Research On Pedestrian Abnormal Behavior Detection In Intelligent Monitoring System

In recent years,with the development of computer technology and the increase of security requirements,the development of intelligent monitoring system is changing with each passing day,and the abnormal behavior detection technology in intelligent monitoring system is the main difference with the traditional video surveillance system.Detecting the abnormal behavior of pedestrians which is based on the video sequence can not only automatically analyze the events in the monitoring scene,but also send a warning message timely according to the abnormal behavior detected by the system.Therefore,this system can not only save the cost of monitoring,but also improve the efficiency of monitoring greatly,which has high practical value and economic value.In this paper,the pedestrian target in the video image sequence is detected and tracked,and the objective feature is used to judge whether it is abnormal behavior on the basis of previous detection and tracking.In this paper,we mainly study the following three aspects:Firstly,the detection of moving target,this paper implements the ViBe algorithm combined with the frame difference method to detect the moving target,and removes small interference points through the morphological post-processing method.Secondly,in this paper,a multi-target tracking algorithm for multi-feature fusion matching is proposed.The algorithm predicts the target position in the next frame by using the Kalman filter,and then through the multi-feature fusion,the matching algorithm determines the degree of matching between the predicted target template and the current target,and realizes the tracking of multiple targets.Thirdly,the detection of pedestrians' abnormal behavior,this paper extracts the features of objects mainly by their physical characteristics,for the four different abnormal behavior,we set the corresponding rules,and finally detect the abnormal behavior and alarm.Finally,I complete the overall algorithm programming through the VS2010 development environment and OpenCV computer vision library,on the platform of MFC which is in the VS2010,I complete the interface development of abnormal behavior detection system,and test the performance of the software system.The test results show that the abnormal behavior detection system designed and implemented in this paper can not only detect and track pedestrians in real time,but also detect the abnormal behavior of pedestrians at the same time,and reports timely to meet the real-time requirements of monitoring system,At the same time,this system has a certain application value.