KR102325251B1 - Face Recognition System and Method for Activating Attendance Menu - Google Patents

Abstract

The present invention relates to a face recognition system and method capable of activating a time and attendance menu, wherein the face recognition system according to the present invention stores the work time and attendance management DB according to the authentication time when a determination that the registered user is a registered user is inputted from the face recognition unit. The stored user's T&A items are analyzed to display the default T&A items on the display unit, and if any one of the T&A items is not input through the user input unit, the authentication time is stored in the default T&A item area, and the T&A items are displayed through the user input unit. By including a control unit for storing the authentication time in the time and attendance item area input by the user when any one of them is input, it is possible to reduce errors in time and attendance management while maximizing the advantages of the face recognition technology even in a situation where time and attendance management is required.

Description

{Face Recognition System and Method for Activating Attendance Menu}

The present invention relates to a face recognition system and method capable of activating a time and attendance menu. In particular, during face authentication, the default time and attendance item of the authenticator is displayed on a display unit according to the day and time, thereby maximizing the advantages of face recognition technology and preventing errors in time and attendance management. A face recognition system and method capable of reducing

Face recognition technology is a technology that determines what kind of person a corresponding face is through automatic detection and analysis of the face region in an image including a face, and has been studied for a long time in the field of pattern recognition and computer vision. From the results of many studies, it has recently been applied to real environments such as surveillance systems, immigration control, or biometrics.

Compared to other biometric recognition technologies, the face recognition technology can recognize a face even without a relatively close approach to the face recognition device, so it is widely used in areas where access management is required. In this case, the current time and attendance management is a method in which a person who wants to be authenticated approaches a face recognition device, selects a menu item such as going to work, and then, when the face is authenticated, time is stored together with the menu item.

However, in the method of selecting a menu item for time and attendance management and authenticating a face to an activated menu item, if the first authenticator and the subsequent user's time and attendance items are different, an error may occur in the time and attendance management of the user.

In order to solve the above problems, the present invention provides a face recognition system capable of reducing errors in time and attendance management while maximizing the advantages of face recognition technology by displaying the authenticator's default time and attendance items on the display unit according to the day and time during face authentication, and The purpose is to provide a method.

Also, according to the present invention, when the user's face is authenticated, the T&A menu is activated and displayed on the display unit. An object of the present invention is to provide a face recognition system and method capable of reducing errors.

In order to achieve the above object, a face recognition system according to an embodiment of the present invention includes a camera unit for capturing an image including a face; a face detection unit for detecting a face image from the image captured by the camera unit; a feature extracting unit for extracting feature point information from the face image; a normalization processing unit for generating an aligned face template by normalizing the face image using the feature point information; a face recognition unit that compares the face template input from the normalization processing unit with the registered face template stored in the registration template DB to determine whether the input face template is a registered user; and when a determination that the registered user is a registered user is inputted from the face recognition unit, a default time and attendance item is displayed on the display unit by analyzing the work time according to the authentication time and the user's time and attendance items stored in the T&A management DB, and the T&A item through the user input unit If any one of these is not input, the authentication time is stored in the default T&A item area, and when any one of the T&A items is input through the user input unit, the authentication time is stored in the T&A item area input by the user. control unit is provided.

The face detection unit detects a face image from the image photographed by the camera unit, and if a plurality of faces are included in the image photographed by the camera unit, the face image can be detected and output only for the largest face, The face recognition system may further include a face tracker configured to display a face tracking by superimposing a face region on the captured image displayed on the display unit using region information of the face image input from the face detection unit.

The method of displaying the default time and attendance item on the display unit may display only the default time and attendance item or display it differently from other time and attendance items so that the user can easily check it.

The time and attendance items stored in the time and attendance management DB may include going to work, going out, returning, and leaving the office.

The controller stores the authentication face template in the authentication template DB according to the user's face authentication, and after a predetermined period elapses, using the plurality of authentication face templates stored in the authentication template DB, the authentication face template is applied to each of the plurality of authentication face templates. to obtain a similarity to each other, and to obtain an average similarity value for each of the plurality of authentication face templates using these similarity values, and to obtain a predetermined number of authentication faces in the order of the largest average similarity value among the average similarity values. Templates are displayed on the display unit, and if there is no input for selecting any one of the authentication face templates displayed on the display unit through the user input unit, the authentication face template having the largest average similarity value is selected as a new registered face template. The registered face template can be updated by storing it in the registration template DB.

