KR102090879B1 - Remote face recognition on long distance and multi channel monitoring system - Google Patents

Abstract

The present invention relates to a system for recognizing a face at a long distance capable of improving reliability by improving an existing system for recognizing a face to classify and recognize faces for unspecified individuals at a long distance. The system for recognizing the face at a long distance and monitoring multiple channels comprises: a fixed type photographing unit; a PTZ camera; a control unit receiving an image from the fixed type photographing unit to detect a face in an area to be monitored, track an area of the detected face, recognize a face against a face DB, and generate an event; and a monitoring unit displaying the event from the control unit as an image.

Description

REMOTE FACE RECOGNITION ON LONG DISTANCE AND MULTI CHANNEL MONITORING SYSTEM

The present invention relates to a face recognition system, and more specifically, by improving a conventional face recognition system so that faces for an unspecified number can be distinguished and recognized at a long distance, remote face recognition and multi-channel monitoring capable of forming a reliability. It aims to provide a system.

In general, security and surveillance video systems such as CCTV and DVR are vulnerable to crime prevention areas around residential areas, schools, and playgrounds that require 24-hour surveillance, and vulnerable locations that cannot be managed by workers at all times, such as banks, department stores, airports, and ports. Establishment of equipment such as surveillance cameras in facilities, etc., and real-time video transmission to the central control center of the remote location connected to the network to display and record, so that the situation manager can prevent various accidents, dangerous situations and crimes through monitoring. In addition to being able to take follow-up measures or remedial measures by collecting various safety checks and other information, it is possible to grasp the anomalies of many surveillance locations with a small number of personnel, as well as providing essential information for future criminal investigations, finding missing persons or missing persons. can do.

In particular, as applications such as threat detection and tracking of real-time objects and collection of evidence of accidents or crimes have been expanded by utilizing CCTV, telebiometrics application technology in wired and wireless information and communication environments and convergence technology combined with physical security technologies such as CCTV It has reached a point when it is absolutely necessary to attempt to develop new technologies. In other words, modeling and implementing a CCTV-based face detection and recognition system security framework model at this point in time that requires the development and standardization of test technology standards for providing reliability to the domestic bio-awareness-based national infrastructure construction project, and implementing it is based on national information protection. And development of bio-recognition standards is an urgent and necessary task to provide reliability and safety.

 To date, face detection and recognition technology has achieved remarkable results, but research on face detection and recognition technology in an environment converged with CCTV is in the beginning. Currently, only faces photographed with CCTV installed in Korea cannot directly reveal the personal characteristics of criminals who show criminal tendencies, and even faces detected and recognized, the recognition rate due to the slowness of face recognition and the effect of light such as lighting There are a number of challenges that need to be overcome because they are vulnerable to reduced recognition rates and criminal arrest due to deterioration, various face poses, and facial expressions.

In order to overcome the limitations of CCTV-based face detection and recognition technology, various face detection and recognition technologies, such as IR-based face recognition technology, which are robust to lighting, have been recently studied. Especially, in order to develop a high-performance face recognition system, it has been discovered through recent research that excellent face detection technology development should be developed first. However, the current face detection method is to find the location of the face in the image. In addition to the problems in face recognition, many problems to be solved, such as face determination, face detection of different sizes, accurate face area search, face detection speed, and result combination, etc. They exist. CCTV (Closed Circuit Television) refers to all communication equipment that only a specific person can receive after transmitting image information collected through a shooting device installed in a certain space through a closed wired / wireless transmission path.

Despite the current security threats caused by CCTV, related laws on CCTV installation and operation are not properly established, and systematic CCTV installation, operation, and management are not implemented.

This CCTV system consists of various types of wired and wireless CCTV cameras connected to the video surveillance server. The CCTV system also supports USB or IP cameras. In particular, in the case of a wireless Wi-Fi camera, it can be installed at any desired location very easily and at low cost. Analog cameras that have been used for a long time can also be used, and the video surveillance server has the ability to digitize the transmitted analog video using a codec algorithm. In general, images are processed in a form suitable for the purpose by a server, and the resulting images are transmitted to various clients such as mobile phones, desktop PCs, laptops / laptops, PDAs through wired and wireless communication networks such as the Internet, and the clients receive the images It will be used for their purposes such as authentication, ID management, crime prevention, evidence collection, terrorism prevention, and access management.

