JP2014016969A - Face matching system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a face matching system that prevents the same face image from being reported successively a plurality of times as a system which gives notice when a face image of a visitor detected in a store or the like is already registered.SOLUTION: A face image of a visitor photographed in a store is detected and segmented, a feature quantity of the segmented face image based upon a face image registered in a face image database is detected, and a mutual similarity value is calculated for the detected feature quantity. When the calculated similarity value exceeds a predetermined threshold, a face image having the highest similarity value is extracted. Registration information on the face image which is thus extracted is confirmed, and when a flag of VIP is set to the registration information, notification processing for VIP is entered.

Description

  Embodiments described herein relate generally to a face matching system.

  In recent years, the image of a customer visiting a store using a face matching technology is photographed with a camera in the store, the face image is detected from the photographed image, the face matching of the detected face image is performed, and a special customer VIP A system that displays (Very Important person) or a suspicious person on a terminal screen and notifies store staff is known (for example, see Patent Document 1).

JP 2007-287138 A

  However, when a person is detected in the shooting range of the camera in the store, the camera keeps track of the person. Therefore, the system is not limited to a customer or a suspicious person who has already been notified. The store staff may have to check each time whether a new customer has come to the store, which requires confirmation by a human system.

  The present invention has been made to solve the above-described problem. For a face image that has already been notified, the same face image is notified a plurality of times by setting a non-notification time of a collation result. An object of the present invention is to provide a face matching system that never happens.

  To achieve the above object, a face collation system according to claim 1 of the present invention detects an image capturing means for capturing a face image output from a camera, and a face image of the image captured by the image capturing means. Face image detecting means, a face collating means for collating the face image detected by the face image detecting means with the face image stored in the face image database, and the result of collation by the face collating means, the face image database A face image detected by the face image detecting means, a face image recognized by the face collating means, and customer information of the recognized face image when it is recognized as any of the face images stored in A notification means for notifying the customer terminal of information, and a collation result non-notification time setting means for setting a time during which the notification information is not notified to the customer terminal, wherein the notification means includes the notification means Thus the notification information has already been notified, and the comparison result the case of non-in notification time set by the non notification time setting means does not notify the notification information to the customer terminal face recognition system.

1 is a schematic configuration diagram of a face matching system according to a first embodiment. Information display screen of customer terminal shown in FIG. Environment setting screen of the management terminal shown in FIG. Flow chart of start processing of face matching system Flow chart of face matching / notification process VIP notification processing flowchart

  The face matching system according to the present embodiment captures an image output from a camera by connecting a camera, a customer server, a face matching server, a management terminal, and a customer terminal installed in a store or the like via a communication line such as a LAN. The face image is detected by and face collation of the detected face image is performed. If a registered important customer (VIP) or suspicious person is detected as a result of the face matching, the face image is displayed on the customer terminal. At this time, in order not to continuously notify the detected face image, the non-notification time of the collation result is set for the face image that has already been notified. Within this non-notification time, the notified face image is not transmitted to the customer terminal. As a result, it is possible to prevent a problem caused by continuously displaying the notified face image on the customer terminal. Hereinafter, it will be described in detail with reference to the drawings.

  FIG. 1 is an example of a schematic configuration diagram of a face matching system 100 according to the first embodiment. In the face matching system 100, a camera 10, a face matching server 20, a customer server 30, a management terminal 40, and a customer terminal 50 installed in a store or the like are connected through a communication line such as a LAN (Local Area Network).

  A plurality of cameras may be used depending on the monitoring range. In addition, each device constituting the system may be expanded as necessary.

  The camera 10 includes a surveillance camera having a LAN I / F (Inter Face). The camera 10 detects a person while tracking a moving object. The detected person image is transmitted to the face matching server 20 via the LAN.

  The face matching server 20 includes a face matching device 21 and a face image database 22. The face matching device 21 includes an image capturing unit (image capturing unit) that captures an image output from the camera 10, a face image detecting unit (face image detecting unit) that detects a face image of the captured image, and this face image detection. A face collating unit (face collating unit) for collating the face image detected by the unit with the face image registered in the face image database 22.

  The customer server 30 is connected to the customer database 31, and in response to a request from the customer terminal 50, customer personal information (customer information) is registered. Further, customer information registered in the customer database is output in response to a request from a customer terminal described later.

  The management terminal 40 is composed of a PC (Personal Computer) and manages registration and deletion of customer face images.

  The customer terminal 50 is constituted by a PC (Personal Computer), and can perform registration of customer personal information (customer information), display of notification information notified from the face matching server, and the like.

  FIG. 2 is an information display screen 50a of the customer terminal 50 shown in FIG. The notification information including the face image detected by the face image detection unit notified (transmitted) from the face collation server 20, the face image recognized by the face collation unit, and customer information of the recognized face image includes the customer terminal 50 information display screens 50a.

  In the captured image 51, the face image detected by the face image detection unit is displayed.

  The image 52 in the database is a face image recognized by the face matching unit, which is a face image that is face-matched with the photographed image 51 and extracted from the face image data. is there.

  The matching rate 53 displays a similarity value that is a matching rate of face matching between the captured image and the face image having the highest matching rate.

  In the customer information field 54, personal information having the highest collation rate as a result of the face collation is displayed. The displayed customer number, name, date of birth, purchase history, and number of visits to the store are examples of display contents, and the side using this system can set them as necessary.

  The operation button 55 is a button for an operator to operate. Operations such as a print button, → (next page display), and ← (return page display) can be performed.

  FIG. 3 is an environment setting screen 40a of the management terminal 40 shown in FIG. In the setting information field 41, environment setting conditions for face matching can be input. For example, it is possible to set a verification time non-notification time, a VIP / suspicious person flag to be displayed, an alarm type setting, and the like.

