Vision based Approaches for Driver Assistance Systems

The number of cars on the roads increases with every passing year. Unfortunately, the number of accidents tends to be proportional with this increase. Nowadays, many vehicles are augmented with conventional driver assistance systems (DAS) such as: sensorial-informative (rear-view alarm, radar, lane departure warning system, etc.), actuation-corrective (anti braking system – ABS, emergency braking assistance – EBA, electronic stability control – ESC), and systemic (parking aids, distance control assist system, etc.). All aforementioned systems are already implemented and designed to improve the traffic safety and to, hopefully, reduce the number of car accidents (particularly fatal ones) mostly caused by human driver errors. Many efforts have been undertaken by joint collaborations between researchers and automotive industry to include Vision Systems on vehicle prototypes in order to enhance safety. For instance, to know the driver’s state, two cameras can be mounted on the dashboard to monitor the driver's gaze, triggering an alert if some drowsiness symptoms are detected. Driver gaze detection technology was thus developed. This paper presents an overview of the most representatives vision-based approaches developed to help the car driver in taking safe actions. Key-Words: vision systems, driver assistance systems, traffic monitoring, driver’s state monitoring.