KR101268102B1 - System and method for video surveillance - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video surveillance system and method, wherein the video surveillance system analyzes trajectory data extracted from input video data to determine whether there is abnormal movement. And a track information generating unit for generating track information of the monitoring target, and a normality determining unit determining whether the movement of the monitoring target is normal based on the track information, wherein the track information includes a track position field and a track direction field. And a hierarchical tree comprising a trajectory velocity field.

Description

Video Surveillance Systems and Methods {SYSTEM AND METHOD FOR VIDEO SURVEILLANCE}

The present invention relates to a video surveillance system and method, and more particularly, to a video surveillance system and method for detecting the steady state of the target object according to the trajectory analysis employing hierarchical space-time modeling.

Recently, the importance of the video surveillance system is increasing with increasing awareness of security in major public facilities such as military facilities, roads, ports, and banks. In particular, unlike conventional video surveillance systems that rely solely on human interpretation of the image, the demand for intelligent video surveillance systems that automatically detect abnormal movements in the surveillance area by analyzing the images is increasing. Many studies have been conducted on this.

At this time, the intelligent video surveillance system basically requires accurate video analysis results, but the efficiency of calculating the results without using a lot of resources is also important.

In this regard, Korean Laid-Open Patent Publication No. 2006-0031832 (hereinafter referred to as "prior art 1") discloses a video surveillance system having a behavior analysis and situational awareness function. That is, in order to extract moving objects from the surveillance area, it includes adaptive background subtraction technique, three frame difference technique mixing, and temporal object hierarchical maintenance technique. According to the present invention, a configuration is performed to classify objects and perform behavior analysis through shape analysis and object tracking information of the extracted objects.

Meanwhile, Korean Patent No. 10-0743927 (hereinafter referred to as 'prior art 2') discloses a method of managing aggregate information of data in a spatial data warehouse. In other words, using the HSaR-tree (Historically Summarized Aggregate R-tree) to store the current aggregate information for the monitoring area, add a link to each node for linking the past aggregate information, the past accounting report of each node Disclosed is a configuration for generating and storing a time summary aggregate table for maintaining aggregate information from past to present in a hierarchical structure.

These prior arts are characterized by tracking location information of the monitored object over time from past to present. That is, only the positional information of the trajectory data is focused, which is different from the present invention. The present invention also aims to simultaneously solve the problem of increasing the accuracy and efficiency of a video surveillance system.

The present invention is to solve the problem of improving the accuracy and efficiency of the above-described video surveillance system, the purpose of which is to use the target system without using a lot of system resources even when the image data is insufficient according to the trajectory analysis employing hierarchical space-time modeling The present invention provides a video surveillance system and method for accurately detecting a normal state.

The video surveillance system according to an aspect of the present invention for achieving the above object is to analyze the trajectory data extracted from the input video data to determine whether there is an abnormal movement, to take a picture of the surveillance target moving in the surveillance area A locus information generation unit configured to generate locus information of the surveillance target from video data, and a normality determination unit determining whether the movement of the surveillance target is normal based on the locus information, wherein the locus information includes a locus position field, And a hierarchical tree including a trajectory direction field and a trajectory velocity field.

The video surveillance method according to another aspect of the present invention for achieving the above object is to analyze the trajectory data extracted from the input video data to determine whether there is an abnormal movement, and to monitor the moving target in the surveillance area. A locus information generator configured to generate locus information of the surveillance target from the captured video data, and a normality determination unit configured to determine whether the movement of the surveillance target is normal based on the locus information, wherein the locus information includes a locus position And a hierarchical tree including a field, a trajectory direction field, and a trajectory velocity field.

The present invention provides a video surveillance system and a video surveillance method for detecting a normal state of an object according to a trajectory analysis, which not only increases the accuracy of the result, but also generates effective trajectory information with a small amount of data. The effect of lowering the computational load is obtained. In addition, the present invention can exhibit excellent performance not only in the off-line system is completely equipped with the image data from the beginning, but also in the online system that receives the image data in real time.

