JP6767683B2 - Face recognition device - Google Patents

Description

The present invention relates to a face recognition device, a face recognition method, and a program.

Conventionally, in various systems such as an entry / exit monitoring system and an attendance management system, personal authentication by face authentication has been performed for a user who passes through a gate or the like.

For example, in Patent Document 1 and Patent Document 2, a face image of a person passing through a gate installed at the entrance of an apartment house or a corporate building is acquired, and the degree of similarity with a pre-registered reference face image of an individual is obtained. A face recognition device that performs personal authentication by collating a face image to be authenticated with a reference face image by calculating and comparing with a threshold value is described.

Further, in this type of face recognition device, a configuration is adopted in which the threshold value used for matching is automatically changed.

For example, in Patent Document 1, the threshold value used for collating a person passing through the gate is automatically changed based on the reference time such as the average return time of the person using the gate and the current time. Further, in Patent Document 2, the number of people passing through the gate is measured, and when the number of people is large, the threshold value used for collating the person passing through the gate is set small.

JP-A-2009-3569 JP-A-2007-156541

By the way, in a relatively large-scale facility, a multi-stage security gate may be constructed by partitioning into a plurality of sections according to a plurality of security levels and installing a gate between each section. When personal authentication is performed by face image matching at a gate installed between certain sections in such a multi-stage security gate, the methods described in Patent Document 1 and Patent Document 2 are used according to the usage state of the gate. The matching threshold of the gate will be changed. Therefore, it was difficult to construct an organized security gate by linking multiple gates.

An object of the present invention is to provide a face recognition device that solves the above-mentioned problems.

The face recognition device according to one embodiment of the present invention is
A face image acquisition unit that acquires a face image of a user passing through a first gate installed at a first boundary that divides adjacent sections, and a face image acquisition unit.
A collation unit that performs face recognition based on a threshold value for the face image,
An operating state acquisition unit that acquires the operating state of the second gate installed at the first boundary, and
A threshold value changing unit that changes the threshold value based on the operating state,
Is configured to include.

The face recognition method according to another embodiment of the present invention
Acquire the face image of the user passing through the first gate installed at the first boundary that divides the adjacent sections.
Face recognition is performed on the face image based on the threshold value.
Acquire the operating state of the second gate installed at the first boundary,
The threshold is changed based on the operating state.
It is configured as follows.

Programs according to other embodiments of the present invention
On the computer
The process of acquiring the face image of the user passing through the first gate installed at the first boundary that divides the adjacent sections, and
The process of performing face recognition based on the threshold value for the face image and
The process of acquiring the operating state of the second gate installed at the first boundary and
The process of changing the threshold value based on the operating state, and
Is configured to do.

By having the above-described configuration, the present invention can construct a systematic security gate by linking a plurality of gates.

It is a schematic block diagram of the multi-stage security gate to which this invention is applied. It is a figure which shows the example of the personal authentication method implemented in each gate of a multi-step security gate. It is a figure explaining the outline of the face recognition apparatus which concerns on 1st Embodiment of this invention. It is a block diagram of the face recognition apparatus which concerns on 1st Embodiment of this invention. It is a figure which shows the example of the gate operation state data used in the 1st Embodiment of this invention. It is a figure which shows the example of the threshold value change rule used in the 1st Embodiment of this invention. It is a figure which shows the example of the threshold value change rule used in the 1st Embodiment of this invention. It is a figure which shows the example of the threshold value change rule used in the 1st Embodiment of this invention. It is a flowchart which shows the outline of the operation of the face recognition apparatus which concerns on 1st Embodiment of this invention. It is a flowchart which shows the operation example of the operation state detection part of the face recognition apparatus which concerns on 1st Embodiment of this invention. It is a flowchart which shows the operation example of the threshold value changing part of the face recognition apparatus which concerns on 1st Embodiment of this invention. It is a block diagram of the face recognition apparatus which concerns on 2nd Embodiment of this invention.

Next, an embodiment of the present invention will be described in detail with reference to the drawings.

[First Embodiment]
FIG. 1 is a schematic configuration diagram of a multi-stage security gate to which the present invention is applied. The multi-stage security gate shown in FIG. 1 is composed of a total of 10 gates G1 to G10. Gates G1 and G2 are installed as entrance-only gates at the entrance of the facility site ST, and gates G9 and G10 are installed as exit-only gates at the exit of the site ST. Further, the gates G3 and G4 are installed as entrance-only gates at the entrance of the building BU in the site ST, and the gates G7 and G8 are installed as exit-only gates at the exit of the building BU. Further, the gate G5 is installed as an entrance-only gate at the entrance of the important room IR in the building BU, and the gate G6 is installed as an exit-only gate at the exit of the important room IR. In this multi-stage security gate, the inside of the important room IR is the third section with the highest security level, the inside of the building BU excluding the important room IR is the second section with the next highest security level, and the building BU is excluded. The inside of the site ST is the first section with the next highest security level. The security level outside the site ST is lower than that of the first section.

