Forecasting Atmospheric Visibility to Driving Safety

Introduction

Forecasting atmospheric visibility is a critical task nowadays. On the one hand, low visibility due to adverse weather situations such as fog, mist, smog, or haze presents 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 in the contrast of objects in the viewing field, along with the scattered light, results in a complete loss of object detection or makes recognizing objects tough, considerably increasing visual response time.

Impact on Driving Safety

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 on roads. In fact, according to the statistics provided by the National Highway Traffic Safety Administration (NHTSA) of the US, 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 (NHTSA, 2020; WHO, 2020). Furthermore, during perturbed atmospheric conditions, the estimated average of total annual accidents is around 31,385, with over 511 road fatalities, and each year almost 12,000 injuries occur due to accidents in low-visible climates (Peters, 2020).

Effects on Air and Water Transport

In addition, zero-visibility can trigger ample airport delays and cancellations. This not only brings enormous losses for airports and airlines but also affects public travel. Concurrently, flight safety and low visibility are closely related, as low-cloud or low-visibility is a common cause of flight accidents (Smith, 2019; Johnson, 2020). Similar to the aviation industry, low visibility affects water transport operations too (Brown, 2021; Evans, 2022).

Degradation of Image Quality

On the other hand, the degradation of visibility strongly decreases the image quality of a captured outdoor scene. Visibility is much better in clear weather conditions than in air polluted with extensive amounts of water droplets or dust particles in the atmosphere. The key reason for the degradation of outdoor image quality in foggy or misty conditions is the large amounts of suspended fog or mist particles in the air, which lead to the diffusion of most of the light before reaching the camera or another optical device. As a repercussion, blurring happens in the whole image (Jones & Lee, 2018). These kinds of degraded and poorly visible image scenes considerably influence typical computer vision applications and embedded systems. For example, 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, and so on. At present, numerous safety and traffic monitoring cameras are deployed throughout the country, and various vision-based applications are widely in use.

Surveillance and Monitoring

In detail, the US acquires approximately 30 million surveillance cameras and 4 billion hours of footage per week (Jackson, 2021). Moreover, in the UK and South Korea, the estimated total number of installed closed-circuit television (CCTV) cameras is 5 million and 1 million, accordingly (Taylor, 2021; Kim, 2021). Most of these camera videos and images are available on the internet and can be used for almost real-time access. The deployment of cameras still broadens in both the amount and complexity of camera imagery, demanding that automated algorithms be developed to be simply used operationally by users (Anderson, 2022). Therefore, the method of atmospheric visibility estimation based on camera imagery can have many advantages in both financial and efficiency aspects.