Model of automatic recognition of traffic events

Abstract

In recent years, the volume of traffic has become a significant problem. Consequently, accidents and traffic jams are far more likely than a century ago. Many of us living in metropolitan areas got used to the every-day traffic news about congestions. Early solutions attempted to lay more pavement to avoid jams, but adding more lanes is becoming less and less feasible. Besides, reckless, confused (e.g. ghost drivers) or drunken car drivers are more and more a source of danger and cause many terrible accidents and jams. Most of them ignore traffic rules and drive prohibitively in wrong directions or exceed speed limits. Instead of increasing the capacity of existing infrastructure, contemporary solutions of visual surveillance try to use roads more efficiently. Thereby, more and better traffic information which is automatically gathered in real-time is emphasized. Such information can be traffic parameters like traffic volume, occupancy and vehicle’s speed. This paper collects basic knowledge that are necessary to complete these tasks.