To enter the important room IR from outside the site ST, the user first enters the site ST through either gate G1 or G2. Next, the user enters the building BU through either the gate G3 or G4, and further enters the important room IR through the gate G5. At this time, personal authentication is performed at the passing gates G1 to G5, and the user can pass through the gate only when the authentication is successful. On the contrary, in order to go out of the site ST from the important room IR, the user first goes out of the important room IR through the gate G6. Next, the user goes out of the building BU through either gate G7 or G8, and then goes out of the site ST through any of gates G9 or G10. At this time, personal authentication is performed at the passing gates G6 to G10, and the user can pass through the gate only when the authentication is successful. In addition, the user can leave the building BU without entering the important room IR, or can leave the site ST without entering the building BU.

Various methods of personal authentication implemented at the gates G1 to G10 can be considered. In the following, for convenience of explanation, it is assumed that the gates G1 to G10 employ the authentication method as shown in FIG. Referring to FIG. 2, face authentication is used at all gates G1 to G10, and gates G1 and G2 adopt two-factor authentication by IC card authentication in addition to face authentication, and gate G5 uses face authentication. In addition, two-factor authentication by password authentication is adopted. Further, FIG. 2 shows which of "large", "medium", and "small" is the initial setting value of the collation threshold value used for face recognition at each gate G1 to G10. "Large" emphasizes safety and can keep the false acceptance rate small. In addition, "small" emphasizes convenience, and the false rejection rate can be kept small. "Medium" is between the two. Here, the collation threshold values are represented by "large", "medium", and "small", but are actually represented by numerical values. At that time, the numerical value corresponding to "large" is the largest, the numerical value corresponding to "medium" is the largest, and the numerical value corresponding to "small" is the smallest. In FIG. 2, the initial setting value of the matching threshold value for face recognition of the gates G1 and G2 is “small”. This is a result of considering that the overall security level is high even if the verification threshold value of face authentication is set to "small" because the gates G1 and G2 perform two-factor authentication of IC card authentication and face authentication. .. For the same reason, the initial setting value of the face recognition verification threshold value of the gate G5 is "small". Further, since the gates G6 to G10 are dedicated exit gates, the initial setting value of the face authentication verification threshold value is "small".

However, if the face recognition verification thresholds of the gates G1 to G10 are fixed at the default values, it is not possible to flexibly respond to changes in the situation. The techniques described in Patent Document 1 and Patent Document 2 are techniques for changing the collation threshold value of each gate G1 to G10 according to the usage state of each gate G1 to G10, but that alone is not sufficient. For example, due to a failure in either face authentication or IC card authentication at gates G1 and G2 that perform two-factor authentication, IC card authentication or one-factor authentication by face authentication has become a situation. In this case, the security of the first section is lower than that in the case of two-factor authentication. Therefore, if the matching threshold value for face recognition of the gates G3 and G4 is left at the initial setting value of "medium", the safety of the second section is also lowered. In addition, if one-factor authentication by password authentication or face authentication occurs due to a failure in either face authentication or password authentication at the gate G5 that performs two-factor authentication, two-factor authentication is performed. Compared to the case of authentication, the security of the third section is reduced. In that case, if the matching threshold value of the face recognition of the gates G3 and G4 is changed to "high" instead of the initial setting value "medium", the safety of the second section is enhanced, and therefore the safety of the third section is increased accordingly. Can be enhanced. In addition to the above-mentioned cases, there are various cases in which it is desirable to change the face recognition collation threshold of the gates G3 and G4 based on the operating state of the other gates. The above has described the case where it is desirable to change the face recognition verification threshold of the gates G3 and G4 based on the operating state of the other gates, but the face recognition matching threshold of the gates other than the gates G3 and G4 may be changed to other gates. Of course, there may be cases where it is desirable to change based on the operating state of the gate. Further, there are cases where it is desirable to change the matching threshold value of face recognition between the gate G3 and the gate G4 according to the operating state of each other. For example, if the gate G4 is stopped and closed due to some kind of obstacle, the only approach road from the first section to the second section is the gate G3, so that there is a high concern that traffic congestion will occur at the gate G3. In that case, if the face recognition verification threshold of the gate G3 is set to "low", the false rejection rate is reduced, the number of passing users per unit time is increased, and traffic congestion can be prevented.

Hereinafter, the gate G3 is taken up as a gate configured to change the matching threshold value of the face authentication based on the operating state of another gate, and the face authentication device applied to the gate G3 will be described in detail as the first embodiment.

