Intelligent security prompting device and method based on behavior characteristic clustering

Abstract

The invention provides an intelligent security prompting device and a method based on behavior characteristic clustering, and the method comprises the steps: detecting each target in each frame of video image, and marking each target through an external rectangle; recording parameter information of each target bounding rectangle in each frame of video picture; intercepting a video frame, respectively calculating the change value of the parameter information of the target circumscribed rectangle of each frame and the target circumscribed rectangle of the same target of the previous frame, and forming a parameter change value array for each target in the video frame; clustering calculation is performed on the obtained multiple arrays, dividing class clusters for targets in the video frame, and selecting the class clusters with the target number smaller than a threshold value; and determining a frame where a target in the selected class cluster is located, and giving a security prompt. According to the security and protection monitoring method, the system operation amount can be reduced, the hardware cost can be reduced, the detection precision can be further improved, and the accuracy of abnormal behavior detection is ensured.

Description

technical field [0001] The invention belongs to the technical field of video monitoring, and in particular relates to an intelligent security prompting device and method based on clustering of behavioral characteristics. Background technique [0002] In smart cities, it is a very important application to conduct video security monitoring for the crowd through cameras widely distributed in the city. Through the analysis of the security video screen in the background, behaviors that endanger public safety and traffic order in the crowd can be found, such as fighting, illegal passing, crowd crowding and stampede, etc., and then give timely alarm reminders, so that the front end can quickly dispatch management personnel or police forces Dispose of to avoid greater impact and hazard. [0003] However, at present, for key areas such as squares and transportation hubs, hundreds of cameras may be deployed. Each camera captures more than 20 frames of security video in real time per ...

AI Technical Summary

Problems solved by technology

[0003] However, at present, for key areas such as squares and transportation hubs, hundreds of cameras may be deployed. Each camera captures more than 20 frames of security video in real time per second, and sends them to the background for analysis. Thousands or even tens of thousands of frames per second screen, the calculation speed is required to reach several million times per second, which brings a huge load to the analysis and alarm prompts in the background

Therefore, it is hoped to reduce the amount of calculations for security video image analysis as much as possible to reduce the load and hardware costs; the clustering algorithm in the prior art has relatively complicated acquisition of parameter information, is not rich enough, and cannot meet the requirements for the accuracy of video analysis

Images