The control unit may store a template value obtained from the authentication face template through a neural network, and displaying the predetermined number of authentication face templates on the display unit includes a plurality of authentication face templates stored in the authentication template DB. A distance is obtained for each of the plurality of authentication face templates using values of and displaying the values of a predetermined number of authentication face templates on the display unit in the order of the smallest average distance value.

According to another embodiment of the present invention, there is provided a face recognition method using a face recognition system including a time and attendance management DB for storing time and attendance information, the method comprising: photographing an image including a face in a camera unit; detecting a face image from the image taken by the camera unit; extracting feature point information from the face image; generating an aligned face template by normalizing the face image using the feature point information; determining whether the input face template is a registered user by comparing the input face template generated in the generating step with the registered face template stored in the registration template DB; when it is determined that the user is a registered user in the determining step, analyzing the work and time according to the authentication time and the user's attendance items stored in a time and attendance management DB and displaying a default time and attendance item on the display unit; and if any one of the T&A items is not input through the user input unit, the authentication time is stored in the default T&A item area. The above object can be achieved by providing the step of storing the authentication time in an item area.

A face recognition system according to another embodiment of the present invention includes: a camera unit for capturing an image including a face; a face detection unit for detecting a face image from the image taken by the camera unit; a face tracking unit displaying face tracking by superimposing the face area on the captured image displayed on the display unit using area information of the face image input from the face detecting unit; a feature extracting unit for extracting feature point information from the face image; a face recognition unit that compares an input face template using the feature point information with a registered face template stored in a registration template DB to determine whether the input face template is a registered user; and a control unit for activating a time and attendance menu immediately upon input of a determination that the user is a registered user from the face recognition unit and displaying the time and attendance menu on a predetermined area of the display unit, wherein the time and attendance menu includes the face of the authenticator on the display unit and the face tracking unit By being displayed only while displayed in , the above object can be achieved.

With the above configuration, the present invention can reduce errors in time and attendance management while maximizing the advantages of face recognition technology even in a situation where time and attendance management is required.

In addition, the present invention can reduce errors in time and attendance management due to user's erroneous authentication recognition by clarifying that attendance management is performed only for an authenticator whose face tracking function is operating.

The present invention can also improve the speed of authenticating the user's face and reduce authentication failures by displaying an average face image among the authenticated face images and selecting a face image to be newly registered from among these images. can do it

1 is a diagram illustrating a block diagram of a face recognition system according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an example of a table of the attendance management DB shown in FIG. 1 .
FIG. 3 is a diagram illustrating an example of a table of the authentication template DB shown in FIG. 1 .
FIG. 4 is a diagram illustrating template values stored in the template value area shown in FIG. 3 in the form of coordinate axes.
5 is a diagram illustrating a flowchart of a face recognition method according to another embodiment of the present invention.

Hereinafter, preferred embodiments of a face recognition system and method capable of updating a registered face template according to the present invention will be described with reference to the accompanying drawings. For reference, in describing the present invention below, terms referring to the components of the present invention are named in consideration of the function of each component, so it should not be construed as limiting the technical components of the present invention. will be

1 is a diagram showing a block diagram of a face recognition system according to an embodiment of the present invention, FIG. 2 is a diagram showing an example of a table of the attendance management DB shown in FIG. 1), and FIG. 3 is FIG. It is a view showing an example of a table of the authentication template DB shown in Fig. 4 is a view showing the template values stored in the template value area shown in Fig. 3 in the form of a coordinate axis.

1 , the face recognition system includes a template storage unit 110 , a camera unit 120 , a face detection unit 130 , a face tracking unit 132 , a feature extraction unit 140 , and a normalization processing unit 150 . , a face recognition unit 160 , a display unit 170 , a menu generation unit 172 , a user input unit 180 , and a control unit 190 . The face recognition system may be specifically divided into a face processing device (not shown) and a management server (not shown) and operated. For example, in FIG. 1 , the template storage unit 110 includes the attendance management DB 112 , the registration template DB 114 , and the authentication template DB 116 , but the time and attendance management DB 112 is stored in the management server. implemented, and the registration template DB 114 and the authentication template DB 116 may be implemented in the face processing device.

