Research On The Method Of Group Abnormal Behavior Recognition

The massive growth of the world population,let the crowd scenes become increasingly frequent,various public occasions,such as holiday population concentration and appeared more and more security risks,these problems in addition to individual behavior and consciously abide by the public safety standards,also need more high-tech means to prevent disasters and groups to ensure public safety.Therefore,the research of video surveillance system automation is particularly important.Based on this,this paper also makes some research for this problem work as follows:(1)In this paper,we propose a method based on optical flow to detect the motion patterns.Delete the background,the foreground is divided into eight parts,in the position,magnitude and direction of flow feature extraction,creative two-dimensional coordinate was divided into four,eight,sixteen quadrant is calculated for each feature vector value,greatly improve the accuracy of system identification,use less time on the detection time,real-time performance has been greatly improved.(2)This paper presents variable dynamics modeling of local features based on the population,the population characteristics of information to establish the optical flow feature trajectory representing the motion.In addition,combined with the local population density and population movement greatly improves the abnormal crowd behavior in the range of recognition and the recognition accuracy and real-time alarm to achieve good the effect of.(3)This paper proposes the use of features based on gray level co-occurrence matrix encoding the violence crowd texture changes,and compared with other people,the consistency of inter metric changes show that the violence is various.The experiment proves that this method can effectively identify the abnormal crowd,and has good real-time performance.