Introduction
In this paper, we propose a novel method for visibility
estimation using only a single image as input. An assumption is
proposed: the extinction coefficient of light is approximately a
constant in clear atmosphere. Such assumption is based on a key
observation – the clear atmosphere is composed of air molecules and
a few small-scale aerosol particles, and distributions of them remain
almost unchanged. Using this assumption with the theory of atmospheric
radiation, the extinction coefficient in clear atmosphere can be estimated.
Based on the dark channel prior, ratio for logarithmic values of two medium
transmission maps respectively in current and clear atmosphere is calculated
using a single image. By multiplying the clear extinction coefficient and the
ratio, we can estimate the extinction coefficient of the input image and then
obtain the visibility value. Compared with other methods that require the explicit
extraction of the scene, our method needs no constraint and performs well in various
types of scenes. Moreover, the actual distance information can also be estimated as
a by-product of this method.
In future work, we plan to explore the relationship
between visibility and air quality by deep learning,
which is expected to be an important complementary
approach for air quality monitoring.
|
Method
|
Result
|
Code
It will be uploaded later. |
Funding
This research is supported by:
- National Natural Science Foundation of China
|
Copyright notice
Please respect the original works, reproduced when the original source of the article and the author information.
More findings can be obtained in Visibility Estimation Using a Single Image (CCCV,2017) and Single Image-Based Scene Visibility Estimation (IEEE ACCESS,2019).
If our works are useful to you, please cite them.
|
If you have any questions, please contact the original author of this article Qin Li, by email: liqin6@csu.edu.cn.
|
|