1 illustrates the structure of a video surveillance system in accordance with the present invention.
2 shows an example of a surveillance area which is the object of a video surveillance system according to the present invention.
FIG. 3 illustrates a process of extracting trajectory information according to hierarchical space-time modeling in FIG. 2. FIG.
4 shows a data structure for storing the information extracted in FIG.
5 shows the flow of a video surveillance method in accordance with the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Throughout the specification, when a part is referred to as being "connected" to another part, it includes not only "directly connected" but also "electrically connected" with another part in between . Also, when an element is referred to as "comprising ", it means that it can include other elements as well, without departing from the other elements unless specifically stated otherwise.

1 is a block diagram illustrating a video surveillance system 100 in accordance with the present invention.

First, referring to FIG. 1, the video surveillance system 100 includes a trajectory information generator 101 and a normality determination unit 102. The video input unit 111, the trajectory data extracting unit 112, and the alarm generating unit 121 shown in FIG. 1 may be included in the video surveillance system 100 according to the present invention or may be located outside.

The video surveillance system 100 receives trajectory data on the movement of the surveillance object from the video input unit 111 and the trajectory data extractor 112. The trajectory data input may be input in real time simultaneously with image capturing, or may be performed using the recorded image. It can be called an online video surveillance system when it is input at the same time as the shooting, and it can be called an offline video surveillance system when using recorded video.

The trajectory information generating unit 101 generates the trajectory information by processing the input trajectory data, and the normality determining unit 102 determines whether the monitoring target is showing abnormal movement based on the generated trajectory information.

Here, the locus data refers to raw data received by the locus information generator 101 such as position information according to time, and the locus information refers to information output based on the locus information generator 101. The locus information generator 101 is also responsible for updating locus information or exchanging locus information with a storage such as a database.

If the normality determination unit 102 determines that the monitoring target is showing an abnormal movement, the alarm generation unit 121 may send an alarm. At this time, there is no restriction on the type of alarm.

2 shows an example of a surveillance area that is the target of the video surveillance system 100. The surveillance area includes an entrance area and an exit area, and it is determined that the movement of the exit area into and out of the entrance area is normal. If movement to or from other areas is detected, it is determined that the movement is abnormal.

3 and 4 illustrate the flow of extracting the trajectory information from the trajectory data inputted by the trajectory information generating unit 101 and the resulting data structure.

As shown in FIG. 3, the trajectory information generating unit 101 generates a trajectory position field (TPF), a trajectory direction field (TDF), and a trajectory speed field (TSF) from the input data. Are extracted, respectively, to generate a hierarchical tree as shown in FIG. The hierarchical tree descends from the upper level to the lower level, where the node containing the trajectory position field has as its child node the node containing the trajectory direction field, and again the node containing the trajectory direction field contains the node containing the trajectory velocity field. It is created in the order and configuration.

In this case, the locus position field is an uncertain propagation distribution derived from the locus segment data. The trajectory data usually represents positional information over time, and the video surveillance system 100 according to the present invention samples this positional information into trajectory segments having a fixed length. This is a more flexible and more flexible way of representing data than the sampling into points. The present invention uses probability calculation as a method of processing this. The locus position field calculated by the locus information generator 101 and put into the hierarchical tree shown in FIG. 4 is an uncertain propagation distribution of the locus path based on the Gaussian distribution of the locus segment.

Nodes including the trajectory position field value again have nodes having trajectory direction field values as child nodes, which are obtained by extracting direction information from the trajectory data to obtain a Markov Random Field (MRF). Nodes having velocity direction fields that are child nodes thereof also have a Markov random field value calculated from velocity information of the trajectory data.

When the hierarchical tree is completed as described above, the trajectory information generation unit 101 clusters the hierarchical tree hierarchically in order to reduce the data weight and reduce the memory resource usage and the calculation load of the system. That is, the trajectory position field level is first clustered, and then the clustering of the trajectory direction field level and the clustering of the trajectory velocity field level are performed. Clustering can use either the PCA guided K-means algorithm or the online K-means algorithm through principal component analysis.

