CN102917207A - Motion sequence based abnormal motion vision monitoring system - Google Patents

Abstract

Translated fromChinese

基于运动序列的异常行为视觉监测系统的主要步骤是提取视野中的运动的人、检测人体的运动部位、判断是否为异常动作。首先提取一段视频，利用该段视频进行高斯背景建模，提取视频中运动的人体，然后提取前景中目标的轮廓和矩特征信息，调用智能比对模块，将前景的特征信息与人的异常动作特征库相比对，判断该前景中是否存在人的异常行为，如果是要监测的异常行为，就发出报警，直到人为解除报警；否则继续分析下一帧。本发明以识别人的异常行为为中心，建立一个可以识别出人的异常动作的智能视频监测系统，该系统可从视频中识别出异常的动作行为，并对异常行为信息进行管理。

The main steps of the abnormal behavior visual monitoring system based on motion sequences are to extract the moving people in the field of vision, detect the moving parts of the human body, and judge whether it is an abnormal action. First extract a video, use the video to perform Gaussian background modeling, extract the moving human body in the video, then extract the contour and moment feature information of the target in the foreground, call the intelligent comparison module, and compare the feature information of the foreground with the abnormal movement of the person Compared with the feature database, it is judged whether there is abnormal human behavior in the foreground. If it is an abnormal behavior to be monitored, an alarm will be issued until the alarm is manually released; otherwise, continue to analyze the next frame. The invention focuses on identifying abnormal behaviors of people, and establishes an intelligent video monitoring system that can identify abnormal behaviors of people. The system can identify abnormal behaviors from videos and manage abnormal behavior information.

Description

The abnormal behaviour visual monitoring system of based on motion sequence

Technical field: patent of the present invention relates to image processing, computer vision, pattern recognition and electronic information field, a kind of passing through video content analysis, therefrom identify the swinging arm, kick the people, wrestle of people, embrace and the abnormal operation of fighting such as fall, prevent the generation of contingency, guarantee good public order.

Background technology: existing monitor and control facility, only have supervisory function bit only, can not identify people's abnormal behaviour, inadequate for the quick reaction capability of accident.Intelligent Video Surveillance Technology is to be derived from computer vision technique, is artificial intelligence study's a branch, and it sets up mapping relations between image and iamge description, thereby makes computer and analyze the content of understanding in the video pictures by Digital Image Processing.Intelligent monitoring technology is applied in the management of public place and some special occasions, can improves the efficient of management, prevent the generation of contingency.

Summary of the invention: for deficiency and the improvement that remedies existing supervisory control system has the defectives such as abnormal behaviour that intelligent monitor system can not be identified the people now, the present invention is centered by identification people's abnormal behaviour, set up an intelligent video monitoring system that can identify people's abnormal operation, wherein people's abnormal operation mainly comprises: swing arm, kick, wrestle, embrace and fall etc.This system is that the behavior act parser of combined with intelligent can realize the intelligentized unusual action behavior that identifies from video take network video monitor and control system as the basis, and the video frequency monitoring system that abnormal behaviour information is managed.

For achieving the above object, the technical solution used in the present invention is: the abnormal behaviour visual monitoring system of based on motion sequence, comprise the video information acquisition module (such as rig camera) that connects successively, video information sending module (such as fiber optic transmitter), video information transmission module (such as optical fiber or coaxial cable), video information receiver module (such as fiber optic receiver) and video information analysis and processing module (such as computer server and siren).According to the difference of application scenario, the remote distance video transfer signal demand adopts optical fiber, closely adopts coaxial cable can satisfy transmission requirement.The video information acquisition module with video information transmission to the video information sending module, the video information sending module is transferred to the video information receiver module by the video information transmission module after with video information coding, the video information receiver module with video information transmission to the video information analysis and processing module.

