Space-time anomaly detection method based on convolutional sparse coding and optical flow

Abstract

The invention relates to the technical field of video monitoring, in particular to a space-time anomaly detection method based on convolutional sparse coding and optical flow. The method comprises two parts of video space-time feature learning and video anomaly detection. According to the algorithm, the convolutional sparse coding representation and the optical flow field are fused to detect the abnormal events in the video from the two aspects of space appearance and time motion, the appearance abnormal score and the motion abnormal score are summed to obtain a total abnormal score with a judgment basis, and finally intelligence of a video monitoring system is achieved through artificial intelligence. The method is used for assisting security personnel in effectively maintaining the safety of social people and reducing unnecessary labor cost. The intelligence is to say that the monitoring system has independent monitoring and independent decision-making capabilities, so that the monitoring system screens and judges the information while collecting the field image information in real time, then the monitoring system sends out alarm information in time when detecting abnormal events such as fighting, panic and the like, and then the harm caused by the abnormal event is reduced.

Description

technical field [0001] The invention relates to the technical field of video monitoring, in particular to a spatio-temporal anomaly detection method based on convolutional sparse coding and optical flow. Background technique [0002] In recent decades, the improvement of the economic level has led to the rapid development of urbanization. As more and more people flood into the city, the current social population has become denser, and it has also been accompanied by various unrest events. occur. In such an environment, people urgently need a safe and reliable social environment to ensure the safety of their lives and property. Therefore, how to ensure the current social public safety has become a hot issue in today's academic and industrial circles. At this stage, relatively complete monitoring systems have been installed in public places with a large flow of people, such as subways, shopping malls, and stations. On-site real-time monitoring, such as figure 1 Common abnor...

AI Technical Summary

Problems solved by technology

[0003] However, such a monitoring system requires human eyes to observe one or more real-time video streams for a long time, which ignores the fact that the human body cannot concentrate and maintain attention for a long time, so that the monitoring quality and efficiency of the monitoring system cannot be guaranteed, and at the same time it cannot Avoiding the waste of a lot of human resources, in view of this, we propose a spatio-temporal anomaly detection method based on convolutional sparse coding and optical flow

Images