When the trajectory information generating unit 101 completes the trajectory information, the normality determining unit 102 determines whether there is an abnormal movement while moving from the upper level to the lower level based on the information. The first level is judged first, and when it is judged abnormal, it stops, and when it is judged normal, the lower level is judged. Decisions at each level are made by determining the maximum a posterior (MAP) and then checking whether it exceeds a preset threshold.

5 is a flow diagram of a video surveillance method according to the present invention. After the video input step (S710) and the trajectory data extraction step (S720) of generating the video data and extracting the trajectory data by capturing the surveillance area, the position, direction, and velocity data according to the time of the trajectory of the surveillance target are extracted. . After generating the trajectory information data based on this (S730), it is determined whether the monitoring target shows abnormal movement (S740), and if it is determined that the abnormal movement (S750).

In this case, the result derived in step S730 is a hierarchical tree including a trajectory position field, a trajectory direction field, and a trajectory velocity field. In step S730, the hierarchical tree is generated and clustered.

In step S740, the upper level of the hierarchical tree is first determined, and if it is determined that the abnormal level is abnormal, the determination operation is stopped.

The foregoing description of the present invention is intended for illustration, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present invention. will be. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. For example, each component described as a single entity may be distributed and implemented, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is shown by the following claims rather than the above description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included in the scope of the present invention. do.

100: video surveillance system
111: video input unit
112: trajectory data extraction unit
101: trajectory information generation unit
102: normality determination unit
121: alarm generator

Claims (12)

In the video surveillance system for determining abnormal movement by analyzing the trajectory information extracted from the input video data,
A trajectory information generation unit for generating trajectory information of the surveillance target from video data of a surveillance target moving in a surveillance region; and
It includes a normality determination unit for determining whether the movement of the monitoring target is normal based on the trajectory information,
The trajectory information is a hierarchical tree including a trajectory position field, a trajectory direction field, and a trajectory velocity field.
The hierarchical tree is generated and clustered hierarchically,
The normality determining unit determines the upper level of the hierarchical tree first, and if it is determined to be abnormal, stops the determination operation, and if it is determined that the normal level determines the lower level of the hierarchical tree. The method of claim 1,
The surveillance zone comprises an inlet zone and an outlet zone,
And the normality determining unit determines whether the movement is normal when the movement of the monitoring target starts in the entrance area and ends in the exit area. The method of claim 1,
The structure of the hierarchical tree goes down from the upper level to the lower level.
The node including the trajectory position field has a node including the trajectory direction field as a child node,
And the node including the trajectory direction field has a node including the trajectory velocity field as a child node. The method of claim 1,
And the locus position field is an uncertain propagation distribution derived from locus segment data extracted from the movement of the surveillance object. The method of claim 1,
And the trajectory direction field is a Markov random field derived from trajectory direction data extracted from the movement of the surveillance object. The method of claim 1,
And the trajectory velocity field is a Markov random field derived from trajectory velocity data extracted from the movement of the surveillance object. delete delete The method of claim 1,
And the normality determining unit determines whether or not the movement of the surveillance target is normal by calculating a maximum posterior probability using the trajectory information generated by the trajectory information generating unit and then crossing a threshold value. In the video surveillance method for determining whether there is abnormal movement by analyzing the trajectory information extracted from the input video data,
a) generating trajectory information of the surveillance object from video data of the surveillance object moving in the surveillance region; and
b) determining whether the movement of the monitoring target is normal based on the trajectory information,
The trajectory information is a hierarchical tree including a trajectory position field, a trajectory direction field, and a trajectory velocity field.
The step a) includes generating the hierarchical tree, and hierarchically clustering the generated hierarchical tree,
In step b), the upper level of the hierarchical tree is first determined, and if it is determined to be abnormal, the determination operation is stopped. delete delete

Images