Motion direction based abnormal behavior detection method in monitoring video

Abstract

The invention discloses a motion direction based abnormal behavior detection method in monitoring video, which solves a problem that a video abnormal behavior is difficult to be detected accurately through analyzing the motion direction. The abnormal behavior detection method mainly comprises the following three parts: preliminary video object acquisition, video object optimization and video object abnormal behavior detection. In the first part, original video frames and background video frames are processed, and a preliminary video object is acquired by using an improved background subtraction method; in the second part, preliminary video object pixels with the moving speed being within a noise boundary threshold are screened, and the accurate position of the video object is acquired; and in the third part, the motion direction of the video object pixels is calculated, and a motion direction entropy is calculated so as to judge whether a behavior is abnormal or not. The abnormal behavior detection method simulates artificial marking, can detect most of the abnormal situations in time and has good real-time performance and effectiveness.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for detecting abnormal behaviors based on motion directions in surveillance videos. Background technique [0002] In recent years, with the increase of various security emergencies, the improvement of public security awareness, and the penetration of the concept of artificial intelligence, intelligent monitoring has attracted more and more attention. The traditional monitoring system mainly realizes the safety management of public places through manual monitoring, which lacks real-time and initiative. In many cases, video surveillance only plays the role of video backup due to unmanned management, but does not fulfill the responsibility of supervision. In addition, with the popularization and widespread deployment of surveillance cameras, traditional manual surveillance methods can no longer meet the needs of modern surveillance. [0003] In response to this p...

AI Technical Summary

Problems solved by technology

The detection method based on human motion features achieves high accuracy by extracting the motion features of the foreground target, but the motion features of the target are not easy to extract and the calculation is complicated; the detection method based on the motion trajectory of the foreground object has a small amount of calculation, but low precision, and is generally used for specific scene

[0005] In general, for the detection of abnormal behavior, there is no particularly efficient method, and continuous exploration is needed

Images