Korean Registered Patent No. 10-1297295 discloses a prior art face recognition security control system, and FIG. 1 is a conceptual diagram for this.

Looking specifically, it comprises a plurality of camera devices (10), repeaters (20), analytical devices (30), and a manager terminal (40), which are CCTVs, and are installed in a plurality of surveillance areas, respectively (11) ) Is made of a switching hub 21 to be connected in bulk. The switching hub 21 receives a surveillance image from a plurality of CCTVs 11 and collectively transmits it to the analysis device 30. The analysis device 30 receives and stores identification target personnel information including blacklist personnel information and detection target information,

It is connected to the repeater 20 and receives the surveillance images of each camera device 10 to recognize the faces of the personnel in the surveillance images. In addition, the analysis device 30 judges whether or not the match is made by comparing the faces of the persons to be identified and the faces of the persons in the surveillance image, and, when matched, the manager or management personnel notifies the identification target, and the manager or management personnel possesses the identification. A pop-up message, an SMS message, and an email message are delivered to the manager terminal 40 that receives the identification information of the target personnel.

As described above, in the case of systems implementing such a conventional face recognition, a problem of recognition accuracy is raised. In particular, in the case of CCTV used in a public space, it not only recognizes images from a long distance, but also a lot of people pass by at a time, which is a practical obstacle to the application of face recognition.

Therefore, the conventional face recognition method shows a certain recognition rate when the camera and the object are matched in a 1: 1 ratio, and is applied in a popular place to use the video wall function and as a multiple, the faces are divided into long distances. The security and control system, which is recognized and displayed, was virtually difficult to appear.

The present invention has been devised to solve the above-mentioned problems, and it is possible to identify short-distance and long-distance human faces and to perform accurate recognition thereof, and remote face recognition that can maximize the efficiency of monitoring during integrated control and It aims to provide a multi-channel monitoring system.

The present invention for solving the above problems, a fixed-type imaging unit (100a) for generating an image signal for the imaging area, PTZ camera (100b) for generating an image signal by tracking and zooming when the coordinates for the identification area are specified, and An image receiving unit 1100 for receiving the data from the fixed photographing unit and the PTZ camera and dataizing it, and a detecting unit 1210 for detecting a face region from the image from the fixed photographing unit and designating an identification region, and an identification region An image recognition unit 1220 for recognizing a face by matching an image from the PTZ camera with the face DB 1020, an event processing unit 1400 for generating an event at the time of face recognition, and an identification area to designate the PTZ camera Remote face recognition and multi-channel monitoring system including a control unit 1000 having a management unit 1500 that transmits a tracking and enlargement control signal to the identification area. Provided.

In addition, the control unit includes an image DB 1010 for storing an image from the image receiving unit, and the event processing unit includes a viewpoint setting unit 1410 for designating a face recognition time point and a thumbnail for the face recognition time point. It may include a thumbnail generator 1420 to generate, and a calling unit 1430 to call a recorded image after the face recognition point from the image DB.

On the other hand, the monitoring unit further includes a real-time image from the control unit, face recognition result information, and a recording image, and the event processing unit overlaps the recording region called in the real-time playback region in the monitoring unit. It is preferable to have the overlapping portion 1440 to be provided.

In addition, the monitoring unit includes an output area 2100 having an integrated image display area for displaying real-time images from a plurality of fixed-type imaging units on the partition area 2110 and an information area for displaying face recognition results and thumbnails. can do.

In one embodiment, the monitoring unit may stop playback of a real-time image in a partition area corresponding to any one of the selected thumbnails, and play a recorded image from a time when a thumbnail is generated from the image DB.

