KR100536747B1 - The video storing system using a ptz camera and the method there of - Google Patents

Abstract

The video security system and method using a PTZ camera according to the present invention can perform a wide range of surveillance using a single PTZ camera and simultaneously capture and record the faces of people in the surveillance zone in high resolution. It can greatly increase the reliability of the security system.

In addition, the present invention can track, detect, and enlarge only a face area of a person using a single PTZ camera, thereby realizing an unmanned and automated AI video security system at low cost.

In addition, the present invention can maximize the effect of the video security system by providing only an intruder's face area along with the alarm signal upon detection of an abnormal intruder.

Description

Video Security System and Method using Pan / Tilt / Zoom Cameras {THE VIDEO STORING SYSTEM USING A PTZ CAMERA AND THE METHOD THERE OF}

The present invention relates to an image security system, and in particular, an image security system using a PTZ camera that can detect, track and photograph only a face image using a PTZ (Pan / Tilt / Zoom) camera to increase the accuracy of security surveillance. And to a method.

Today, video security systems are becoming increasingly important as the need to protect or monitor human or technical resources in a complex and rapidly changing modern society. In particular, the use of video security systems is increasing dramatically for surveillance of visitors and securing evidence in corporations, government offices, and banks.

Surveillance performance of such a video security system is represented by the automation of the system and the resulting video. The need for automation has arisen from the growing security and surveillance of surveillance areas, and from the problems of early security and surveillance systems that require the watcher to monitor all surveillance areas.

For example, alarming and tracking an intruder when an intruder's movement is detected in a surveillance area is an example of intruder detection and tracking automation. This automation is becoming essential to meet the surveillance needs of many complex monitoring areas in modern society.

A typical video security system consists of a large number of stationary cameras to monitor intruders. The structure of such a video security system is a structure that is difficult to achieve automation of a surveillance system such as motion detection and tracking of the intruder.

In a video security system using multiple fixed cameras, motion detection by each stationary camera is possible within the limited field of view of each stationary camera, but it is difficult to automate motion detection of the entire surveillance system, especially advanced automation to continuously track an object. The function is difficult to equip.

In addition, the general video security system used a method of recording and recording a large area using a fixed fixed camera. In this case, the fixed camera has a limited resolution, so even if an intruder's image is captured, the face image of the person captured by the fixed camera drops to a very low resolution, resulting in a poor face discrimination.

As a result, it is difficult to secure the surveillance video of the video security system due to the low resolution, and to expand the application of face recognition due to the low resolution.

In order to secure such disadvantages, a method of introducing a digital camera and increasing the resolution of a fixed camera may be applied, but in this case, the amount of image data increases exponentially, resulting in an increase in recording cost.

In order to compensate for this, a video security system using a single or a plurality of high resolution PTZ cameras in combination with a general camera has been proposed, but it requires at least two or more cameras, which increases the cost of the video security system and expands all moving objects. Since the recording is performed by recording a normal object and an object, only an object is recorded and a face may be missing.

In particular, such a conventional video security system uses a relatively simple method of detecting and tracking a changed part of an image by using a color difference from a recorded background image, and does not provide a selectivity for a human face.

The present invention is to solve the above problems, an object of the present invention by using a PTZ camera using a PTZ camera to selectively record only the face image by detecting and tracking only the face part of the person It is to provide a video security system and method.

According to one aspect of the image security system using a PTZ camera according to the present invention for realizing the above object, the photographing unit for performing the vertical and horizontal rotation and zoom-in / zoom-out to shoot and transmit the image of the surveillance zone; An image capture unit configured to detect and select a face region of the captured image of the surveillance zone, and control the photographing unit to acquire an image to enlarge the selected face region and to compress the obtained image; Contains wealth.

In addition, according to one aspect of the security method of the image security system using a PTZ camera according to the present invention, a detection step of detecting and selecting a face region of the image photographing the surveillance zone, and the position of the selected face image for a predetermined time Tracking the image, and controlling the photographing unit to magnify, obtaining an enlarged face image, compressing and storing the enlarged face image, and detecting a face region of the image of the surveillance zone again after a certain time. And selecting.