The template storage unit 110 is a storage space for storing time and attendance information and templates, and includes a time and attendance management DB 112 , a registration template DB 114 , and an authentication template DB 116 . The time and attendance management DB 112 may store items necessary for time and attendance management, for example, information such as going to work, going out, returning and leaving work. An example of a table of the attendance management DB 112 is shown in FIG. 2 . The registration template DB 114 may store registered face templates stored for registration, their normalized feature point information, their template values, and time information at which they are stored. The authentication template DB 116 includes authentication face templates according to the user's face authentication, their normalized feature point information, their template values, time information at which they are stored, and size information related to the face detected by the face detection unit 130 , etc. This can be saved. An example of a table of the authentication template DB 116 is shown in FIG. 3 .

The camera unit 120 is installed near the door to take an image including a face.

The face detection unit 130 basically detects a face image existing in the image captured by the camera unit 120 . Here, the image captured by the camera unit 120 is one image including a face. The face detection unit 130 may detect a face image from an image input from the camera unit 120 using a Haar-like feature algorithm. In this case, the face parts (head, eyes, nose, mouth, and beard) can be individually detected and determined as the final face. The face detection unit 130 may improve the final face detection result by finely tuning the face image. When a plurality of faces are included in the image captured by the camera unit 120 , the face detection unit 130 detects and outputs a face image only for the largest face.

The face tracking unit 132 may display the face tracking by superimposing the face region on the captured image of the camera unit 120 displayed on the display unit 170 using the information of the face region input from the face detection unit 130 . have. The face tracking unit 132 may display the face region overlapping the captured image of the camera unit 120 in an oval shape according to the face shape.

The feature extraction unit 140 extracts feature point information from the user's face image. The feature extractor 140 may extract 68 feature points, ie, landmarks, from the face image using, for example, a 3D face geometric model learned in advance.

The feature extraction unit 140 finds eyebrows, eyes, nose, mouth regions and surrounding feature points in the face image and maps them to the face image. Of course, the feature points are not limited to the specific parts listed above. The feature extractor 140 may be specifically implemented as a pattern classifier such as a support vector machine (SVM). The feature extractor 140 may recognize the degree of left and right tilt of the face based on the positions of the two eyes.

The feature extraction unit 140 may also include a configuration for extracting a feature vector from the landmark information, and recognizing the user's facial expression and the current emotional state of the user by using the feature vector. In this case, the feature extraction unit 140 may extract a feature vector from the landmark information by using angle and distance ratio information between the landmarks. In particular, when extracting a feature vector, it is important to reduce the amount of computation by minimizing the dimension of the feature vector. Accordingly, the speed of recognizing the user's facial expression can be improved compared to the existing facial recognition technology.

The feature extraction unit 140 is also configured to express a face image using, for example, the texture features of the face image, the morphological features of the entire face image, and the morphological features of body parts (eyes, nose, mouth, etc.). You can create feature vectors. The feature extractor 140 may also convert the face image from an RGB image to a brightness-chrominance (Y/C) image, and generate a feature vector from the converted image. Here, the color difference image refers to an image in which a difference between colors is expressed by excluding a brightness value and using only pure color information. The feature extractor 140 may generate a feature vector using only the color difference image excluding the brightness image in order to reduce the effect of lighting.

The normalization processing unit 150 generates an aligned face image by normalizing the face image using feature point information extracted from the face image. The normalization processing unit 150 performs affine transformation of the face image using feature point information, that is, a two-dimensional affine transformation or a three-dimensional affine transformation to perform image normalization processing, such as face alignment and luminance correction. . In this case, the face image may be rotated so that the center of the eye and the center of the mouth are vertical, and the size of the face image may be corrected so that the distance between the center of the eye and the center of the mouth is 40 pixels. Accordingly, even if the detected face image is a side view instead of a front view, it is possible to convert this image to generate a front face image.

The face image aligned by the normalization processing unit 150 may be stored in the template storage unit 110 as a face template. In this case, normalized feature point information obtained from the aligned face image may be stored in the template storage unit 110 in a separate area instead of or together with the face template. The normalization processing unit 150 divides the face image into a plurality of blocks with the normalized characteristic point information, and then generates a normalized characteristic vector of each block, and the control unit 190 stores the normalized characteristic vector of each block into the template storage unit 110 . ) can be stored in the normalized feature point information area. In this embodiment, the normalized feature point information obtained from the face template and the aligned face image is distinguished, but the normalized feature point information obtained from the aligned face image is also derived from the aligned face image, so it is included in the face template. Be careful.