FIG. 3 is a diagram illustrating an outline of the face recognition device 100 according to the first embodiment of the present invention. Referring to FIG. 3, the face recognition device 100 according to the present embodiment is a user 141 passing from the first section to the second section through a gate G3 installed between the first section and the second section. It is configured to perform face authentication and transmit the authentication result to the gate G3.

The gate G3 performs a predetermined operation according to the received authentication result. The operation of the gate G3 when the authentication result is received is arbitrary. For example, the gate G3 automatically opens and closes the door attached to the gate G3 according to the authentication result. Alternatively, the gate G3 includes a gate bar that can be opened and closed manually, and unlocks the gate bar for a certain period of time only when the authentication result of successful authentication is received. Alternatively, the gate G3 visually or acoustically outputs a passability message from the alarm attached to the gate G3 according to the authentication result. Further, the gate G3 may be a stationary type or a movable portable type. Further, the gate G3 may be a device independent of the face recognition device 100, or may be a device integrated with the face recognition device 100.

FIG. 4 is a block diagram of the face recognition device 100. Referring to FIG. 4, the face recognition device 100 stores the interface unit 111 with the gate G3, the interface unit 112 with the camera unit 142, the communication interface unit 113, the operation input unit 114, the screen display unit 115, and the storage. It is composed of a unit 120 and an arithmetic processing unit 130.

The interface unit 111 is configured to send and receive signals such as the result of face authentication to and from the gate G3. Signals can be sent and received either by wire or wirelessly.

The interface unit 112 is configured to send and receive signals to and from the camera unit 142. Signals can be sent and received either by wire or wirelessly. As shown in FIG. 3, the camera unit 142 is preset at a position and an angle of view at which the face portion of the user 141 moving from the first section to the second section through the gate G3 can be imaged. The camera unit 142 is an imaging device including a CCD (Charge Coupled Device) element, a CMOS (Complementary Metal-axis Semiconductor) element, or the like that outputs image data obtained by imaging the face portion of a person passing through the gate G3. .. The image data captured by the camera unit 142 is, for example, a black-and-white shade image of 256 gradations, but may be a color image.

The communication interface unit 113 is a communication device that performs data communication with an external device such as a face recognition device or a terminal (not shown) provided for each of the gates G1 to G2 and G4 to G10 other than the gate G3 of FIG. is there. The communication method may be either wired or wireless.

The operation input unit 114 is an input device such as a keyboard or a numeric keypad, and is configured to detect an operator's operation and output it to the arithmetic processing unit 130.

The screen display unit 115 is a screen display device such as an LCD (Liquid Crystal Display) or a PDP (Plasma Display Panel). The screen display unit 115 is configured to display various information such as operation menus on the screen in response to an instruction from the arithmetic processing unit 130.

The storage unit 120 is a storage device such as a hard disk or a memory. The storage unit 120 is configured to store processing information and a program 121 required for various processes performed by the arithmetic processing unit 130.

The program 121 realizes various processing units by being read by the arithmetic processing unit 130 and executed. The program 121 is read in advance from an external device (not shown) or a storage medium (not shown) via a data input / output function such as the communication interface unit 113, and is stored in the storage unit 120.

The main processing information stored in the storage unit 120 includes the reference face image data 122, the collation threshold value 123, the gate operation state data 124, and the collation threshold value change rule 125.

The reference face image data 122 is data in which a reference face image used when performing face authentication and a user ID are associated with each other. In the reference face image data 122, at least one face image of a person having the user ID may be held as the reference face image in association with the user ID. Alternatively, the reference face image data 122 may be associated with the user ID and the facial feature amount extracted from the face image of the person having the user ID as the reference face information. Here, the facial feature amount is a numerical value indicating the features of each part of the facial image for recognizing the positional relationship and shape of each part such as the eyes, nose, mouth, and eyebrows that make up the face, and the similarity judgment between the images. It is used when performing such as.

The collation threshold value 123 is a threshold value used for collating the user's face image captured by the camera unit 142. In the present embodiment, the collation threshold value 123 is common to all user IDs.

The gate operation state data 124 is data representing the operation state of the gates G1 to G10 installed in the facility. FIG. 5 shows an example of gate operation state data 124. In this example, the gate operation state data 124 exists for each gate installed in the facility, and is composed of a gate ID, an operation state, an authentication history, and the number of successful authentications per unit time. The gate ID is gate identification information that uniquely identifies the gate. The operating status indicates whether it is operating or stopped, and if it is operating, it indicates whether it is operating with two-factor authentication or one-factor authentication, and if it is stopped, it is open to freely pass through the gate. Alternatively, it indicates whether the vehicle is closed, which is physically blocked. That is, the operating state represents any of four types: operating with two-factor authentication, operating with one-factor authentication, open, and closed. For example, when the gate cannot be used due to a breakdown or construction, it is said to be closed or open. Further, at the gates G1, G2, and G5 that perform two-factor authentication, when the state is switched to the state of performing one-factor authentication due to a failure or the like, the operating state is switched from operating with two-factor authentication to operating with one-factor authentication.

