Video-based discriminant model of traffic congestion

Aiming at the increasing congestion of traffic,this paper proposed a method employing video detection to judge the extent of congestion directly. The traffic characteristic parameters such as the occupancy of road,mean-variance of occupancy, adopted the absolute value of variation of occupancy and studied their influence on the occurrence of roads crowded events. Then proposed a kind of transportation state partition method,consequently,presented a new model of judging traffic congestion. Testing by the real video data collected on spot,the results of the proposed model is in good agreement with the judgment made by human,and thus the validity of method is demonstrated.