Campus global safety monitoring analysis method based on video AI

Abstract

The invention provides a campus global safety monitoring analysis method based on video AI, which can automatically realize judgment of a target video source and relevance. The method comprises the following steps: training a user model; acquiring camera installation information, wherein the camera installation information comprises installation position information, irradiation direction information and view angle range information; generating associated path information between the cameras according to the camera installation information; identifying a suspect by adopting a user model to generate suspect initial position information, and generating a suspect prediction track list according to the initial position information and the associated path information; identifying the suspect byadopting a camera on a preset track list and updating a predicted track list; and generating a suspect actual track list and displaying the suspect actual track list to a user. The beneficial effectsof the invention are that the method can meet the actual needs of users of a network video monitoring system for searching targets efficiently, visually, highly practically, completely and accurately.

Description

technical field [0001] The invention relates to the field of video monitoring, in particular to a method for analyzing campus security monitoring based on video AI. Background technique [0002] In the current video intelligent analysis, the key frame extraction and combination technology of video images is used for image target tracking and retrieval. Based on the characteristics of the target color, shape, texture, and motion direction, the target is extracted from the image background, and then the behavior of the target is analyzed. The characteristics are analyzed, and the identification and comparison of targets are realized according to these characteristics. [0003] Since different video sources have different backgrounds, orientations, illumination angles and ranges, it is difficult to track and find targets across cameras. A common practice in the prior art is for the user to manually search among multiple video images according to the features of the tracked tar...

AI Technical Summary

Problems solved by technology

[0005] In the existing technology, since the target video source needs to be manually selected, although some technical means of automatically extracting target features and automatically importing target features are used, this still cannot solve the continuous tracking and searching of targets across cameras, and cannot automatically generate target movement. track, the next video source to be searched cannot be automatically selected according to the current main target

[0006] Therefore, after adopting the current video target recognition and search technology, the labor intensity of the video surveillance staff has been reduced, but the efficiency of video target search and the degree of automatic processing are still very low.

Images