Research On Intelligent Video Security System

With the popularization of high-definition video cameras and the application of machine vision algorithms,the intelligent video surveillance technology has developed rapidly.Especially in recent years,public safety problems have become more frequent Is a new requirement for intelligent security.However,the current intelligent video security system often needs human intervention to detect abnormal behaviors in the monitoring range,increases the workload of related personnel,easily leads to human error and is not conducive to improving monitoring efficiency.Based on the above situation,this paper firstly introduces the development and status quo of intelligent video security system in detail,clarifies the monitoring requirements and summarizes the existing problems.Then,the existing security system can not automatically determine whether the target has abnormal behavior or not,In this method,an inter-frame difference method is used to process the monitoring video in real time to detect the existence of a moving object.Once a moving object is found,the object is extracted to improve the SIFT signature verification target consistency and to calibrate the target area.After the target is determined,an adaptive updating feature distribution Compared with the original compression tracking algorithm,this paper improves the feature distribution update method of the algorithm to make it adaptively update and improve the tracking anti-blocking ability.At the same time,in view of the fact that the tracking window can not be updated in real time with the target change Problem,the use of targets to improve the scale of SIFT features adaptive tracking to improve the tracking stability,experimental results show that the improved algorithm and the original algorithm processing speed is basically the same,to meet the real-time tracking requirements;tracking process,the target tracking sequence as a The convolutional model input line(off-line model parameter acquired),for detecting the presence or absence of abnormal target behavior.Finally,according to the system design scheme,introduce the software programming environment,select the hardware equipment,and test the system in the laboratory environment to verify its effectiveness and real-time.The system studied in this paper can automatically detect the moving target,track the target and detect the abnormal behavior without manual intervention,reduce the burden on the relevant personnel and improve the security level of the monitoring area.In addition,the system can be ported to the hydrological station after combining the specific environment,,Prisons and other places requiring a higher level of security,lay the foundation for the further development of intelligent video security system.