The key step of abnormal operation recognition methods of the present invention is to extract the motive position of the human body that moves in the visual field, human body, determine whether abnormal operation.At first extract one section video, utilize this section video to carry out the Gaussian Background modeling, extract the human body that moves in the video, then the profile of target and moment characteristics information are called intelligent comparing module in the extraction prospect, and the characteristic information of prospect and people's abnormal operation feature database are compared, judge the abnormal behaviour that whether has the people in this prospect, if the abnormal behaviour that will monitor is just sent warning, until artificial the releasing reported to the police; Otherwise continue to analyze next frame.

Being described in detail of several committed steps wherein is as follows:

1. set up people's abnormal operation feature database;

People's abnormal operation feature database is the priori of identification people's abnormal operation, so the behavioural characteristic storehouse of Erecting and improving is conducive to improve the accuracy of the abnormal operation of identifying the people.The method for building up of people's abnormal operation feature database mainly is to collect the photo that different people is done the different visual angles of same abnormal operation, and the feature database that corresponding feature consists of this action is extracted in these actions.

2. extraction video-frequency band, segmentation object;

What video analytics server received from fibre optic receiver in this invention is a bit of video, utilizes this section video to carry out the Gaussian Background modeling, obtains the human body that moves in this video channel, extracts the position of human body or the human body of motion with this from the background of complexity.

3. extraction target signature, the contrast characteristic storehouse is made identification and is judged;

Profile and the moment characteristics of the target that the extraction second step is partitioned into, and the abnormal operation feature in the feature database that this feature and the first step are set up compares, utilize Euclidean distance between feature to judge the similarity of the two, when similarity is enough high, just think that this target is abnormal operation, otherwise do not think that the target of extracting is abnormal operation.

4. send warning, recording exceptional information;

If the 3rd walks out of existing warning, then this section video is saved in assigned address take the date as filename, and ejects corresponding alarm signal at monitoring interface, or send corresponding audible ringing signal.

The present invention sets up an intelligent video monitoring system that can identify people's abnormal operation centered by identification people's abnormal behaviour, this system can identify unusual action behavior from video, and abnormal behaviour information is managed.

Description of drawings:

Fig. 1 is hardware composition frame chart of the present invention.

Fig. 2 is operation principle block diagram of the present invention.

Embodiment:

As shown in Figure 1: hardware composition of the present invention mainly contains: video information acquisition module (such as rig camera), video information sending module (such as fiber optic transmitter), video information transmission module (such as optical fiber or coaxial cable), video information receiver module (such as fiber optic receiver), video information analysis and processing module (such as computer server and siren).According to the difference of application scenario, the remote distance video transfer signal demand adopts optical fiber, closely adopts coaxial cable can satisfy transmission requirement.

The image transmitting that video camera is taken is to fiber optic transmitter, fiber optic transmitter with video information coding after by Optical Fiber Transmission to the fibre optic receiver in the control room, the intelligent video analysis server obtains the video information of each passage video camera from fibre optic receiver by timeslice mode in turn, video information is presented on the monitoring interface of server, meanwhile this section video information is carried out intellectual analysis, if note abnormalities action behavior, then send alarm signal at monitoring interface.This system can utilize commercially available rig camera, fiber optical transceiver and server to form.

The recognition methods of above-mentioned intelligent vision abnormal operation is written as the form of application software, and moves at server.Whether the video information of at first checking each passage can correctly gather, and sets up the abnormal behaviour database; Then operation exception action recognition application program is monitored each passage.

As shown in Figure 2: extract first one section video, utilize this section video to carry out the Gaussian Background modeling, extract the human body that moves in the video, then the profile of target and moment characteristics information are called intelligent comparing module in the extraction prospect, and characteristic information and the property data base of prospect compared, judge the abnormal behaviour that whether has the people in this prospect, if the abnormal behaviour that will monitor is just sent warning, until artificial the releasing reported to the police; Otherwise continue to analyze next frame.