(1) Non-notification time of verification result: 10
It is an example when the non-notification time of the collation result is set to 10 minutes.

(2) VIP / suspicious person flag:
1: VIP, 2: Suspicious person, 3: VIP, and whether the face image extracted from the suspicious person face image database is VIP or whether to display when a suspicious person is extracted Alternatively, it is possible to select whether to display when a VIP or a suspicious person is extracted from a pull-down menu.

(3) Alarm type:
1: VIP, 2: Suspicious person alarm type can be selected from a pull-down menu.

  FIG. 4 is a flowchart of activation processing of the face matching system. After the system is started, the environment setting file of the face matching system is read and the operating environment is set (S01).

  After setting the operating environment, the face image (matching template) registered in the face image database is developed on the memory (S02). In this process, the registered face image is developed on the memory in order to perform the similarity calculation process at high speed. This process is performed once when the face matching system is activated, for example, in the morning when the operator activates the face matching system. Alternatively, it is automatically activated by setting a predetermined activation time.

  After the development of the face image (verification template) and personal verification (customer management number, VIP / suspicious person), the face verification process is started. This processing means that a processing program for face matching processing is activated and expanded on the memory. By performing such development on the memory, it becomes possible to perform high-speed processing when face matching needs to be performed.

  Through the above processing, the face collation system activation processing is completed.

  FIG. 5 is a flowchart of face collation / notification processing by the face collation server 20. It is checked whether a processing end request has been received from the operator (S20). As a result of this confirmation, if a request for termination of processing has not been received (No in S20), it is confirmed whether there is a person detection image transmitted from the camera 10 (S21).

  As a result of the confirmation, if there is no detected person image (No in S21), the process loops until a detected image is received. When the detected image is received (Yes in S21), the face image is detected, and the detected face image is cut out (S22).

  Face collation is performed between the cut face image (detected face image) and the face image (collation template) registered in the face image database (S23). The face matching method is performed by detecting a feature amount between a detected face image and a face image registered in the face image database 22 and calculating a mutual similarity value for the detected feature amount. . A face image having the highest similarity value is extracted from the face images having the calculated similarity value exceeding the set threshold value.

  The registration information of the face image extracted in this way is confirmed (S24). In the embodiment, the case where the registration information is registered in the customer database 31 is shown. However, the present invention is not necessarily limited to this, and the same result can be obtained even if the face image is registered in the face matching server 20 for registering the face image. be able to. If the VIP flag is set in the registration information of the face image extracted in this way, the process proceeds to VIP notification processing (S25). On the other hand, when the suspicious person flag is set, the process proceeds to a suspicious person notification process (S26).

  If neither the VIP flag nor the suspicious person flag is set, the process ends.

  In step S21, while waiting for the face matching server 20 to receive the detected image (looping), the camera 10 detects and tracks the store visitors until a person is detected (S30). ) When a person is detected (Yes in S30), the detected image detected is transmitted to the face matching server 20 via the LAN 60. The subsequent processing is as described above.

  FIG. 6 is a flowchart showing the VIP notification process. First, it is confirmed whether the face image has already been notified (S40). As a result of the confirmation, if it has already been notified (Yes in S40), it is confirmed whether it is within the non-notification time of the collation result (S41). As a result of this confirmation, if it is within the non-notification time of the collation result (Yes in S41), the process ends as it is. This is because the verification result has already been notified and is not continuously notified based on the gist of the present invention. On the other hand, if it is not within the non-notification time of the collation result (No in S41), a VIP visitor notification is made on the screen (S42).

  As described above, according to the present invention, when a face image of a customer at a store or the like is detected and the visitor is registered, in the face verification system that notifies registration information together with the face image of the visitor, By setting a non-notification time of the collation result for a face image that has already been notified, it is possible to provide a face collation system in which the same face image is not notified multiple times.

DESCRIPTION OF SYMBOLS 100 Face collation system 10 Camera 20 Face collation server 22 Face image database 30 Customer server 31 Customer database 40 Management terminal 50 Customer terminal 60 Network

Claims (3)

Image capturing means for capturing a face image output from the camera;
Face image detecting means for detecting a face image of the image captured by the image capturing means;
Face collation means for collating the face image detected by the face image detection means with the face image stored in the face image database;
As a result of the collation by the face collating means, when it is recognized as any of the face images stored in the face image database, the face image detected by the face image detecting means, the face image recognized by the face collating means And notification means for notifying the customer terminal of notification information comprising customer information of the recognized face image,
Collation result non-notification time setting means for setting a time not to notify the notification information to the customer terminal;
With
The notification means includes
A face verification system that does not notify the customer terminal of the notification information when the notification information has already been notified by the notification unit and is within the non-notification time set by the verification result non-notification time setting unit.   The notification unit also notifies the customer terminal of the notification information when the notification information has already been notified and is not within the non-notification time set by the verification result non-notification time setting unit. The face matching system according to 1. The verification means includes
Feature amount detection means for detecting a feature amount between a detected face image detected by the face image detection means and a face image registered in the face image database;
Similarity calculation means for calculating mutual similarity with respect to the feature quantity detected by the feature quantity detection means;
A facial image extraction unit that extracts a facial image having the highest similarity among facial images in which the similarity calculated by the similarity calculation unit exceeds a set threshold value,
The customer terminal is
A detected face image detected by the face image detecting means;
A face image extracted by the face image extraction means;
A similarity value between the detected face image and the face image extracted by the face image extracting means;
Registration information of the extracted face image;
The face matching system according to claim 1, wherein:

Images