According to the present invention, by detecting a face region and acquiring an image for recognition of a face through a camera that is distinguished from each other, the recognition rate for remote people is increased and its reliability can be dramatically improved. In order to monitor multi-channel images, it is possible to maximize the efficiency in the processing of complex information by economically constructing the exposure and reproduction of images for the partition area.

1 is a view showing a prior art security control system.
2 is a configuration diagram of a remote face recognition and multi-channel monitoring system according to the concept of the present invention.
3 is a block diagram for explaining an embodiment of the control unit in the remote face recognition and multi-channel monitoring system of the present invention.
4 is a block diagram showing an embodiment of the output unit of FIG. 3.
5 is a block diagram showing an embodiment of the event processing unit of FIG. 3.
6 is a view showing an area of a screen output to a monitoring unit in the remote face recognition and multi-channel monitoring system of the present invention.

Hereinafter, a remote face recognition and multi-channel monitoring system according to the present invention will be described in more detail with reference to the accompanying drawings.

However, the embodiments described below are only for explaining in detail that the person skilled in the art to which the present invention pertains can easily implement the invention, thereby limiting the protection scope of the present invention. Does not mean

Throughout the specification and claims, when a part includes a certain component, this means that other components may be included, not excluded, unless specifically stated to the contrary.

Basically, the present invention detects a face in an area to be monitored by receiving an image from a fixed-type imaging unit, a PTZ camera, and the fixed-type imaging unit, tracks the detected face region, recognizes a face against a face DB, and detects an event. Provided is a remote face recognition and multi-channel monitoring system including a control unit to generate and a monitoring unit to display an event from the control unit as an image.

The present invention proposes the concept of image control by capturing images from various locations and remotely collecting these image signals from the control unit, and for this purpose, a plurality of cameras can be applied to a plurality of locations.

In the concept of the present invention, the camera is composed of a fixed type camera such as a fixed CCTV and a PTZ camera capable of pan-tilt zoom, and the fixed camera and the PTZ camera each detect and recognize a face area at a remote location. It generates a video signal that can be performed respectively. Specific embodiments thereof will be described later. However, the camera is any type of equipment that can generate a video signal.

In addition, the locations of the cameras may be placed in various places for various purposes, such as security or access management, as well as in public places, as well as in places where people, such as offices or factories, can enter and leave faces. Will be able to.

2 is a configuration diagram for a remote face recognition and multi-channel monitoring system according to the present invention.

The fixed-type photographing unit 100a may be applied with a camera such as a CCTV that can recognize a predetermined position at a relatively long distance and wide angle. At this time, the fixed-type photographing unit 100a does not necessarily mean a fixed position at any one location, and may also change a direction to be directed according to the needs of the administrator. Therefore, it should be understood that the meaning of the fixed type in the present invention is relative to the PTZ camera 100b tracking the identification area in real time.

The PTZ camera 100b tracks and enlarges the coordinates and / or contours of the identification area, which is the face area, so that face recognition can be performed when an area of the face is detected according to the generation of the image signal of the fixed image capturing unit 100a. And extract the face. Accordingly, the PTZ camera 100b means a camera having a Pan-Tilt-Zoom function.

In the prior art, image information is generated to detect and recognize a face in one image capturing area through a single CCTV, but in this case, it is as described above that face recognition in a long distance is virtually impossible. Here, the long distance means a distance within about 15m, but such a range may be set according to the performance of the camera.

Therefore, in the present invention, the face can be detected in the photographing area through the fixed type photographing unit 100a, and the face is compared with the DB information through tracking and magnification in the PTZ camera 100b for the detected area. By enabling it, it is possible to recognize the face of a remote person, and a detailed description thereof will be described later.

The fixed photographing unit 100a and the PTZ camera 100b are not necessarily matched 1: 1, and in some cases, a plurality of fixed photographing units 100a and any one PTZ camera 100b are matched. Could be considered. The combination of the fixed-type photographing unit 100a and the PTZ camera 100b may be configured in plural at a remote location according to the number of required locations to transmit a video signal to the control unit 1000 through the relay unit 200.

