Identification of Abnormal Human Behavior in Intelligent Video Surveillance System
Bo Zhai · 2018
Focusing on the security issues of current group activities, strengthening the prevention of group incidents is the focus of current thinking.Combined with the above requirements, crowd abnormality recognition algorithms have begun to enter people's field of vision and are valued.In this regard, this paper combines crowd abnormality and intelligent monitoring video acquisition principles, proposes a crowd recognition algorithm based on preference distribution model, uses KL distance similarity to complete the labeling of common attribute labels, and then uses the preference distribution model to complete the classification of abnormal behavior, and get the judgment of abnormal behavior.Finally, through the test verification method, the verification of the above model is completed and its feasibility is proved.With the continuous increase of modern intelligent monitoring, video images show a geometric growth rate.How to combine these intelligent monitoring images to complete the detection of abnormal behaviors among them and provide references and reference for the current group safety management is the focus of current thinking and research.Video surveillance is an important part of security work.The traditional video surveillance system mainly uses the on-duty personnel to manually identify and analyze the scene images displayed by the video, in order to find abnormal and suspicious information on the screen.Obviously, the efficiency of this type of video surveillance recognition detection method is extremely low, and it is only possible to adopt video playback methods afterwards, making it difficult to realize real-time monitoring and occupying a large amount of time and manpower.In addition, the concentration of attention of personnel on duty is greatly affected by subjective factors such as the degree of fatigue, which can easily lead to incorrect judgments, resulting in misstatements, omissions, and other consequences.In summary, the traditional video surveillance system has great limitations in terms of the timeliness and accuracy of the alarm, and it is difficult to meet the needs of real-time monitoring and analysis, and lacks the capability of timely and accurate alarming and rapid emergency response.With the continuous advancement of science and technology, intelligent video surveillance technology has made great progress in recent years.By using computers instead of traditional labor to identify, track, and analyze video images, automatic monitoring of abnormal and suspicious information is realized, Can quickly and accurately carry out alarm and emergency response processing, so as to play a more powerful security role.