can be recovered according to ERC to calculate the transmission map and then solve the haze imaging mode. Experiments show that ERC method outperforms state-of-the-art methods in PSNR, SSIM and visual perception effects.

Highlights

I N weather with fog, haze, and soot, there are numerous particles in the air, with a radius between 1 and 10 microns
This paper proposes a new prior to overcome the weakness of the dark channel prior
Based on scattering analysis of the extreme reflectance channel, the pixels of a hazy image can be separated into B1 and B2, and use the dark channel prior and the bright channel prior to estimate the corresponding transmission maps

Summary

INTRODUCTION

I N weather with fog, haze, and soot, there are numerous particles in the air, with a radius between 1 and 10 microns. Among the various priors for single-image dehazing, the most popular is the dark channel prior [43] proposed by He et al Their prior is based on the following key observation: most local patches in outdoor haze-free images contain pixels whose intensities are very low in at least one color channel. Numerous particles absorb and scatter light, and dark channels are no longer dark Based on this prior, given a hazy image, the transmission map can be estimated directly and a haze-free image can be restored. Given a hazy image, the transmission map can be estimated directly and a haze-free image can be restored Motivated by this idea, Yeh et al [44] proposed a pixel-based dark channel prior to estimate transmission maps and a bright channel prior to estimate atmospheric light, to improve computing speed and more accurately estimate airlight.

DARK CHANNEL PRIOR

ESTIMATING ATMOSPHERIC LIGHT

EXPERIMENTAL RESULTS

NATURAL IMAGES EXPERIMENT

CONCLUSION