Claims (8)

Translated fromChinese

1.基于运动序列的异常行为视觉监测系统,其特征在于：包括依次连接的视频信息采集模块，视频信息发送模块，视频信息传输模块，视频信息接收模块和视频信息分析处理模块；视频信息采集模块将图像传输给视频信息模块，视频信息模块将视频信息编码后通过视频信息传输模块传输给视频信息接收模块，视频信息接收模块将信息传输给视频信息分析处理模块。1. The abnormal behavior visual monitoring system based on the motion sequence is characterized in that: it includes a video information collection module connected in sequence, a video information sending module, a video information transmission module, a video information receiving module and a video information analysis and processing module; a video information collection module The image is transmitted to the video information module, and the video information module encodes the video information and transmits it to the video information receiving module through the video information transmission module, and the video information receiving module transmits the information to the video information analysis and processing module.2.如权利要求1所述的基于运动序列的异常行为视觉监测系统,其特征在于：所述的视频信息采集模块为监控摄像机。2. The abnormal behavior visual monitoring system based on motion sequence as claimed in claim 1, characterized in that: the video information collection module is a surveillance camera.3.如权利要求1所述的基于运动序列的异常行为视觉监测系统,其特征在于：所述的视频信息发送模块为光纤发送器。3. The abnormal behavior visual monitoring system based on motion sequence as claimed in claim 1, characterized in that: said video information sending module is an optical fiber transmitter.4.如权利要求1所述的基于运动序列的异常行为视觉监测系统,其特征在于：所述的视频信息传输模块为光纤或同轴电缆。4. The abnormal behavior visual monitoring system based on motion sequence as claimed in claim 1, characterized in that: said video information transmission module is an optical fiber or a coaxial cable.5.如权利要求1所述的基于运动序列的异常行为视觉监测系统,其特征在于：所述的视频信息接收模块为光纤接收器。5. The abnormal behavior visual monitoring system based on motion sequence as claimed in claim 1, characterized in that: the video information receiving module is an optical fiber receiver.6.如权利要求1所述的基于运动序列的异常行为视觉监测系统,其特征在于：所述的视频信息分析处理模块为计算机服务器及警报器。6. The abnormal behavior visual monitoring system based on motion sequence as claimed in claim 1, characterized in that: the video information analysis and processing module is a computer server and an alarm.7.基于运动序列的异常行为视觉监测识别方法，采用如权利要求1所述的监测系统，具体步骤是：首先提取一段视频，利用该段视频进行高斯背景建模，提取视频中的运动目标，然后提取前景中目标的轮廓和矩特征信息，调用智能比对模块，将前景的特征信息与人的异常动作特征库相比对，判断该前景中是否存在人的异常行为，如果是要监测的异常行为，就发出报警，直到人为解除报警；否则继续分析下一帧。7. The abnormal behavior visual monitoring and recognition method based on motion sequence adopts the monitoring system as claimed in claim 1, and the specific steps are: first extract a section of video, utilize this section of video to carry out Gaussian background modeling, extract the moving target in the video, Then extract the contour and moment feature information of the target in the foreground, call the intelligent comparison module, compare the feature information of the foreground with the abnormal human action feature library, and judge whether there is any abnormal human behavior in the foreground, if it is to be monitored If abnormal behavior occurs, an alarm will be issued until the alarm is manually released; otherwise, continue to analyze the next frame.8.如权利要求7所述的基于运动序列的异常行为视觉监测识别方法，其特征在于：所述的人的异常动作特征库的建立方法主要是收集不同人做同一异常动作的不同视角的照片，对这些动作提取相应的特征构成该动作的特征库。8. The method for visual monitoring and recognition of abnormal behavior based on motion sequences as claimed in claim 7, characterized in that: the establishment method of the abnormal behavior feature library of people is mainly to collect photos of different perspectives of different people doing the same abnormal behavior , to extract the corresponding features for these actions to form the feature library of the action.

Images