The relay unit 200 may include all network relay means conforming to the Transmission Control Protocol / Internet Protocol (TCP / IP) communication protocol for efficient management of the remote, and WiBRO (WIreless BROadband internet) and WiMAX (WiBRO) as the wireless Internet. World Interoperability for Microwave Access). In addition, a communication method according to a power line communication (PLC), an RS485 interface or an RS232 communication protocol, or a communication method of various wired or wireless such as Zigbee protocol (IEEE 802.15.4) and Bluetooth communication may be included. . In addition, it is needless to say that the CCTV line itself may constitute a relay network, for example, a wired or wireless connection line from a photographing device.

The control unit 1000 receives the image signals from the fixed type photographing unit 100a and the PTZ camera 100b through the relay unit 200 described above, and in particular, transmits a control command for pan, tilt, or zoom in particular. It performs a function, and functions to output information output from the control unit 1000 to the monitoring unit 2000. The control unit 1000 is preferably composed of one server or a PC, but may be considered when installed in a plurality of locations depending on the case.

The monitoring unit 2000 may mean a display provided in the control room, but may also refer to a mobile terminal, laptop, PC, etc. that can receive and output information from the control unit 1000 by wire or wirelessly at a remote location. . The monitoring unit 2000 may divide and display information from the plurality of fixed photographing units 100a or the PTZ camera 100b or the control unit 1000 into one screen, which will be described later.

3 is a block diagram for explaining the control unit in the remote face recognition and multi-channel monitoring system of the present invention.

'Parts' described in the present invention may mean an algorithm that functions as well as a predetermined device.

The image receiving unit 1100 receives the images of the fixed photographing unit 100a and the PTZ camera 100b for each location, and performs a function of information processing, and can be converted into a predetermined file format. File formats of general video signals are known as MPEG, AVI, ASF, WMV, MOV, VIVO, RM, Yuv4mpeg, RoQ, OGG / OGM, 3gp, etc., and video codecs are MPEG-1 (VCD), MPEG-2 (DVD ), DivX, 3ivX, MS MPEG-4 V1, V2, Xvid Windows Media Video 7/8 (WMV1 / WMV2), Windows Media Video 9 (WMV3), Sorenson 1 (SVQ1), Sorenson3 (SVQ3), Intel Indeo (3.1) , 3.2, 4.1, 5.0), MJPEG, AVID, VCR2, ASV2, VIVO 1.0, 2.0, I263, H263, FLI / FLC, RealVideo 1.0, 2.0, 3.0, 4.0, HuffYUV, H.264 codec It is understood that. The control unit 1000 may be capable of converting video signals of various types of cameras at various locations into a single format so as to be integrated and managed.

In some cases, the image receiving unit 1100 may decode and encode audio signals input at each location to output them together. However, such a voice signal is not necessarily an essential element.

The image of the image receiving unit 1100 is stored in the image DB 1010 in time series, and the stored image may be called based on the event time point and reproduced in the monitoring unit 2000 as will be described later.

First, the detection unit 1210 performs a function of detecting a face region through the input of an image signal from the fixed type imaging unit 100a. Various known techniques may be applied to the detection of the face region, so a detailed description thereof will be omitted.

The thus-detected area is designated as the identification area, and when the coordinates and the outline of the identification area are designated in the entire fixed area imaging unit 100a, the management unit 1500 identifies the PTZ camera 100b through the relay unit 200. The tracking and zooming control signals for the area are transmitted.

Accordingly, the PTZ camera 100b generates a video signal through the pan / tilt / zoom function for the identification area and transmits it to the image receiving unit 1100, and the recognition unit 1220 faces the face image for the identification area. It performs the function of performing face recognition in preparation for the DB 1020.

The output unit 1300 may output such face recognition result information to the monitoring unit 2000, and the recognition time point may be understood as an event occurrence time. The occurrence of the event in the event processing unit 1400 will be described later. At this time, the output unit 1300 may perform a function of outputting a real-time image from the fixed photographing unit 100a, outputting a result at an event point, and playing a video by calling the recorded image.

The management unit 1500 may perform a function of designating an identification area, transmitting a control signal to the PTZ camera 100b, and calling the image DB 1010.