The authentication history is a history of personal authentication, and a set of a user ID of a person who has succeeded in authentication at the gate and an authentication time is saved as a history. In addition, the authentication history includes information on whether or not a predetermined specific person has been successfully authenticated. The user ID of a person who is regarded as a specific person is predetermined. The number of successful authentications per unit time represents the number of successful authentications in the most recent predetermined time (for example, 10 minutes).

The matching threshold value changing rule 125 is a rule for changing the matching threshold value 123 for face recognition of the gate G3 based on the operating state of another gate. The collation threshold change rule 125 describes a condition part that describes the conditions that the operating state of the other gate should satisfy, and a conclusion that describes the changed value of the face recognition collation threshold value of the gate G3 when the condition of this condition part is satisfied. Includes departments and their rule priorities. However, the collation threshold value change rule 125 is not limited to the rule format including the condition portion and the conclusion portion as described above, and may be any format.

6 to 8 show an example of the matching threshold change rule 125. Rule 125-1 stipulates that if the operating states of both gate G1 and gate G2 are open (condition part), the face recognition verification threshold of gate G3 is set to "large", and its priority is 1. Is. The smaller the number, the higher the priority. That is, priority 1 is the highest. Rule 125-2 stipulates that if both the gate and the gate G2 are operating with two-factor authentication (condition part), the face recognition threshold of the gate G3 is set to "medium", and its priority is It is 4. Rule 125-1 states that if Gate G1 and Gate G2 are shut down and opened due to a failure or construction, the safety of the first compartment will be reduced, which will reduce the safety of the second compartment. This is a rule for preventing a decrease in the security of the second section by setting the verification threshold value of the face recognition of the gate G3 to "large". Further, rule 125-2 is a rule for returning the collation threshold value of face recognition of gate G3 to the initial state when the failure of gate G1 and gate G2 is repaired and the operation is restarted. In the face recognition device that controls the gate G4, the matching threshold value for face recognition of the gate G4 is changed according to the same rules as those of rules 125-1 and 125-2.

Rule 125-3 stipulates that if the operating states of both gate G1 and gate G2 are operating with one-factor authentication (condition part), the matching threshold value for face authentication of gate G3 is set to "large", and its priority is set. Is 1. Rule 125-3 is that when gate G1 and gate G2 are degenerated to one-factor authentication, the security of the first compartment is reduced, which causes the security of the second compartment to be reduced. Therefore, the face recognition of the gate G3 is matched. This is a rule for setting the threshold value to "large" to prevent a decrease in the safety of the second section. When the operating state of the gate G1 and the gate G2 returns to the operating state by the two-factor authentication, the matching threshold value of the face authentication of the gate G3 is returned to the initial state according to the rule 125-2.

Rule 125-5 stipulates that if the operating state of the gate G5 is operating with one-factor authentication (condition part), the matching threshold value of the face authentication of the gate G3 is set to "large", and the priority is 1. Is. Rule 125-6 stipulates that if the operating state of the gate G5 is operating by two-factor authentication (condition part), the matching threshold value of the face authentication of the gate G3 is set to "medium", and the priority is 4 Is. Rule 125-5 is a rule for increasing the security of the second section in order to compensate for the decrease in the security of the third section when the gate G5 is degenerated to one-factor authentication. Further, Rule 125-6 is a rule for returning the collation threshold value of the face authentication of the gate G3 to the initial state when the gate G5 returns to the state of two-factor authentication. In the face recognition device that controls the gate G4, the matching threshold value for face recognition of the gate G4 is changed according to the same rules as those of rules 125-5 and 125-6.

Rule 125-7 stipulates that if the authentication of a specific person is successful at the gate G1 or the gate G2 (condition part), the matching threshold value of the face authentication of the gate G3 is set to "large", and the priority is 2. is there. The user ID of a person who is regarded as a specific person is predetermined. Rule 125-8 stipulates that if the authentication of a specific person is successful at the gate G9 or the gate G10 (condition part), the matching threshold value of the face authentication of the gate G3 is set to "medium", and the priority is 2. is there. Rule 125-7 is a rule for increasing the safety of the second section in which a specific person may enter the first section from the entrance of the site. Further, Rule 125-8 is a rule for returning the verification threshold value of face recognition of the gate G3 to the initial state when a specific person goes out of the premises. In the face recognition device that controls the gate G4, the matching threshold value for face recognition of the gate G4 is changed according to the same rules as those of rules 125-7 and 125-8. In Rule 125-7, the importance of a particular person does not matter. However, the specific person may be divided into a plurality of groups according to the importance, and the change width of the matching threshold value may be adjusted according to the importance of the specific person who entered the first section.

