KR20190088596A - Statistical population control system based on multiple camera image analysis and its method - Google Patents

Abstract

The present invention relates to a statistical population control system based on multi-camera image analysis and a method thereof, and more particularly, to a statistical population control based on multi-camera image analysis, which is a system capable of controlling population information generated for a predetermined time by calculating a numerical value of a statistical average from information on incoming, maintaining, and outflow population from images input from two or more cameras.

Description

TECHNICAL FIELD The present invention relates to a statistical POPULATION CONTROL SYSTEM,

The present invention relates to statistical population control based on multi-camera image analysis, and more particularly, to statistical population control based on multi-camera image analysis. More particularly, the present invention relates to a statistical population control method for determining a human object in an image input to two or more cameras, And controlling the population information generated for a predetermined period of time.

In general, to classify human objects in image information, various public algorithms such as Histogram of Oriented Gradient (HOG), Histogram of Optical Flow (HOF), Principal Component Analysis (PCA), and Radial Basis Function Neural Networks (RBFNNS) And analyzes the existence of the corresponding object.

Therefore, as shown in Korean Patent Laid-Open No. 10-2016-1631694, "HOG-PCA Feature and RBFNNs Pattern Classifier Detection Method Using Pedestrian Detector", the feature vector is generated from the training input image using the HOG algorithm, and the PCA algorithm And the final pedestrian can be detected using the RBFNNs pattern classifier.

At this time, the image area is divided into a plurality of blocks so that accurate pedestrian position can be measured.

However, since a single camera is installed and classified according to prescribed conditions for classifying human objects, there is a problem that it is difficult to accurately measure human objects located behind or below the obstacle.

In addition, the technique can classify human objects only in a space free from obstructions, and has difficulty in extracting information on the inflow, maintenance, and outflow of human objects, which are active in the image, as numerical values.

Therefore, it is difficult to classify the state according to the activity information of the inflow, maintenance and outflow of the human object in the image.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to provide a method and apparatus for detecting a human object by using a pattern classifier and detecting inflow, And the amount of flow in the specific space where detection is completed is provided as a numerical value.

It is still another object of the present invention to provide a technique for calculating a statistical average of numerical values using two or more cameras, and a statistical average calculation of a higher accuracy when using a plurality of cameras.

Yet another object of the present invention is to provide a current state value and a cumulative distribution value by converting the extracted human object into numerical values according to the state classification of influent, maintenance, and outflow.

Another object of the present invention is to confirm the population activity in a specific space using the population cumulative distribution map.

According to an aspect of the present invention, there is provided a method for acquiring a plurality of camera groups, Analyzing the collected images; Packetizing the analyzed data packet and transmitting it to the gateway; And collecting data transmitted from the gateway at the control server.

According to another aspect of the present invention, there is provided an image processing apparatus comprising: The method comprising: detecting a human object for analyzing the collected image information; analyzing an inflow amount, a holding amount, a flow amount analysis, a statistical average calculation, and a real-time connection through the detected human object; A database for storing numerical values generated by the information analysis unit; A gateway for transmitting the image data and the database; a control server for collecting the composite image and the statistical population information; .

According to another aspect of the present invention, there is provided an information processing apparatus including a database for accumulating and updating an inflow range, a human object inflow state, a real-time confirmation unit, an inflow amount and a maintenance amount increase count, A database for accumulating and updating an incremental count inferring the flow rate; .

The present invention is advantageous in that statistical population control can be performed because the inflow amount, the maintenance amount, and the outflow amount of the population in a specific space are generated based on the data collected through the multi-camera image analysis and stored in the database.

In addition, when statistical population control is performed using two or more cameras, a higher accuracy can be provided in calculating the statistical average.

In addition, there is an advantage that the inflow, maintenance, and outflow states of human objects can be classified and quantified with only image information without a separate sensor mechanism.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram illustrating a configuration of a statistical population control system based on a multi-camera image analysis according to a preferred embodiment of the present invention,
2 is a configuration and a flowchart showing a real-time data processing method according to a preferred embodiment of the present invention,
3 is a flowchart illustrating a method of generating statistical population control data according to a preferred embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will now be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram illustrating a configuration and a connection relationship of a statistical population control system based on multi-camera image analysis according to a preferred embodiment of the present invention. The present invention includes a multi-camera configuration unit 110 installed in a specific space, 210, a transmission unit 310, a gateway 410, a control server 510, and a database 610.

The multiple camera configuration unit 110 installed in the specific space is a device for collecting image information. The detailed configuration includes a specific space classifying unit 111 where multiple cameras are installed, two or more multiple camera configuration units 112, .

The multi-camera configuration unit 110 installed in the specific space supports wired and wireless at the same time. When the power is being input, the wired power is used. However, when the power is cut off, the power is supplied using a separate power supply device Real-time image information is collected.

In the image information collecting apparatus such as the camera, the specific spatial classifying unit 111 can generate an infinite number of randomly generated values as IDs by the user or the manufacturer. The generated ID is stored in the database and used as a primary key to collect information collected in each specific space.