The recognition unit 1220 may not be able to perform recognition because the image extracted from the PTZ camera 100b and the image contrasted with the face DB 1020 do not correspond. In this case, it is preferable that the event processing unit 1400 outputs an unrecognizable result, and the management unit 1500 performs a function of storing and updating the extracted image of the PTZ camera 100b as a new face DB 1020.

FIG. 4 is a block diagram illustrating an embodiment of an output unit in the remote face recognition and multi-channel monitoring system of FIG. 3.

In the case of the monitoring unit 2000 provided in the control room, an administrator can monitor real-time images and information on event processing. To this end, the output unit 1300 continuously outputs images from the fixed-type imaging unit 100a in real time. It may include a real-time video transmission unit 1310 to output. The real-time image transmission unit 1310 may mainly perform a function as a CCTV that outputs an image from the fixed-type imaging unit 100a, but the image of the PTZ camera 100b is real-time according to the administrator's selection or the output method selection. It may also be considered to be output as.

In addition, when an event such as designation of an identification area or completion of face recognition occurs, a recognition result transmission unit 1320 may be displayed to display information about the event. The recognition result transmission unit 1320 functions to display the information of the face DB 1020 matching the image extracted from the video signal of the PTZ camera 100b to the monitoring unit 2000, where there is no matching result value. In this case, the management unit 1500 can store and update the new face DB 1020 as described above.

Meanwhile, the recorded image transmission unit 1330 may be further included, and the recorded image transmission unit 1330 may be stored in a fixed photographing unit from a predetermined point in time according to an administrator's selection from the image DB 1010 or designation of an event occurrence time. A function of playing an image from 100a) or an image from PTZ camera 100b may be performed. In the present invention, as will be described later, it is possible to have a split screen in which multi-channel images can be relayed in real time, and it is efficient that the reproduction of the recorded images is implemented in a manner overlapping with real-time images. Detailed description thereof will be described later.

5 is a block diagram for explaining an embodiment of the event processing unit in the remote face recognition and multi-channel monitoring system of the present invention.

The event time point may mainly refer to a time point at which the identification area is extracted from the video signal from the fixed image capturing unit 100a, but this may mean a case in which the search of the matching face DB 1020 in the recognition unit 1220 is completed. It might be. The event time point is generated by the time point setting unit 1410 and face recognition and a specific playback time point may be designated based on the time point of the event occurrence.

In addition, the thumbnail generation unit 1420 may generate a thumbnail image for easy identification by the administrator based on the time point at which the event occurred in the viewpoint setting unit 1410 so that a call from the image DB 1010 is possible.

In addition, the calling unit 1430, when the administrator searches for an image through the monitoring unit 2000 and calls an image based on a past viewpoint, transmits the video from the corresponding viewpoint as a recorded video from the output unit 1300 You can do it.

Meanwhile, in the present invention, the overlapping unit 1440 may be additionally included, and the overlapping unit 1440 is configured to reproduce a recorded image based on a specific event occurrence time in the playback area of the real-time image of the monitoring unit 2000. Can perform a function. In this case, the playback of the recorded video may be considered even through a separate pop-up window, but the overlapping unit 1440 may increase the efficiency of video management by overlapping the recorded video in the real-time playback area having the corresponding identifier. .

6 is a view showing an area of a screen output to the monitoring unit in the remote face recognition and multi-channel monitoring system of the present invention.

The monitoring unit 2000 may have a predetermined output area 2100, and the output area 2100 may have a divided area for each location or for each shooting area. Each divided region may mean a photographing region of the fixed photographing unit 100a having a predetermined identifier.

The output area 2100 may be divided into an integrated video display area and an information area for displaying its own information such as a menu (no reference number). Basically, since the present invention displays complex images on one display, most of the area can be occupied by the integrated video display area, but this ratio is optional. The integrated image display area includes a divided area, and the divided areas may be implemented as a matrix. Each of the divided image regions may be understood as a partition region 2110 corresponding to each of the fixed-type imaging unit 100a and / or the PTZ camera 100b with a predetermined coordinate value.