The face recognition unit 160 compares the face template of the image input from the camera unit 120 with the registered face template stored in the registration template DB 114 to determine whether the face template of the input image is a registered user. The face recognition unit 160 includes a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), a Restricted Boltzmann Machine (RBM), a deep Various deep learning techniques such as, but not limited to, Deep Belief Networks (DBNs) and Deep QNetworks may be used. In the present embodiment, the face recognition unit 160 may be configured as a convolutional neural network composed of several layers. In this case, a locally connected convolution layer may be used to effectively extract local features in the face region.

The face recognition unit 160 may determine the similarity of the face by using feature point information of the face template. In this case, the face recognition unit 160 may determine the similarity of all registered face templates stored in the face template of the input image, and determine whether the user is a registered user by using each similarity result. For example, if there are four registered face templates stored in the registration template DB 114, and the face similarity is determined to be 97%, 85%, 80%, and 75% for each, the registered user based on the maximum value of 97% cognition can be determined. If the reference value for determining whether the user is a registered user is 95%, since the maximum value of the similarity is 97%, the registered user may be determined.

The face recognition unit 160 specifically calls the face template processed by the normalization processing unit 150 and the registered face template stored in the registration template DB 114 to provide data in block units to the convolutional neural network, and block template for each block. The similarity can be obtained for each block by calculating the value, and the total similarity can be obtained by summing them all up. In this case, the similarity may be calculated using the normalized feature vector or template value of each block stored in the registration template DB 114 .

Meanwhile, the face recognition unit 160 may perform face verification in relation to training. Face verification verifies that two face templates for the same person are the same person. Through this, it is possible to reduce the rate of misrecognition generated in the face recognition process in the image.

The training may train the facial feature points using a plurality of learning models. The plurality of learning models may include at least one of a user's face learning model, another's face learning model, a lighting learning model, and a pose learning model. For example, the training may train the facial feature points by dividing the learning data into at least one region based on the feature points of the face detected from the learning data for the plurality of learning models.

Training may store a representative feature vector for each facial feature point in a buffer as many as the number of facial feature points used in the training data. For example, in the case of the same person due to the characteristics of the face, since the data of the facial feature point is formed around a specific vector, the representative feature vector serving as the center may be stored in a separate buffer to make it the central point of learning.

In training, different images of the same user are input as learning data according to a predetermined number of inputs, and output values for facial feature points, facial feature points for lighting changes, and facial feature points for pose changes can be obtained.

On the display unit 170 , the face region of the face tracking unit 132 is superimposed on the captured image of the camera unit 120 , and thus it appears as if the face is tracked. In addition, the menu item output from the menu generating unit 172 may be displayed on the display unit 170 .

The menu generating unit 172 generates attendance, outing, returning, and leaving the attendance items, and provides it to the display unit 170 . In this case, the menu generating unit 172 may provide only the default T&A item to the display unit 170 , and may also make it possible to better check the current user's T&A status by blinking the corresponding T&A item. In addition, when a touch pad is provided, all of the T&A items may be displayed, but only the default T&A items may be displayed in large and bold colors.

The user input unit 180 may input T&A items, for example, going to work, going out, returning, leaving work. When the default time and attendance item generated by the menu generator 172 and displayed on the display 170 is erroneously displayed, the user may input the correct time and attendance item through the user input unit 180 . The user input unit 180 includes not only buttons disposed outside the display unit 170 , but also a touch pad disposed inside the display unit 170 .

In the registration mode, the control unit 190 stores the generated registration face template processed by the normalization processing unit 150 in the registration template DB 114 . And in the training mode, the controller 190 controls the face recognition unit 160 to compare the verification face template generated by processing in the normalization processing unit 150 of the same user with the registered face template stored in the registration template DB 114. Running training may be performed, and values of face verification according to the training may be set in a convolutional neural network.

In the authentication mode, the control unit 190 controls the face recognition unit 160 to compare the generated face template processed by the normalization processing unit 150 with all of the face templates registered in the registration template DB 114 to obtain the input image. Let the face template determine if it is a registered user.

