Driver Drowsiness Identification by Means of Passive Techniques for Eye Detection and Tracking

The aim of this paper is to describe a system whose final goal is to detect if a driver is drowsy, in order to prevent potentially danger situations. The system is based on the processing of the driver´s face image, acquired by a web cam installed on the dashboard of the car. After a brief introduction explaining the connection of the present work to the European project REFLECT, the relationship between drowsiness condition and fatal car crashes is discussed. Then, an overview of the most used techniques for face and eye detection is given, and the developed algorithm is described in detail. Finally, preliminary results of in-laboratory and in-car tests are presented and commented.