Hereinafter, an image security system and method using a PTZ camera of the present invention will be described with reference to the accompanying drawings.

1 is a block diagram showing a schematic configuration of a video security system using a PTZ camera according to a preferred embodiment of the present invention.

As shown, an image security system using a PTZ camera includes a photographing unit 1, an image capturing unit 2, an image compressing unit 3, an image storing and transmitting control unit 4, a database 5, and a communication interface. The part 6 is provided.

The photographing unit 1 adjusts the camera angle up and down, left and right, and wide angle to narrow angle according to the camera control signal (position control signal and zoom-in / zoom-out control signal) transmitted from the image capturing unit. The image is taken, and the captured image is transmitted to the signal converter 21.

The photographing unit uses a PTZ camera whose camera angle is adjustable up and down, left and right, and wide angle to narrow angle.

The image capturing unit 2 detects only a face region of the image photographed by the capturing unit 1, controls the capturing unit 1 so that the detected face region is captured to a predetermined size or more, and at this time, the captured image Send it to the compression unit.

For this operation, the image capturing unit 2 receives a digital image signal from the image capturing unit 2 and a signal converting unit 21 for converting an analog image signal transmitted from the photographing unit 1 into a digital image signal. A face image detection unit 22 for detecting an area, and a face image magnification control unit 23 for controlling the photographing unit 1 to enlarge and photograph the detected face area.

At this time, the image capturing unit 2 divides the operation state of the image security system into a first operation state for detecting a face area of people present in the surveillance area and a second operation state for enlarging the detected face area. .

The image capture unit 2 calls the face image detector 22 in the first operation state, and transmits the digital image signal of the signal converter 21 to the face image detector 22.

When the called face image detector 22 completes the detection of the face region, the image capture unit 2 controls the image capture unit 1 to enlarge and track the detected face region. Call and change the operating state of the video security system to the second operating state.

Then, after magnified tracking for a predetermined time, the image of the face is transferred to the image compression unit 3, and then the operation state of the image security system is changed to the first operation state, and the face image detection unit is configured to detect another face image. Call (22).

The face image detector 22 operating in the first operating state tracks and detects the positions of the faces of people existing in the surveillance area from the digital image signal received through the signal converter 21. One of the detected face images is selected.

The face image enlargement control unit 23 operating in the second operation state calculates a camera PTZ parameter for controlling the selected face to be maintained at the center of the camera and controlling the face image to be enlarged to a predetermined size or more set by the user. The camera control signal (position control signal and zoom-in / zoom-out control signal) is generated and transmitted to the imaging unit 1 according to the calculated camera PTZ parameter to continuously control the imaging unit 1.

The face image enlargement control unit 23 acquires the face image tracked and enlarged in the second operation state through the signal converter 21 for a predetermined time, and at least one or more face images obtained at this time are the image compression unit 3. To send).

As such, the face image detector 22 and the face image magnification control unit 23 of the image capture unit 2 operate while being exclusively switched according to the image security system.

The image compressing unit 3 compresses the face image received from the image capturing unit 3, and the face image storing and transmitting control unit 4 stores the compressed face image in the database 5, where an abnormal intruder exists. In this case, the compressed face image and the alarm signal are transmitted through the communication interface 6.

At this time, a local area network (LAN), a wide area network (WAN), a public switched telephone network (PSTN), a packet switched network (PSDN), an asymmetric digital subscriber line (ADSL), and an integrated communication network (ISDN) are used.

2 is a flowchart illustrating a security method of a video security system using a PTZ camera according to a preferred embodiment of the present invention.

It is checked whether the operating state of the video security system using the PTZ camera is the first operating state (step S1).

As a result of the check, when the operating state of the video security system is the first operating state, the face image detecting unit 22 is operated to track and detect the position of the people face image existing in the surveillance zone. One of the detected face images is selected (step S2).