Rule 125-9 stipulates that if the number of successful authentications per unit time at gate G1 and gate G2 is equal to or greater than a predetermined value (condition part), the matching threshold value for face recognition of gate G3 is set to "small", and its priority The degree is 3. Rule 125-10 stipulates that if the number of successful authentications per unit time at gates G1 and G2 falls below a predetermined value (condition part), the face recognition verification threshold of gate G3 is set to "medium". Its priority is 3. Rule 125-9 states that if the flow rate of the user passing through the gates G1 and G2 increases, the flow rate of the user passing through the gate G3 may also increase. Therefore, face recognition is performed before the flow rate of the gate G3 increases. This is a rule for setting the collation threshold value of "Small" to lower the false rejection rate and prevent traffic congestion at the gate G3 in advance. Further, Rule 125-10 is a rule for quickly returning the collation threshold value of face recognition of Gate G3 to the initial state when the inflow amount of users from Gate G1 and Gate G2 decreases. In the face recognition device that controls the gate G4, the matching threshold value for face recognition of the gate G4 is changed according to the same rules as those of rules 125-9 and 125-10.

Rule 125-11 stipulates that if the operating state of the gate G4 is closed (condition part), the matching threshold value for face recognition of the gate G3 is set to "small", and the priority is 3. Rule 125-12 stipulates that if the operating state of the gate G4 is operating with one-factor authentication (condition part), the matching threshold value of the face authentication of the gate G3 is set to "medium", and its priority is 4. .. Rule 125-11 sets the collation threshold for face recognition of gate G3 to "small" because when gate G4 is shut down and closed, the only access path from the first section to the second section is gate G3. This is a rule for lowering the false rejection rate and preventing traffic congestion at the gate G3. Further, Rule 125-12 is a rule for returning the collation threshold value of the face authentication of the gate G3 to the initial state when the closed state of the gate G4 is resolved. In the face recognition device that controls the gate G4, the face recognition threshold value of the gate G4 is changed according to the same rules as those of rules 125-11 and 125-12.

Referring to FIG. 4 again, the arithmetic processing unit 130 is an arithmetic processing unit having a microprocessor such as an MPU and its peripheral circuits. The arithmetic processing unit 130 is configured to realize various processing units in cooperation with the hardware and the program 121 by reading the program 121 from the storage unit 120 and executing the program 121. The processing unit realized by the arithmetic processing unit 130 includes an input unit 131, a face image acquisition unit 132, a collation unit 133, an output unit 134, an operation state detection unit 135, and a threshold value change unit 136.

The input unit 131 is configured to input the reference face image data 122, the collation threshold value 123, and the collation threshold value change rule 125 from the outside through the communication interface unit 113 or the operation input unit 114 and store them in the storage unit 120. Further, the input unit 131 is configured to input the gate operation state data 124 in the initial state of each gate G1 to G10 from the outside through the communication interface unit 113 or the operation input unit 114 and store it in the storage unit 120. For the gate operation status data in the initial state, the corresponding contents are set in the gate ID and the operation status columns, and the columns of the authentication history and the number of successful authentications per unit time are blank.

The face image acquisition unit 132 is configured to input image data obtained by imaging the face of a person to be authenticated from the camera unit 142 and acquire (detect) the face image to be authenticated from the image data. There is. The face image acquisition unit 132 acquires a face image existing in the image data by matching a template showing the outline of a general human face with the image data. In addition to template matching, various known face detection algorithms may be used.

The collation unit 133 calculates the degree of similarity between the face image to be authenticated acquired by the camera unit 142 and the reference face image included in the reference face image data 122 for each user ID, and for each calculated user ID. Based on the result of comparing the similarity of the above with the collation threshold 123, it is determined whether the face image to be authenticated matches any reference face image included in the reference face image data 122, or does not match any reference face image. It is configured to determine. As an example of the degree of similarity, the mutual correlation coefficient between facial images can be used. In that case, the similarity is increased when the face portions of the same person are included in the two-face images, and the similarity is decreased when the face portions of different persons are included in the two-face images. As the similarity between face images, a known similarity calculation technique other than the mutual correlation coefficient can be used. For example, the collation unit 133 extracts a facial feature amount from the face image to be authenticated, and collates the reference face feature amount included in the reference face image data 122 with the face feature amount extracted from the face image to be authenticated. Then, it may be configured to calculate the degree of similarity between the reference face image and the face image to be authenticated.

Further, the collation unit 133 fails to authenticate when there is no reference face image in the reference face image data 122 whose similarity with the face image to be authenticated acquired by the camera unit 142 is equal to or higher than the collation threshold 123. It is configured to generate an authentication result and transmit it to the output unit 134. Further, when there is one or more reference face images whose similarity with the face image to be authenticated acquired by the camera unit 142 is equal to or higher than the matching threshold value 123, the collation unit 133 determines that the user ID has the maximum similarity. It is configured to generate an authentication result of successful authentication including the authentication time and transmit it to the output unit 134 and the operation state detection unit 135.

