Method for extraction of dynamic parameters of pedestrian traffic based on digital image processing

In order to extract dynamic parameters for evacuation research,in this paper,we process the video scene of pedestrian movement.Firstly,we extract the foreground using mixed Gaussian models.Secondly,we use mean-shift algorithm to effectively segment and exactly track the connected pedestrian objects in the video which records the people moving with a free velocity.As a result,we get the trajectories which contain all the spatial and temporal information about the pedestrian movement in the video scene.And through the trajectories of pedestrians,we obtain the velocity variation characteristics of pedestrians who move with a free velocity.The results show that the method can extract characteristic parameters of pedestrian movement easily,rapidly and exactly.This method can provide effective means to extract fundamental data such as the velocity,pedestrian density and flow in the evacuation experiments and real evacuation scenes.