In order to track and enlarge the face image of the selected person, the face image magnification control unit 23 is called, and the operation state of the video security system is changed to the second operation state (step S3).

The called face image magnification control unit 23 controls the face image to be zoomed in and taken beyond the size set by the user and a position control signal for controlling the face to be maintained at the center of the camera while continuously tracking the positional change of the face of the selected person. The zoom-in / zoom-out control signal is transmitted, and a face image of the selected person is acquired (step S4).

The obtained face image is compressed, the compressed face image is stored in the database 5 (step S5), and the compressed face image is transmitted through a communication network. If an abnormal intruder exists, an additional alarm signal is transmitted through the communication network (step S6).

Checking the passage of time, confirming that a predetermined time has elapsed (step S7), and if the predetermined time has elapsed, reestablish the operating state of the video security system in order to track and detect the position of the faces of people existing in the surveillance area. The operation state is changed to one operation state, and the face image detection unit 22 is called (step S8).

In FIG. 2, for convenience, the operation is terminated after the operation of step S8. However, in the actual video security system, the security method of the video security system is continuously performed while the operating state of the system is repeatedly switched at a predetermined time interval.

Hereinafter, the face detection tracking algorithm used in the face image detection unit 3 used in the image security system using a PTZ camera and the face magnification tracking algorithm used in the face image enlargement control unit 4 will be described in detail.

At this time, the face detection tracking algorithm is operated when the system is in the first operating state in the video security system. In the first operating state, the PTZ camera 1 observes a large area in a fixed state, tracks and detects faces of people appearing on the observation screen, and selects one of the detected faces. Then, the face magnification tracking algorithm is called to magnify and track the face of the selected person.

The face zoom tracking algorithm, which is called, controls the PTZ (Pan / Tilt / Zoom) parameter of the PTZ camera 1 from time to time, so that the PTZ camera 1 continuously tracks and enlarges the face area of the selected person.

After a certain time elapses, the operating state of the video security system is automatically changed back to the first operating state, and the execution right is passed to the face detection tracking algorithm.

3 is a flowchart illustrating a face detection tracking algorithm in detail according to a preferred embodiment of the present invention.

When the operation state of the video security system becomes the first operation state and the face detection tracking algorithm is started, the tracking list is initialized first to an empty state (step S11).

The tracking list is a list of face areas currently being tracked.

In order to receive new images (step S12) and to perform location tracking on the new image for each of the areas registered in the current tracking list, the face color with respect to the position of the face area in the previous image is displayed. The maximum matching position is extracted by matching the most similar positions of the color model YUV (step S13).

A partial image is extracted from the extracted maximum matching position (step S14), and a pattern recognition technique is applied to determine whether the extracted partial image is a human face. To this end, a cosine transform is performed on the extracted partial image to extract feature values (step S15).

The facial similarity is calculated by inputting the extracted feature value to a pattern recognizer such as a neural network trained on the face shape (step S16).

If the face similarity is low according to the calculated neural network evaluation value, it is deleted from the tracking list record, and if the face similarity is high according to the calculated neural network evaluation value, the position of the newly obtained area is updated with the record of the tracking list. (Step S17).

The step for performing location tracking of each area registered in the current tracking list is performed for all tracking areas. Therefore, the tracking of the plurality of regions is performed.

When the tracking of the previously tracked area is completed as described above, the face detection function is called to check whether there is a newly detected face (step S18).

The called face detection function first detects the movement area through steps S19 to S111.

An image difference is calculated using the difference between the currently input image and the image immediately input, and color and pattern information (step S19).

At this time, the calculated image difference pixels are clustered (step S110), and the grouped pixels are detected as a motion region (step S111).

When the movement area is detected as described above, the area most similar to the face is detected as the candidate area through the following steps S112 to S114.

A color feature is extracted for the detected motion region (step S112), and the similarity is calculated by comparing the extracted color feature with the face color model (step S113).

As a result of the calculation, only regions having high similarity to the face color model are selected and detected as candidate regions (step S114).