The output unit 134 is configured to transmit the authentication result generated by the collating unit 133 to the gate G3 through the interface unit 111. Further, the output unit 134 may transmit the authentication result to an external terminal via the communication interface unit 113 and / or display it on the screen display unit 115.

The operation state detection unit 135 is configured to periodically detect data representing the operation state of the gate installed in the facility and store it in the storage unit 120. The operation state detection unit 135 is configured to detect the gate operation state related to the gate G3, and is configured to detect the gate operation state related to the gates G1 to G2 and G4 to G10 other than the gate G3. There is. The operating state detection unit 135 is also referred to as an operating state acquisition unit.

The operation state detection unit 135 detects the gate operation state related to the gate G3 as follows. When the operation state of the gate G3 is input from the administrator through the operation input unit 114 or the communication interface unit 113, the operation state detection unit 135 sets the input operation state as the operation state of the gate operation state data 124 related to the gate G3. It is stored in the storage unit 120. Further, when the authentication result of successful authentication is transmitted from the collation unit 133, the operation state detection unit 135 registers the authentication result in the authentication history of the gate operation state data 124 related to the gate G3. At this time, the operation state detection unit 135 compares the user ID included in the authentication result with the user ID of the person regarded as the specific person, and if they match, the information indicating that the identity authentication of the specific person is successful. Is recorded in the authentication history of the gate operation state data 1240 related to the gate G3. Further, the operation state detection unit 135 calculates the number of successful authentications per unit time based on the authentication result of the authentication success transmitted from the collation unit 133, and registers it in the gate operation state data 124 related to the gate G3.

On the other hand, the operation state detection unit 135 periodically transmits the gate operation state data 124 of the gate G3 stored in the storage unit 120 to the face recognition devices of the gates G1 to G2 and G4 to G10 through the communication interface unit 113. Further, the operation state detection unit 135 receives the gate operation state data 124 of the gates G1 to G2 and G4 to G10 periodically sent from the face recognition devices of the gates G1 to G2 and G4 to G10 to the storage unit 120. The stored gate operation state data 124 of the gates G1 to G2 and G4 to G10 is updated. That is, the operation state detection unit 135 detects the gate operation state related to the gate other than the gate G3 by receiving the gate operation state data 124 periodically transmitted from the face recognition device of the other gate.

The threshold value changing unit 136 is configured to change the matching threshold value 123 based on the gate operation state data 124 and the matching threshold value changing rule 125 stored in the storage unit 120. Specifically, the threshold value changing unit 136 determines whether or not the condition unit of the collation threshold value change rule 125 is satisfied based on the gate operation state data 124, and the conclusion unit of the collation threshold value change rule 125 in which the condition unit is satisfied. The matching threshold value 123 is changed according to the above. When the condition parts of the plurality of matching threshold value changing rules 125 are satisfied at the same time, the threshold value changing unit 136 changes the matching threshold value 123 according to the conclusion part of the matching threshold value changing rule 125 having the highest priority among them. Further, when a plurality of matching threshold value changing rules 125 having the highest priority exist, a predetermined standard is followed. For example, when the criterion of emphasizing safety is preset, the threshold value changing unit 136 has one of the conclusion parts of the plurality of matching threshold value changing rules 125 having the highest priority, the value of the changed matching threshold value 123. Priority is given to the one with the largest value. On the contrary, when the criterion of emphasizing convenience is preset, the threshold value changing unit 136 has one of the conclusion parts of the plurality of matching threshold value changing rules 125 having the highest priority, the value of the changed matching threshold value 123. Priority is given to the one with the smallest value.

FIG. 9 is a flowchart showing an outline of the operation of the face recognition device 100. Referring to FIG. 9, first, the input unit 131 of the face recognition device 100 inputs the reference face image data 122 from the outside and stores it in the storage unit 120 (step S1). Next, the input unit 131 inputs the collation threshold value 123 from the outside and stores it in the storage unit 120 (step S2). Next, the input unit 131 inputs the matching threshold value change rule 125 from the outside and stores it in the storage unit 120 (step S3). Next, the input unit 131 inputs the gate operation state data 124 in the initial state of each gate G1 to G10 and stores it in the storage unit 120 (step S4). The processes of steps S1 to S4 are preparatory processes for starting the operation of face recognition at the gate G3, and may be omitted once they have been performed.