When the face template of the input image is a registered user, the control unit 190 analyzes the user's time and attendance items stored in the work and attendance management DB 112 according to the authentication time and displays the default time and attendance item on the display unit 170 . do. For example, if the attendance time is until 9 o'clock, if there is no attendance time in the attendance item of the attendance management DB 112 by 10 o'clock, the menu generator 172 may be controlled to generate and display 'go to work' as the default attendance item. can If there is an outing time in the outing item of the attendance management DB 112 from 10:00 to 17:00, the menu generating unit 172 may be controlled to generate and display 'return' as the default time and attendance item. From 17:00, if the face template of the input image is a registered user, the menu generating unit 172 may be controlled to generate and display 'leave work' as the default time and attendance item. However, in this case, if the user's face is continuously input, the leave time is updated, and the time at which the face is authenticated at the end of the day is set as the leave time.

The controller 190 displays the T&A item on a predetermined area of the display unit 170 , but controls the menu generator 172 to display the T&A item only while the face of the authenticator is displayed by the face tracking unit 132 . . Accordingly, it can be clarified that only the current authenticator is in attendance management.

If any one of the T&A items is not input through the user input unit 180 , the controller 190 stores the authentication time in the default T&A item area, and when any one of the T&A items is input through the user input unit 180 , the user The authentication time can be stored in the T&A item entered by .

When the face template of the input image is a registered user, the controller 190 stores the face template and/or normalized feature point information of the input image in the face template area and/or feature point information area of the authentication template DB 116 . Then, the template value calculated through the convolutional neural network in the face recognition unit 160 may be stored in the template value area, and the storage time may be stored in the time information area. Also, the controller 190 may store size information related to the face detected by the face detector 130 in the size information area of the authentication template DB 116 . In this case, the size information may be the maximum horizontal and/or vertical size of the face image or the number of pixels in the face image.

The controller 190 may update the registered face template stored in the registration template DB 114 after a predetermined period, for example, one month has elapsed. The control unit 190 calls the template values of the corresponding user, for example ID1, stored in the authentication template DB 116 . A display showing the template values stored in the template value area shown in FIG. 3 in the form of a coordinate axis is shown in FIG. 4 . Here, blue is the template value of the registered face template currently stored in the registration template DB 114 , and all other colors are template values of the face templates stored in the authentication template DB 116 .

The controller 190 obtains a similarity to each other among template values obtained for a predetermined period, obtains an average similarity value for each of a plurality of authentication face templates using the similarity values, and obtains an average similarity value from among the average similarity values. The face template with the largest template value (green) is selected as the new registered face template. In this case, the controller 190 displays a predetermined number of authentication face templates on the display unit 170 in the order of the largest average similarity value, and the authentication face templates displayed on the display unit 170 through the user input unit 180 . If any one of these is not input, the registered face template may be updated by selecting the authentication face template having the largest average similarity value as a new registered face template and storing it in the registration template DB.

Here, the average similarity value is, for example, when there are three authentication face templates, the similarity between the first authentication face template and the second authentication face template is 82, and the similarity between the second authentication face template and the third authentication face template is 90, and if the similarity between the first authentication face template and the third authentication face template is 86, the average similarity of the first authentication face template is (82+86)/2 = 84, and the average similarity of the second authentication face template is ( 82+90)/2 = 86, and the average similarity of the third authentication face template is (90+86)/2 = 88. Accordingly, the authentication face template having the largest average similarity value among them becomes the third authentication face template.

The controller 190 then stores all information on the face template selected as the registered face template in the registration template DB 114 . In this case, the controller 190 may maintain previously stored information of the registered face template as it is. Accordingly, the controller 190 first processes the face recognition using the newly updated face template in the registration template DB 114, and when the similarity is low, the registered face template previously stored in the registration template DB 114 is further added. can be used for facial recognition.

In order to obtain a template value (green) having the largest average similarity value, the controller 190 obtains a distance from each other using the template values in detail, and the average distance for each of the plurality of authentication face templates using the value of these distances. The value may be obtained, and a face template having a template having the smallest average distance value among the average distance values may be selected as a new registered face template. In this case, the controller 190 may exclude from the registered face template candidates in advance, those whose average values obtained from each other are equal to or greater than a certain threshold. That is, those outside a predetermined deviation from the overall distribution of template values of the face templates stored in the authentication template DB 116 may be excluded from the registered face template candidates. As shown in FIG. 4 , the two template values on the left are closer to the registered face template value than the template values located on the right, but farther from all template values stored in the authentication template DB 116 . Accordingly, the face template selected as the new registered face template may further reflect changes in the face image according to aging or changes in seasons.

