Suspicious act detecting method based on video analysis

Abstract

The invention relates to a suspicious act detecting method based on video analysis. The method provided by the invention comprises the following three steps: detecting a human body objective, modeling according to track of the human body objective, and extracting and classifying features of the human body objective. The method provided by the invention adopts a computer-assisted means and a video analysis technology to intelligently detect suspicious acts in a monitoring video for timely detection and early warning, so that the threat of suspicious actions to a monitored place can be effectively reduced; and at the same time, the suspicious act detecting method provided by the invention is easy for installation and convenient to use and has obvious economic and social benefits.

Description

technical field [0001] The invention relates to a suspicious behavior detection method based on video analysis. Background technique [0002] At present, cameras widely used in banks, stores, parking lots, etc. usually can only find and investigate suspicious persons through video playback after an abnormal situation occurs, and cannot report to the police in real time. If the suspicious behavior of the human body in the surveillance video can be intelligently detected, the alarm can be reported in time when the incident occurs, and the loss of life and property can be avoided. [0003] There are many existing suspicious behavior detection methods. In "Intelligent Monitoring Algorithm Based on Walking Trajectory", Zhang Ruiyu et al. proposed an abnormal behavior recognition method based on walking trajectory. Moving human body is detected, and whether someone is suspicious is judged by tracking and recording the person’s walking trajectory, but this method mainly detects lo...

AI Technical Summary

Problems solved by technology

[0003] There are many existing suspicious behavior detection methods. In "Intelligent Monitoring Algorithm Based on Walking Trajectory", Zhang Ruiyu et al. proposed an abnormal behavior recognition method based on walking trajectory. Moving human body is detected, and whether someone is suspicious is judged by tracking and recording the person’s walking trajectory, but this method mainly detects loitering behavior and has a single function; Zhang Jin et al. used trajectory The extraction method detects and analyzes loitering events, but this method can only detect loitering behavior and has a single function; Zhou Weibo et al. proposed a video surveillance system targeting pedestrians in "Recognition of Pedestrian Abnormal Behavior Based on Trajectory Feature Analysis", using The trajectory characteristics of pedestrians are used to judge whether abnormal behavior occurs, but the trajectory model elements are simple, and the false alarm rate and missing alarm rate for complex behaviors are high; Hu Weiming et al. The neural network establishes the trajectory model of the moving target, learns the trajectory model through a series of trajectory points, and predicts the direction and position of the moving target at the next moment according to the current trajectory points and model parameters, so as to detect suspicious movement directions of traffic vehicles. location, and whether there are suspicious persons in the parking lot, etc., but this method only uses the direction and position characteristics of the target in the trajectory, and it is difficult to detect more complex suspicious behaviors; Hu Zhilan et al. This paper proposes an abnormal behavior detection method based on the direction of motion, which uses block motion directions to describe different actions, and uses support vector machines to classify abnormal behaviors in real-time surveillance videos. The calculation complexity is small, and real-time monitoring can be realized. The detection effect of target occlusion is poor; Yin Yong et al. proposed an abnormal behavior recognition algorithm based on improved Hu moment in "Abnormal Behavior Recognition Based on Improved Hu Moment", mainly for jumping, accelerating running, falling, squatting, Six suspicious behaviors of waving and holding foreign objects are recognized, but this method needs to extract a relatively fine human body outline, which is difficult to achieve in a complex environment

In general, some existing technologies have a single function and can only detect specific suspicious behaviors; some models are simple and cannot adapt to complex environments; some are not real-time or accurate, and false alarms and Missing alarm phenomenon

Images