When the area most similar to the face is detected as the candidate area as described above, only the area whose shape is very similar to the face is registered as the detection result through the following steps S115 to S119.

A partial image of the detected candidate region is extracted (step S115), and a feature value is extracted by performing cosine transform on the extracted partial image (step S116).

The extracted feature value is input to a pattern recognizer such as a neural network trained on a face shape to calculate a face similarity evaluation value (step S117).

Candidate areas with high face similarity are selected and registered as detection results (step S118).

If only the area very similar to the face shape is registered as the detection result, the position tracking is repeated again in step S119.

If the detection result comes out, it is additionally registered in the tracking list (step S119).

When registration is completed in the tracking list, it is checked whether the video security system is terminated, and if the video security system is operating, the process proceeds to step S12 again (step S120).

4 is a flowchart illustrating a detailed face magnification tracking algorithm according to an exemplary embodiment of the present invention.

It transmits the position control signal and moves up and down (Pan / Tilt) so that the face of the person (hereinafter, the target) is at the center of the photographing unit 1 from the image coordinate of the face region (hereinafter, the target area) of the person to be enlarged and tracked. The zoom-in / zoom-out control signal is transmitted to adjust the zoom of the photographing unit 1 such that the target area is displayed on the display unit above the value set by the user (step S21).

The new images are input (step S22), and the color model (YUV) for the face color is matched with the most similar to the tracking procedure in the face detection tracking algorithm, based on the position where the existing face area is located. The maximum matching position is extracted (step S23).

A partial image is extracted from the extracted maximum matching position (step S24), and a pattern recognition technique is applied to determine whether the extracted partial image is a human face. To this end, a cosine transform is performed on the extracted partial image to extract feature values (step S25).

The extracted feature values are input to a pattern recognizer such as a neural network trained on the face shape to calculate face similarity (step S26).

If the face similarity is low according to the calculated neural network evaluation value, the list is deleted from the tracking list. If the face similarity is high according to the calculated neural network evaluation value, the position of the newly obtained area is updated with the record of the tracking list. (Step S27).

As a result of the tracking, when the target moves, the PTZ parameters are recalculated so that the target is centered again (step S28).

The position control signal and the zoom-in / zoom-out control signal according to the recalculated PTZ parameters are transmitted to control the photographing unit 1 (step S29).

When the shooting of the image of the target area is completed, it is confirmed whether a predetermined time has elapsed, and if the predetermined time has not elapsed, the process proceeds to step S22 again. The face detection tracking algorithm is called in order to change to the state and to pass the execution right to the face detection tracking algorithm (step S29).

Although described above with reference to a preferred embodiment of the present invention, those skilled in the art will be variously modified and changed within the scope of the invention without departing from the spirit and scope of the invention described in the claims below I can understand that

As described above, an image security system and method using a PTZ camera according to the present invention can perform a wide range of surveillance using a single PTZ camera and simultaneously record and record the faces of people in the surveillance zone in high resolution. This can dramatically increase the reliability of the security system through face recognition.

In addition, the present invention can track, detect, and enlarge only a face area of a person using a single PTZ camera, thereby realizing an unmanned and automated AI video security system at low cost.

In addition, the present invention can maximize the effect of the video security system by providing only an intruder's face area along with the alarm signal upon detection of an abnormal intruder.

1 is a block diagram showing a schematic configuration of a video security system using a PTZ camera according to a preferred embodiment of the present invention.

2 is a flowchart illustrating a security method of a video security system using a PTZ camera according to a preferred embodiment of the present invention.

3 is a flow chart for explaining in detail the face detection tracking algorithm according to a preferred embodiment of the present invention.

4 is a flowchart illustrating a detailed face zoom tracking algorithm according to a preferred embodiment of the present invention.

