Research On Detecting Abnormal Behavior Of Human Body In Video Surveillance

With the sustainable development of social economy level and science and technology level,various countries or regions in the world in public places laying a large number of surveillance cameras,used to protect the relative safety of the scope of surveillance.But only rely on increase the number of surveillance cameras,and could not do intelligent analysis,the occurrence of riots in society is still unable to effectively avoid.Therefore,people for the surrounding and their own security requirements are getting higher and higher.In recent years,video intelligent analysis technology research heat continues to rise,The large-scale application of this technology will improve the security system and will be of great significance.In view of the current public video surveillance calls lag of the video,as well as the video monitor function unification characteristic,In order to analyze the behaviors of high-risk groups in real time,Improve the real-time monitoring and analysis to determine the accuracy.In this thesis,a method for analyzing and detecting human anomalous behavior based on regional optical flow energy is proposed.The main work of this thesis is as follows:(1)In order to make the system recognize the foreground object and background effectively,a new foreground extraction algorithm,Vibe algorithm,is used to obtain the foreground region of the video sequence.And through the preprocessing of the algorithm itself,the optimization scheme enables the background to adaptively update the background model,more accurate tests can be performed for subsequent tests,making the simulation test can be closer to the results of experimental prediction.(2)In order to effectively monitor the behavioral targets in the monitoring area,the nearest neighbor method is used to mark the foreground region,accurately extract the motion area;Marked the area on the monitor screen where target actions take place,so that the system can effectively identify and detect the behavior of the region.(3)In order to identify the behavior target in the region more accurately,the Lucas-kanada method is used to calculate the optical flow in the moving region.Calculated the optical flow energy for the target's behavior in the marked area,and uses the energy amplitude in subsequent chapters to determine behavioral anomalies.(4)In order to make the energy amplitude map more intuitive and objectively reflect the human behavior,a weighted energy histogram is used to describe the human behavior,and the weighted energy amplitude is used to judge whether or not the behaviors of the behavioral targets are abnormal.(5)In order to make the algorithm and behavior detection method can effectively detect the behavior of each environment,a large number of simulation experiments will be carried out on the video sequence in different scenes.Finally,the validity of the method is verified by the experimental results.