Study On The Theory Of Pedestrian Evacuation Based On Video Detection And Cellular Automata

Walking is regarded as an independent mode of transportation, and whatever the vehicle adopted, walking cannot be spared at the trip origin and destination. In the large stadium, people participate in activities on feet, various types of facilities must meet the needs of walking, and the level of service of walking is directly related to the quality of the activities. Through study on group behavior of pedestrian flow caused by the interaction of individual pedestrian micro-behaviors, the relationship between micro-characteristics and macro-characteristics of pedestrian flow behavior is established. The characteristics and phenomena of group behavior under different environments are considered to be the focus of the study and hot topics of pedestrian evacuation. It has important theoretical significance, can provide valuable advice for pedestrian safety evacuation and reasonable suggestions for overall structure of buildings. In the dissertation, pedestrian video detection methods, pedestrian evacuation model under normal circumstances and emergency situations are analyzed, the main content of this dissertation is summarized as follows:(1) According to the different traffic states, a moving-based pedestrian flow detection method is provided based on pedestrian traffic characteristics for un-crowded scenes and a head-based pedestrian flow detection method is proposed based on pedestrian movement characteristics of crowded scenes. At the low-density state:the traditional Gaussian model (GMM) is improved by controling weighting parameter and variance, the possibility of moving foreground into the background model due to traffic conflict is effectively reduced; the target tracking approach of multiple moving targets merger or separation is improved by Kalman filter and Mean-Shift algorithm; BP neural network is utilized to classify movement individual, thus get the motion information of the pedestrian; at the high-density state:a head-based pedestrian flow detection method is proposed based on mixed-color model (RGB, HSV, YUV color space), head contours is extracted by Canny Algorithm and Wavelet transform, Hough transform is used to locate head by head shape features and statistics the number of pedestrian. At last, the validity and advantage of pedestrian video detection in this dissertation are verified through actual experiment.(2) Study on the pedestrian evacuation model under normal circumstances, a dynamic parameters model is established based on cellular automata (CA) in this paper, the Cognition-parameter is introduced to the traditional dynamic parameter model to simplify tactically the decision-making process of pedestrians, which can reflect the pedestrian judgment on the surrounding conditions and decide the pedestrian’s choice of action, compared with the previous model through different pedestrian distribution, the simulation results prove that this improvement makes sense, because for the choice of exit, in addition to the requirements of the spatial distance, the density around exit is also a very important influencing factor; pedestrian evacuation with and without obstacle are studied: if there is no obstacle, the optimal exit layout is given in this paper, and the effect of exit width, pedestrian density, and system size on evacuation time in the evacuation system are discussed with single exit and multiple exits; in the pedestrian evacuation model with obstacle, the effect of different obstacles layouts on evacuation time is studied, then the relationship between obstacle moving time and evacuation time are simulated at different time step. At the same time, in order to verify the reliability and practicality of the model, the actual evacuation experiments are proposed by video detection method and model simulation, the process of simulation is consistent with the actual evacuation process.(3) Study on the pedestrian evacuation model under emergency circumstances, the pedestrian’s visual field affected, pedestrian evacuation with Extrusion, pedestrian evacuation with fire is considered in this paper. If the pedestrian’s visual field affected, the pedestrian evacuation characteristics are discussed considering the cases in which pedestrians herd evacuation without evacuation sign or pedestrians move along the wall with evacuation sign available, the effect of pedestrian sight radius is introduced, and the effect of pedestrian sight radius, pedestrian density, exit width on evacuation time are discussed; if pedestrian evacuation with Extrusion, the pedestrian evacuation model with cellular capacity variable under extrusion is constructed based on CA model, Direction-parameter and Herd-parameter are proposed, the effect of exit width, system size and pedestrian density on evacuation time are analyzed; if pedestrian evacuation with fire, the pedestrian evacuation model with fire based on Extrusion is established, the model takes into account the impact of fire on the field value of system and pedestrian evacuation behavior, the effect of spread time of fire, exit width, system size and pedestrian density on the number of deaths is discussed. At last, pedestrian evacuation process of three cases is simulated, the model and update rules are reasonable and conform to reality.