A sudden crowd anomaly detection method based on instantaneous energy

Abstract

The invention discloses a sudden crowd abnormality detection method based on instantaneous energy. The motion feature point extraction method based on pixel statistical analysis is used to extract the feature points of the image motion area, using kinetic energy as the basic energy feature of the crowd, and based on the energy block in the The kinetic energy difference between two consecutive frames is used to extract the instantaneous energy features of the crowd's motion state, so as to detect crowd abnormalities. The above sudden crowd anomaly detection method can not only effectively reduce the number of acquired motion feature points to avoid subsequent processing of some meaningless feature points, but also reflect the texture characteristics of the foreground motion area to reflect the distribution of crowd density, making the extraction The motion feature points are more representative and uniform; whether it is the response speed when an abnormality occurs or the description of the movement state changes of the crowd, the above sudden crowd anomaly detection method has shown outstanding advantages.

Description

An abnormal detection method of sudden crowd based on instantaneous energy technical field The present invention relates to the field of intelligent video surveillance, be specifically related to a kind of abnormal detection of sudden crowd based on instantaneous energy method. Background technique [0002] With the increasingly tense public safety situation, automatic detection of abnormal events in crowd scenes will Common security is of great significance, so intelligent video surveillance has become an important research direction in the field of computer vision. target The difficulty of the previous research direction mainly focused on how to automatically obtain the effective characteristic information of individuals or groups, and to analyze the characteristic information. to make a correct understanding and judgment. In order to try to solve the problem of automatic identification of crowd anomalies in surveillance videos, currently at home and abroad Many ...

AI Technical Summary

Problems solved by technology

The corner detection method can obtain the feature points of the foreground motion area more comprehensively, but the feature points obtained for some local areas are too concentrated, which affects the extraction of subsequent motion features

The uniform distribution method is to artificially mark the particle points evenly in the plane of the foreground motion area, and consider these particle points as the feature points of the crowd motion image. Although this method can obtain the feature points of the foreground motion area evenly, it cannot reflect the foreground motion area. The texture changes in the medium make the extraction of feature points rough

[0004] Anomaly detection algorithm based on kinetic energy features, but the response time of kinetic energy features is slow, and the description of sudden anomaly features is not significant

The sudden anomaly detection algorithm based on the acceleration feature, the calculation amount of the acceleration feature is too large, it cannot meet the real-time requirements, and the robustness is poor

When the crowd is abnormal, the most obvious feature of the crowd state change is the sudden change of the crowd's kinetic energy, which makes the energy of the crowd jump from one energy level to another. Research

Images