The controller 190 may also use the size information of the face image detected by the face detection unit 130 to exclude from the registered face template candidates in advance even when the size information is smaller than a predetermined threshold value. Thereby, it is possible to remove an error due to the normalization process.

The controller 190 may also obtain a registered face template for each predetermined time interval using time information in which the face template is stored. Accordingly, it is possible to reflect the change of the face according to time during the day and the change of illumination that occurs during the night and day section.

5 is a diagram illustrating a flowchart of a face recognition method according to another embodiment of the present invention.

In the registration mode, the control unit 190 stores the generated registration face template processed by the normalization processing unit 150 in the registration template DB 114 (S502).

Since the control unit 190 controls the face recognition unit 160 in the training mode in the training mode, the face recognition unit 160 registers the verification face template generated by processing in the normalization processing unit 150 of the same user to the registration template DB 114 . ) performs deep learning training compared with the registered face template stored in (S504).

Since the controller 190 controls the face recognition unit 160 in the authentication mode in the authentication mode, the face recognition unit 160 registers the face template generated by processing in the normalization processing unit 150 in the registration template DB 114 . It is determined whether the face template of the input image is a registered user by comparing all of the face templates (S506).

If the face template of the input image is a registered user, the control unit 190 analyzes the work hours according to the authentication time and the user's attendance items stored in the attendance management DB and displays the default time and attendance items on the display unit 170 (S508) ). In this case, the face tracking unit 132 performs face tracking by superimposing the face region on the captured image of the camera unit 120 displayed on the display unit 170 using the face region information input from the face detection unit 130 . can be displayed If any one of the T&A items is not input through the user input unit 180 , the controller 190 stores the authentication time in the default T&A item area, and when any one of the T&A items is input through the user input unit 180 , the user The authentication time can be stored in the T&A item entered by .

The controller 190 also stores the face template of the image input to the face template area of the authentication template DB 116 as an authentication face template (S510). In this case, the controller 190 stores the normalized feature point information in the feature point information area of the authentication template DB 116, stores the template value obtained through the convolutional neural network in the face recognition unit 160 in the template value area, and also The storage time can be stored in the time information area. Also, the controller 190 may store size information related to the face detected by the face detector 130 in the size information area of the authentication template DB 116 .

The controller 190 determines whether a predetermined period, for example, one month has elapsed (S512). When a predetermined period has elapsed, the controller 190 reads a plurality of authentication face templates stored in the authentication template DB 116 ( S514 ).

The controller 190 calculates a distance from each of the plurality of authentication face templates, and determines only values of the authentication face templates whose average distance value is equal to or less than a predetermined average distance value as registration template candidates (S516). In this case, the controller 190 may further determine, as registration template candidates, only values of the authentication face template whose size information stored in the size information area of the authentication template DB is equal to or greater than a predetermined size value.

The controller 190 obtains a degree of similarity for each of the registered template candidates determined from among the plurality of authentication face templates stored in the authentication template DB 116, and displays a predetermined number of authentication face templates in the order of increasing the average similarity value to the display unit 170 ) is displayed (S518).

If any one of the authentication face templates displayed on the display unit 170 is not input through the user input unit 180, the controller 190 selects the authentication face template having the largest average similarity value as a new registered face template, and the user When any one of the authentication face templates displayed on the display unit 170 is input through the input unit 180, the input authentication face template is selected as a new registered face template (S520). In order to select a new registered face template, the controller 190 obtains a distance from each of the values of the authentication face template of the registered template candidates selected from among the values of the plurality of authentication face templates stored in the authentication template DB 116 in detail, A face template having a template value having the smallest average distance value may be selected as the new registered face template.

The controller 190 updates the registered face template by storing the selected new registered face template in the registration template DB 114 (S522).

The embodiments of the present invention described above are merely illustrative of the technical idea of the present invention, and the protection scope of the present invention should be interpreted by the following claims. In addition, a person of ordinary skill in the art to which the present invention belongs will be able to various modifications and variations without departing from the essential characteristics of the present invention, all technical ideas within the scope equivalent to the present invention are should be construed as being included in the scope of the rights of

110: template storage unit 112: time and attendance management DB
114: registration template DB 116: authentication template DB
120: camera unit 130: face detection unit
132: face tracking unit 140: feature extraction unit
150: normalization processing unit 160: face recognition unit
170: display unit 172: menu generation unit
180: user input unit 190: control unit