The multi-camera configuration unit 112 includes a camera capable of capturing two or more images. Multiple cameras installed collect statistical population control information with high accuracy by overlapping image recognition range in specific space.

The image analysis unit 210 is configured to have the same number of the specific spatial classifiers 111 as the configuration is completed, and all the functions are included in the multiple camera configuration unit 110.

The image analyzing unit 210 analyzes histograms of an HOG (Histogram of Oriented Gradient), Histogram of Optical Flow (HOF), Principal Component Analysis (PCA), and RBFNNS Radial Basis Function Neural Networks) to detect and digitize human objects.

In this case, the value to be digitized is the real-time position information of the human object detected in the image, the inflow amount of the human object sensed in the set inflow range, the retention amount detected in real time when the inflow human is located in the specific space, And produces a calculated runoff when the human object does not increase the hold amount.

The transmission unit 310 packetizes the data transmitted from the image analysis unit 210 and transmits the packetized data to the gateway 410.

The transmission unit 310 supports wired and wireless at the same time. When the power is being input, the transmission unit 310 uses the wired power. However, when the power is interrupted, the transmission unit 310 receives power using a separate power supply unit and performs real-

The transmission unit 310 is configured to have the same number of the image analysis units 210 as configured, and all the functions are included in the multiple camera configuration unit 110.

The gateway 410 delivers the packet data transmitted from the transmitter 310 to the control server 510.

The gateway 410 supports wired and wireless at the same time. When the power is being input, the gateway 410 uses the wired power. However, when the power is interrupted, the gateway 410 receives the power using a separate power supply and performs real-time transmission and reception .

The control server 510 converts the packetized human object detection information, the inflow amount, the holding amount, and the outflow amount into a physical property value in a table configured in the database, and stores the converted physical object value in the database 610.

2 is a configuration and a flowchart illustrating a real-time data processing method according to a preferred embodiment of the present invention. Basically, this procedure can be performed by the statistical population control system based on the above-described multi-camera image analysis, A statistical average deriving unit 230 using human object detection information, an image synthesizing and outputting unit 240, and a transmitting and receiving unit 250.

The group-based real-time connection checking unit 220 performs a group ID check 221 classified as a specific space and checks connection status of the video information collecting device in the specific space in real time (222).

If no connection is detected at this time, it is repeatedly checked (223) whether or not the connection is continuously checked. If a connection is detected, a pattern classifier to which a public algorithm is applied is executed from the collected image information.

The statistical average deriving unit 230 using the human object detection information may perform the shatter object detection 231 by applying the collected image information to a pattern classifier to which a public algorithm is applied and analyze the inflow amount of the human object 232 in a specific space, , A maintenance amount analysis (233), and an outflow amount analysis (234).

Inflow Analysis 232 collects the human object detection amount in the interval set by the image inflow range, holding amount analysis unit 233 is the image information acquisition unit is installed by adding the human object (Human Object) detection amount from the image overall range interval calculates (235) a statistical average (statistical average) divided by _(Total Camera).

The outflow amount (244) is calculated by subtracting the inflow amount with the count increase when the maintenance amount is decreased from the statistical average derived from the inflow amount analysis (232) in the statistical average derived from the maintenance amount analysis (233) And the statistics are averaged by increasing the count and stored in the database 238.

At this time, if the connection with the image information collecting device is confirmed, the final remaining population value 237 derived from the last image 236 is recorded as final information in the database 238.

The image synthesizing and outputting unit 240 performs notation graphic synthesis 241 on the original image so as to clearly identify the puppet object in the real time connected image information collecting apparatus and transmits the image outputting 242 information to the transceiving unit 243 And transmits the packetized data to the gateway 251.

The gateway 251 receiving the packet data from the transceiver unit 250 delivers all the collected packet data to the control server 252. The control server 252 transmits the packetized human object detection information, The flow rate is converted into a physical property value in a table configured in the database and is stored in the database 253.

FIG. 3 is a flowchart illustrating a method of generating statistical population control data according to a preferred embodiment. This procedure can be basically performed by the above-described real-time data processing method.

In the first step S 230, an image collection device such as multiple cameras installed in a specific space sets a certain pixel range in the image as an inflow range.

In the second step S 231, a pattern classifier using a public algorithm is executed to utilize the human object detected in the inflow range as a basis for determining inflow amount.

The third step (S 232) detects the inflow state of the human object in the inflow range in real time, and performs continuous monitoring when it is not detected.

In a fourth step S 233, the inflow amount is counted in order to grasp the inflow amount of the human object in real time.

At this time, the inflow increase count is stored in the database in real time.

In the fifth step (S 234), it is determined that the human object flowing into the specific space is maintained in the specific space, and the maintenance amount increase count is performed to grasp the total inflow amount.

At this time, the maintenance amount increase count is stored in the database in real time, and the object detection (S 232) is performed in real time to continuously perform the inflow amount and the maintenance amount increase count.

In the sixth step S 235, the state of the maintenance amount is determined, and it is determined whether or not the inflow or outflow occurs when the human object infiltrating into the inflow range is held in the specific space. When a human object is detected in the inflow range, It is judged that the human object is inflowed, and the inflow amount increase count and the maintenance amount increase count are performed and stored in the database.

