Forecasting atmospheric visibility is a critical task nowadays. One the one hand, low visibility due to adverse weather situations such as fog, mist, smog, or haze present numerous challenges to drivers, including strong negative visual effects and decreasing the ability to see through the air which is truly relevant to driving safety. These challenges arise due to the reduction of the luminance contrast. A rapid decrease of the contrast of objects in the viewing field, the scattered light result in a complete loss of object detection or recognizing objects becomes tough and visual response time considerably increases.

Moreover, because of the reductions in the optical fidelity of the aging eye older drivers mostly suffer. Particularly in dense fog, the scattered light becomes more detrimental and small objects approach the visibility threshold. This, in turn, leads to fatal incidents in roads. In facts, according to the statistics provided by National Highway Traffic Safety Administration (NHTSA) of US that 28% of all crashes occur in inconvenient weather conditions and the World Health Organization (WHO) declared that one person dies every 25 seconds because of road injuries [x1]. Furthermore, during perturbed atmospheric conditions, the estimated average of total annual accidents is around 31385, over 511 road fatalities, and each year almost 12,000 injuries due to accidents in low-visible climate [P1].

In addition, zero-visibility can trigger ample airport delays and cancellations. This not only brings enormous losses for airports and airlines but also affects the public travels. Concurrently, flight safety and low-visibility are closely relative, since low-cloud or low-visibility is a common cause of flight accidents [E3, E10, E13, E14, E16]. Similar to the aviation industry, low-visibility affects water transport operations too [E4, E22]. On the other hand, the degradation of visibility strongly decreases the image quality of a captured outdoor scene. Since visibility is much better in a clear weather condition than air polluted with an extensive amounts water droplets or dust particles in the atmosphere. The key reason for the degradation of outdoor image quality in foggy or misty conditions is large amounts of suspended fog or mist particles in the air, which lead to the diffusion of most part the light prior to reaching the camera or another optical device. As a repercussion, blurring happens in the whole image [1, 2]. These kinds of degraded and poor visible image scenes considerably influence typically computer vision applications and embedded systems. Say, navigation, tracking applications, and monitoring systems; outdoor recognition, detection, and segmentation; intelligent transportation systems; optical and camera-based applications which are used in the military and security agencies so on. Hereby, at present, numerous safety and traffic monitoring cameras deployed throughout the country, and various vision-based applications widely in use.

In details, US acquires approximately 30 million surveillance cameras and 4 billion hours of footage per week [lnk1]. Moreover, in UK and South Korea, the estimated total number of installed closed-circuit television (CCTV) cameras is 5 million and 1 million, accordingly [lnk2, lnk3]. Most of these camera videos and images are available on the internet and can be almost used real-time access. Deployments of cameras still broaden in both the amount and complexness of camera imagery demand that automated algorithms be developed to be simply used operationally by users [p4]. Therefore, the method of atmospheric visibility estimation based on camera imagery can have many superiorities in both financial and efficiency aspects.