Claims (11)

a camera unit for capturing an image including a face;
a face detection unit for detecting a face image from the image captured by the camera unit;
a feature extracting unit for extracting feature point information from the face image;
a normalization processing unit generating an aligned face template by normalizing the face image using the feature point information;
a face recognition unit that compares the face template input from the normalization processing unit with the registered face template stored in the registration template DB to determine whether the input face template is a registered user; and
When a determination that the user is a registered user is inputted from the face recognition unit, a default time and attendance item is displayed on the display unit by analyzing the work time according to the authentication time and the user's attendance items stored in the T&A management DB, and the time and attendance items are selected through the user input unit A control unit configured to store the authentication time in the default T&A item area if any one is not input, and to store the authentication time in the T&A item area input by the user when any one of the T&A items is input through the user input unit including,

The control unit is
The authentication face template is stored in the authentication template DB according to the user's face authentication, and after a predetermined period has elapsed, the plurality of authentication face templates are respectively similar to each other using the plurality of authentication face templates stored in the authentication template DB. , obtains an average similarity value for each of the plurality of authentication face templates using these similarity values, and selects a predetermined number of authentication face templates in the order of the largest average similarity value among the average similarity values. If there is no input for selecting any one of the authentication face templates displayed on the display unit through the user input unit, the authentication face template having the largest average similarity value is selected as the new registration face template. and updating the registered face template by storing it in a DB. delete delete delete delete According to claim 1,
The control unit stores the template value obtained from the authentication face template through a neural network,
Displaying the predetermined number of authentication face templates on the display unit is to obtain a distance from each of the plurality of authentication face templates using values of the plurality of authentication face templates stored in the authentication template DB, and the distances An average distance value for each of the plurality of authentication face templates is obtained using a value of A face recognition system comprising displaying. A face recognition method by a face recognition system including a time and attendance management DB for storing time and attendance information, the method comprising:
photographing an image including a face by a camera unit;
detecting a face image from the image taken by the camera unit;
extracting feature point information from the face image;
generating an aligned face template by normalizing the face image using the feature point information;
determining whether the input face template is a registered user by comparing the input face template generated in the generating step with the registered face template stored in the registration template DB;
when it is determined that the user is a registered user in the determining step, analyzing the work and time according to the authentication time and the user's time and attendance items stored in a time and attendance management DB and displaying a default time and attendance item on the display unit; and
If any one of the T&A items is not input through the user input unit, the authentication time is stored in the default T&A item area. A face recognition method comprising the step of storing the authentication time in an area,

storing the authentication face template in the authentication template DB according to the user's face authentication;
obtaining a degree of similarity for each of the plurality of authentication face templates by using the plurality of authentication face templates stored in the authentication template DB after a predetermined period has elapsed;
obtaining an average similarity value for each of the plurality of authentication face templates by using the obtained similarity value;
displaying a predetermined number of authentication face templates on the display unit in the order of the largest average similarity value among the obtained average similarity values; and
If there is no input for selecting any one of the authentication face templates displayed on the display unit through the user input unit, selecting the authentication face template having the largest average similarity value as a new registered face template and storing the selected face template in the registration template DB and updating the registered face template. delete a camera unit for capturing an image including a face;
a face detection unit for detecting a face image from the image taken by the camera unit;
a face tracking unit displaying face tracking by superimposing the face area on the captured image displayed on the display unit using area information of the face image input from the face detecting unit;
a feature extracting unit for extracting feature point information from the face image;
a face recognition unit that compares an input face template using the feature point information with a registered face template stored in a registration template DB to determine whether the input face template is a registered user;
and a control unit for activating a time and attendance menu immediately upon input of a determination that the user is a registered user from the face recognition unit and displaying it on a predetermined area of the display unit;
The time and attendance menu is displayed only while the face of the authenticator is displayed on the face tracking unit on the display unit,

The control unit is
The authentication face template is stored in the authentication template DB according to the user's face authentication, and after a predetermined period has elapsed, the plurality of authentication face templates are respectively similar to each other using the plurality of authentication face templates stored in the authentication template DB. , obtains an average similarity value for each of the plurality of authentication face templates using these similarity values, and selects a predetermined number of authentication face templates in the order of the largest average similarity value among the average similarity values. If there is no input for selecting any one of the authentication face templates displayed on the display unit through the user input unit, the authentication face template having the largest average similarity value is selected as the new registration face template. and updating the registered face template by storing it in a DB. delete delete

Images