Abstract
The number of deaths and injuries from traffic accidents has been rapidly increasing. Most of those accidents occur at and nearby intersections. It is very important to detect accidents automatically, and to rescue injuries quickly. This effort should decrease deaths and ease traffic jams. In order to analyze accident events, a video camera was used to record traffic activities all day. This traffic monitoring system, the method for tracking vehicles, and the method for detecting accidents by using the Hidden Markov model are described. Experimental results achieved by applying this method to video sequences of two different real accidents are presented. (A*)