Next, the face image acquisition unit 132 of the face recognition device 100 captures the image data obtained by imaging the face portion of the user 141 moving from the first section to the second section, and captures the image data through the interface unit 112. It is input from the unit 142, and the face image to be authenticated is acquired from the image data (step S4). Next, the collating unit 133 of the face recognition device 100 calculates the degree of similarity between the face image to be authenticated and the reference face image included in the reference face image data 122 and compares it with the matching threshold 123 to be authenticated. It is determined whether the face image of the above matches any of the reference face images included in the reference face image data 122, or does not match any of the reference face images (step S6). At this time, the collation unit 133 generates an authentication result of authentication failure when there is no reference face image in the reference face image data 122 whose similarity with the face image to be authenticated is equal to or higher than the verification threshold 123. It is transmitted to the output unit 134. Further, the collation unit 133 succeeds in authentication including the user ID having the maximum similarity and the authentication time when there is one or more reference face images whose similarity with the face image to be authenticated is the collation threshold value 123 or more. The authentication result of is generated and transmitted to the output unit 134 and the operation state detection unit 135. Next, the output unit 134 of the face recognition device 100 transmits the authentication result to the gate G3 through the interface unit 111 (step S7).

If the authentication is successful, the operation state detection unit 135 registers the authentication result in the authentication history of the gate operation state data 124 related to the gate G3, and records that fact in the authentication history when the recognition of the specific person is successful (step). S8). If the authentication is successful, the operation state detection unit 135 calculates the number of successful authentications per unit time, and updates the number of successful authentications per unit time in the gate operation state data 124 of the gate G3 (step S9). Then, the face recognition device 100 returns to step S5 and repeats the same process as the above-described process.

Further, the face recognition device 100 executes the process shown in FIG. 10 in parallel with the process shown in FIG. Referring to FIG. 10, the operation state detection unit 135 of the face recognition device 100 has received an input from the administrator to change the operation state of the gate G3, and has the time to transmit the gate operation state data of the gate G3. Whether or not the gate operation status data is received from another gate is always detected (steps S11 to S13).

When there is an input from the administrator to change the operating state of the gate G3, the operating state detection unit 135 updates the operating state in the gate operating state data 124 related to the gate G3 stored in the storage unit 120 according to the input. (Step S14), the process returns to step S11, and the same process as described above is repeated. When the time has come to transmit the gate operation state data related to the gate G3, the operation state detection unit 135 transmits the gate operation state data of the gate G3 stored in the storage unit 120 through the communication interface unit 113. The data is transmitted to the face recognition devices of the gates G1 to G2 and G4 to G10 (step S15), and the process returns to step S11 to repeat the same process as described above. The gate operation status data related to the gate G3 is transmitted, for example, when a predetermined time has elapsed from the time of the previous transmission, the gate operation status data of the gate G3 is updated by steps S8, S9, and S14 after the time of the previous transmission. It can be done at times. Further, when the operation state detection unit 135 receives gate operation state data from another gate through the communication interface unit 113, the operation state detection unit 135 updates the gate operation state data related to the other gate stored in the storage unit 120 with the received gate operation state data. Then, the process returns to step S11 and the same process as described above is repeated.

Further, the face recognition device 100 executes the process shown in FIG. 11 in parallel with the process of FIGS. 9 and 10. With reference to FIG. 11, the threshold value changing unit 136 of the face recognition device 100 determines whether or not the time for changing the collation threshold value has arrived (step S21). The collation threshold value can be changed, for example, when a certain time has elapsed from the time of the previous change, or when the gate operation state data 124 of another gate is changed after the time of the previous change.

When it is time to change the collation threshold value, the threshold value change unit 136 examines the collation threshold value change rule 125 in which the condition unit is satisfied (step S22). That is, the threshold value changing unit 136 reads the matching threshold value changing rule 125 stored in the storage unit 120 one by one and collates it with the gate operation state data 124 to determine whether or not the condition unit of the rule is satisfied. judge. When there is no matching threshold value changing rule 125 for which the condition unit is satisfied (YES in step S23), the threshold value changing unit 136 returns to step S21 and repeats the same processing as described above.

On the other hand, when there is one or more matching threshold value changing rules 125 for which the condition unit is satisfied, the threshold value changing unit 136 updates the matching threshold value 123 according to the conclusion unit of the matching threshold value changing rule for which the condition unit is satisfied (step S24). Then, the threshold value changing unit 136 returns to step S21 and repeats the same processing as described above.

As described above, according to the face recognition device 100 according to the present embodiment, the collation threshold value used for face recognition performed at the gate G3 installed between the first section and the second section is set. Automatically change according to the operating status of other gates such as gates G1 and G2 installed between the outside of the site and the first section and gate G5 installed between the second section and the third section. Therefore, it is possible to construct an organized security gate by linking a plurality of gates.

In the above description, the gates G1 to G5 are dedicated entrance gates, and the gates G6 to G10 are dedicated exit gates, but these gates G1 to G10 may be gates that are used for both entrance and exit. The face recognition device of the gate, which is used for both the entrance and the exit, is configured to perform face recognition of the person entering and face recognition of the person leaving. Further, the collation threshold value may be common to the entrance and the exit, or may be set separately.