In the seventh step (S 236), it is determined whether or not the amount of holding of the human object is equal or decreased when the holding amount is not increased as compared with the holding amount recorded in the existing database, as described above. And real-time confirmation is performed by returning to the real-time confirmation of the human object inflow state.

At this time, if the hold amount is decreased, the eighth step (S 237) is performed to count down the amount of maintenance, and it is determined that the object has flowed out from the specific space, and the flow rate increase count and the tenth step (S 239) Lt; / RTI >

The eleventh step (S 240) calculates the statistical average by updating the inflow amount increase count, the maintenance amount increase count, the inflow amount decrease count, and the flow increase count information in real time.

The twelfth step (S 241) updates the information in the database and checks the real-time system connection state. If the connection is maintained, the database information update is continued.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the invention as defined by the appended claims. It is obvious that you can do it.

110: Multiple camera components installed in a specific space
210: Image analysis unit 310: Transmission unit
410: Gateway 510: Control server
610: Database
220: real-time connection confirmation unit for each group 230: statistical average derivation unit
240: image synthesis and output unit 250: transmission and reception unit

Claims (3)

An image analysis unit 210, a transmission unit 310, a gateway 410, a control server 510, and a database 610, which are installed in a specific space.
In the multiple camera configuration unit 110 installed in the specific space, a user or an author can generate an infinite number of randomly generated values as IDs, and the generated IDs are collected in a database to collect information collected in each specific space A multi-camera configuration unit 112 including a specific space classifying unit 111 used as a primary key, and a camera capable of capturing two or more images;
The image analysis unit 210 analyzes the image information collected by the multiple camera configuration unit 110 installed in the specific space. The real-time position information of the human object detected in the image as a value to be digitized using the pattern classifier, An image analysis module that generates an output amount that is calculated when the human object detected in the set inflow range does not increase the holding amount; (211);
A transmitter 310 for packetizing and transmitting the data generated by the image analyzer 210; A gateway 410 for collecting and delivering the information; A database 610 for storing in the table the human object detection information, the inflow amount, the holding amount, and the outflow amount converted into the physical property values; And a statistical population control system based on the multi-camera image analysis. A method and system for real-time data processing for statistical population control based on multi-camera image analysis,
A group-by-group real-time connection checking unit 220, a statistical average deriving unit 230 using human object detection information, a video synthesizing and outputting unit 240, a transceiving unit 250;
The group-based real-time connection checking unit 220 may include a group ID checking unit 221 classified into a specific space, a checking unit 222 in real time for checking the connection status of the image information collecting device in the specific space, An identifying unit 223;
The statistical average deriving unit 230 using the human object detection information applies the collected image information to a pattern classifier to which a public algorithm is applied to obtain a human body object detection unit 231, A statistical averaging unit 235 for collecting human object detection quantities in the entire image range and dividing the detected human object quantities by the number of installed image information collecting apparatuses 235 A determination unit 236, a final residual population value calculation unit 237 derived from the last image, and a primary storage database 238 for transmission, which are connected to the image information collecting apparatus, );
The image synthesizing and outputting unit 240 may include a notation diagram combining unit 241, a video output unit 242, and a transmitting / receiving unit 243 );
A gateway 251 for receiving and delivering packet data from the transmission / reception unit 243, and a control server 252 for converting the packetized data into a property value. The control server 252, the control server 252, A database 253 for storing human object detection information, an inflow amount, a maintenance amount, and an outflow amount in a table; Wherein the real-time data processing method comprises the steps of: As a method of generating statistical population control data,
A step S 230 of setting a certain pixel range in the image as an inflow range by an image collection device such as multiple cameras installed in a specific space;
Executing (S 231) a pattern classifier using a public algorithm to utilize a human object detected in the inflow range as a basis for inflow determination;
(S 232) detecting the inflow state of the human object in the inflow range in real time and performing continuous monitoring if the inflow state is not detected;
(S 233) performing inflow amount increase counting for grasping the total inflow amount by grasping the inflow amount of the human object in real time;
A step (S 234) of determining a total inflow amount by performing a maintenance amount increase count by determining that the human object flowing into the specific space is maintained in the specific space;
The human body object is detected when the human body object is detected in the inflow range and the human body object is inflowed when the holding amount is increased. (S 235) judging the inflow amount increase count and the maintenance amount increase count and storing it in the database;
If human body inflow is detected but compared with the retention amount recorded in the existing database, if the retention amount does not increase, it is judged whether the retention amount is equal or decreased. If the retention amount is the same, (S 236);
(S 237) if the holding amount is decreased, performing the holding amount decrease count and judging that the object has flowed out from the specific space;
Flow rate increase count step (S 238);
The maintenance amount reduction count step (S 239)
(S 240) calculating the statistical average by updating the inflow amount increase count, the maintenance amount increase count, the inflow amount decrease count, and the flow increase count information in real time; (S 241) judging whether the system is connected or not to maintain or suspend the connection; And generating the statistical population control data.

Images