<Description of the symbols for the main parts of the drawings>

1: PTZ camera 2: Image capture unit

3: face capture unit 4: image compression unit

5: Image storage and transmission control unit 6: Database

7: communication interface

Claims (11)

In the video security system using a PTZ camera (Pan / Tilt / Zoom), A photographing unit which photographs and transmits an image of a surveillance area by performing vertical rotation and zoom-in / zoom-out; The face region is detected from the image photographing the surveillance zone, and the photographing unit is controlled for a predetermined time so that the detected face region is enlarged and photographed to obtain a plurality of enlarged face images. An image capture unit for detecting a face region of the captured image of the surveillance area; An image compression unit to compress the obtained image; And An image storage and transmission control unit for storing the compressed face image in a database and transmitting the compressed face image to a communication network as an alarm signal, The image capture unit A face image detector for capturing an image of the surveillance area and detecting a face region from the captured image; And And a face image magnification control unit configured to control the photographing unit so that the detected face region is tracked and enlarged and photographed, thereby obtaining a plurality of enlarged face images. delete delete The method of claim 1, wherein the photographing unit, Video security system using a PTZ camera, characterized in that using one PTZ camera. In the security method of a video security system using a PTZ camera, Detecting a face area of the captured image of the surveillance area; Acquiring a plurality of enlarged face images while tracking a location of the selected face image and controlling a photographing unit to enlarge the predetermined face image; Compressing and storing the plurality of enlarged face images, and detecting a face region of the captured image of the surveillance zone when a predetermined time elapses; And Transmitting the compressed face image together with an alarm signal through a communication network; Detecting the face area Initializing the tracking list and receiving a new image; If a new image is input, tracking a position in the new image for each of the areas recorded in the tracking list; Invoking a detection function when the position tracking in the new image is completed; If the detection function is called, using the difference between the current image and the previous image, color information, and pattern information, registering a region having a high face similarity as a detection result; And Security method of a video security system using a PTZ camera comprising the step of recording the detection result added to the tracking list. delete delete 6. The method of claim 5, wherein tracking the location If a new image is input, finding a maximum matching position having the most similar face color information based on the position where the face region exists in the previous image; Extracting a partial image of the maximum matching position and extracting feature values by performing cosine transform on the extracted partial image; Calculating a face similarity evaluation value by inputting the extracted feature value to a pattern recognizer trained on a face shape; And If the evaluation value is better, deleting from the record of the tracking list, and if the evaluation value is high, updating the record of the tracking list. The method of claim 5, wherein the registering of the region having high face similarity as a detection result comprises: If a new image is input, calculating an image difference from the previous image, grouping the calculated image difference pixels, and detecting the clustered pixels as a motion area; Extracting a color feature of the detected motion region and comparing the extracted color feature and face color information to detect a region having high similarity as a candidate region; Extracting feature values by extracting partial images of candidate regions, performing cosine transform on the extracted partial images, extracting feature values, and inputting the extracted feature values into a pattern recognizer trained on face shape to evaluate face similarity And after calculating the value, selecting only a candidate region having a high face similarity evaluation value and registering the candidate region as a detection result. The method of claim 5, wherein the obtaining of the enlarged face image is performed. Transmitting a position control signal for causing the face of the selected person to be at the center of the PTZ camera and a camera control signal for enlarging the face area of the selected person; If a new image is input, tracking a position in the new image for each of the areas recorded in the tracking list; And As a result of the location tracking, when the face area of the selected person is moved, recalculating the camera PTZ parameters so that the face area of the selected person is centered again, and transmitting a camera control signal according to the calculated camera PTZ parameters. Security method of a video security system using a PTZ camera comprising. The method of claim 5, wherein the obtaining of the enlarged face image is performed. If a new image is input, finding a maximum matching position having the most similar face color information based on the position where the face region exists in the previous image; Extracting a partial image at the maximum matching position and extracting feature values by performing cosine transform on the extracted partial image; Calculating a face similarity evaluation value by inputting the extracted feature value to a pattern recognizer trained on a face shape; And If the evaluation value is low, deleting from the record of the tracking list, and if the evaluation value is high, updating the record of the tracking list.