Each partition area 2110 functions to output a real-time video image from the real-time video transmission unit 1310 and may include a control area 2120 for controlling each video.

Meanwhile, the recognition result and event information, etc. may be displayed in the information area, where thumbnail images at the event time may be exposed by the thumbnail generator 1420. Recorded images stored from the image DB 1010 can be output according to the administrator's input, and the thumbnails are generated from the image DB 1010 by dragging or clicking a predetermined thumbnail image to the integrated image display area. The video can be played back from the time the event occurred.

At this time, it is preferable that the address of the partition area of the identifier in which the event has occurred is automatically designated, and accordingly, the playback of the real-time video is stopped in the corresponding partition area and the recorded video is played according to the input of the administrator. It may also be considered that the control region 2120 is displayed only when the recorded image is played back.

The remote face recognition and multi-channel monitoring system according to the present invention described above has a recognition rate for remote people compared to the prior art by allowing the detection of the face region and the acquisition of the image for face recognition through a camera that is distinguished from each other and specialized. Increased and its reliability can be improved dramatically.

In addition, for the monitoring of multi-channel images, it is possible to economically arrange the exposure and reproduction of images to the partition area and maximize efficiency in processing complex information.

In the above, the present invention has been described in detail based on the embodiments and the accompanying drawings. However, the scope of the present invention is not limited by the above embodiments and drawings, and the scope of the present invention will be limited only by the contents described in the claims below.

100a ... Fixed type shooting part 100b ... PTZ camera
200 ... relay 1000 ... control
1010 ... Video DB 1020 ... Face DB
1100 ... Video receiving unit 1210 ... Detection unit
1220 ... Recognition section 1300 ... Output section
1310 ... real-time video transmitter 1320 ... recognition result transmitter
1330 ... Recording video transmission unit 1400 ... Event processing unit
1410 ... Point setting unit 1420 ... Thumbnail generation unit
1430 ... Calling part 1440 ... Overlapping part
1500 ... Management Department 2000 ... Client Department

Claims (5)

A fixed type photographing unit 100a generating an image signal for the photographing area;
A PTZ camera 100b for tracking the identification area and generating an enlarged image signal;
The fixed type photographing unit and an image receiving unit 1100 receiving an image signal from a PTZ camera, a detection unit 1210 designating an identification area, a recognition unit 1220 for recognizing a face, and an event at the time of face recognition are generated. A control unit 1000 having an event processing unit 1400 and a management unit 1500 for transmitting a control signal to the PTZ camera;
It includes; a monitoring unit (2000) having an output area (2100) consisting of an integrated video display area and an information area to output real-time video, face recognition result information, and recorded video from the control unit.
The control unit,
The detection unit detects the face region in the imaging area of the video signal that is firstly shot and transmitted from the fixed-type imaging unit and designates it as an identification area. To generate and transmit the image signal in the second order, and recognizer compares the face image with the face DB in the second image signal for the identification area from the PTZ camera, but if there is no matching result, the PTZ camera Save and update the extracted image as new face data,
The integrated video display area,
It is composed of a plurality of partition regions 2110 divided for each location, and the partition regions output a real-time image image to the playback region, but each has a control region 2120 to reproduce real-time images in the partition region according to a control input. A remote face recognition and multi-channel monitoring system that interrupts and reproduces the image by overlapping it in the corresponding partition area from the time when the thumbnail is generated.
According to claim 1,
The control unit,
It has an image DB (1010) for storing the image from the image receiving unit,
The event processing unit,
A viewpoint setting unit 1410 for designating a face recognition time point, a thumbnail generation unit 1420 for generating a thumbnail for a face recognition time point, and a calling unit 1430 for calling a recorded image after the face recognition time point from the image DB Remote face recognition and multi-channel monitoring system having a.
According to claim 1,
The monitoring unit,
A long-distance face recognition and multi-channel monitoring system that displays real-time images from a plurality of fixed-type imaging units in the integrated image display area in the partition area 2110 and displays face recognition results and thumbnails in the information area. delete delete

Images