Crowd Scene Understanding And Behavior Analysis Based On Visualcomputer

Crowd density estimation and behavior analysis in video sequence has been avery important new branch in mode recognition and computer vision area. Theresearch and progress in this new aspect are helpful to establish an intelligentizedsystem and network. Intelligent vision monitoring system has drop many governments’attentions, as an effective monitoring method. The order of intelligent visionmonitoring is to describe and analysis the crowd’s behavior in the complex scene byalgorithm itself instead of manual work. With the social activities in the public fieldsgetting more frequently than before, the accidents caused by high crowd density werereported more frequently. For the reasons making analysis on crowd by computervision technic has become a valuable research aspect. This thesis focus on modelingof the scene, estimating of the crowd density, moving objects detection and anomalyanalysis in intelligent vision monitoring. In this paper, the algorithm of crowd densityestimation and abnormal behavior analysis has been innovated, to improve theaccuracy and get a good experiment results. The method has good versatility androbustness, practical significance which is suitable for real-time intelligent monitoringsystem.