Abnormal state detection method for medium and high density crowd

Abstract

The invention discloses an abnormal state detection method for medium and high-density crowds. Firstly, the feature points are extracted and tracked, and the speed of the feature points is calculated according to the coordinates of the feature points in the previous frame and the current frame of the monitoring video image, and then the feature points are obtained by graph analysis. Behavior consistency between points, based on behavior consistency, remove discrete points, cluster the remaining feature points, and then calculate the crowd aggregation degree, crowd movement intensity, and crowd movement direction variance of all feature points in the cluster, and then monitor according to several frames These three parameters in the video image and the size and increase of the number of crowd clusters determine the abnormal state of the crowd. The invention analyzes the crowd state based on the graph analysis method, and realizes the detection of three crowd states: the crowd evacuation state, the crowd gathering state and the crowd disturbance state.

Description

technical field [0001] The invention belongs to the field of crowd gathering state detection, and more specifically, relates to a method for detecting abnormal state of medium and high-density crowds. Background technique [0002] With the improvement of people's living standards and the development of spiritual and cultural needs, large-scale and highly dense crowds can be seen everywhere: shopping malls, temples, railway stations and various celebrations. These public places have limited space, but there are often crowds of people and bustling scenes, behind which there are huge hidden safety hazards. Among them, the frequency of serious crowd abnormalities such as parades, gang fights, and stampedes is increasing rapidly. People's lives and properties have been seriously damaged. Therefore, in the face of the severe situation, how to effectively prevent the occurrence of crowd abnormalities and control the development of accidents is a major scientific and social issue. ...

AI Technical Summary

Problems solved by technology

A large number of people interact with each other in medium-to-high-density scenes, resulting in higher frequency of conflicts and collisions, blurred moving targets, mutual occlusion, and more complex behaviors of different people, all of which make the detection of the state of medium-to-high-density crowds more complicated.

[0004] At present, the traditional crowd abnormal state detection algorithm mainly has the following deficiencies: 1) The detected crowd objects are very limited

2) The parameters of the algorithm itself are complex, and whether the parameters are properly selected will directly affect the detection results of the algorithm

3) Algorithms are often limited by factors such as occlusion, crowding, and low resolution between pedestrians, and ignore the interaction between pedestrians

Images