Further, in the above description, the verification threshold value for face authentication is common to all user IDs, but a verification threshold value may be set for each user ID. When setting the collation threshold value for each user ID, the user ID to be changed may be specified in the collation threshold value change rule. For example, in rule 125-7 of FIG. 7, when the authentication of a specific person is successful at the gates G1 and G2, the matching threshold value of the face recognition of the gate G3 is set to “high”, the matching threshold value corresponding to the specific person is changed. It may be excluded from.

Further, in the above description, the face recognition device corresponding to one-to-one for each gate G1 to G10 is used to authenticate the person passing through each gate G1 to G10, but all the gates or a plurality of a part thereof. The authentication at the gate of the above may be configured to be performed by one common face recognition device.

Further, in the above description, the collating unit 133 collates the face image to be authenticated with the reference face images of all the registered users registered in advance, but the registered users to be collated may be limited. For example, the users who pass through the gate G3 of FIG. 1 are limited to the users who have passed through the gate G1 or the gate G2 and entered the first section. Therefore, the collation unit 133 of the face recognition device corresponding to the gate G3 may perform collation only for the users recorded in the authentication history of the gate operation state data 124 related to the gate G1 and the gate G2. Specifically, the collation unit 133 detects the user recorded in the authentication history of the gate operation state data 124 related to the gate G1 and the gate G2 as a visitor of the first section, and a reference face image for each registered user. Identity authentication is performed by selecting the reference face image of the visitor from the data 122, calculating the similarity between the selected reference face image and the face image of the user passing through the gate G3, and comparing it with the threshold value. I do. Alternatively, in order to further limit the registered users to be collated, the users who went out of the premises from the gate G9 or the gate G2 may be removed from the visitors of the first section detected as described above. Further, the user who has passed through the gate G3 or the gate G4 but has not passed through the gate G7 or the gate G8 may be removed from the visitors of the first section detected as described above. By limiting the users who pass through the gate G3 to the visitors of the first section in this way, the authentication accuracy can be improved. Further, when limiting the users passing through the gate G3 to the visitors of the first section in this way, the collation threshold value may be set larger.

[Second Embodiment]
FIG. 12 is a block diagram of the face recognition device according to the second embodiment of the present invention. Referring to FIG. 12, the face recognition device 200 according to the present embodiment includes a face image acquisition unit 201, a collation unit 202, an operation state acquisition unit 203, and a threshold value change unit 204.

The face image acquisition unit 201 is configured to acquire a face image of a user passing through a first gate installed at a first boundary in a plurality of sections. The face image acquisition unit 201 can be configured in the same manner as, for example, the face image acquisition unit 132 of FIG. 4, but is not limited thereto.

The collation unit 202 is configured to perform face authentication based on a threshold value for the face image acquired by the face image acquisition unit 201. The collating unit 202 can be configured in the same manner as, for example, the collating unit 133 in FIG. 4, but is not limited thereto.

The operating state acquisition unit 203 is configured to acquire the operating state of the second gate installed at the second boundary different from the first boundary. The operating state acquisition unit 203 can be configured in the same manner as, for example, the operating state detecting unit 135 of FIG. 4, but is not limited thereto.

The threshold value changing unit 204 is configured to change the threshold value for matching used by the matching unit 202 based on the operating state of the second gate acquired by the operating state acquisition unit 203. The threshold value changing unit 204 can be configured in the same manner as, for example, the threshold value changing unit 204 of FIG. 4, but is not limited thereto.

The face recognition device 200 according to the present embodiment configured as described above operates as follows. That is, first, the face image acquisition unit 201 acquires the face image of the user passing through the first gate installed at the first boundary in the plurality of sections. Next, the collation unit 202 performs face authentication on the face image acquired by the face image acquisition unit 201 based on the threshold value. Next, the operating state acquisition unit 203 acquires the operating state of the second gate installed at the second boundary different from the first boundary. Then, the threshold value changing unit 204 changes the threshold value for matching used by the collating unit 202 based on the operating state of the second gate detected by the operating state acquisition unit 203.

As described above, according to the present embodiment, the matching threshold value used for face recognition performed at the first gate installed at the first boundary in the plurality of compartments is set to a second boundary different from the first boundary. Since it is automatically changed according to the operating state of the second gate installed in, a plurality of gates can be linked to construct an organized security gate.

Although the present invention has been described above with reference to each of the above embodiments, the present invention is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the structure and details of the present invention within the scope of the present invention.

For example, in the first embodiment, when a specific person (important person) is successfully authenticated at the gate G1 or the gate G2 and the specific person enters the first section ST from the outside, the thresholds of the gate G1 and the gate G2 are set. It may be increased, or the security level may be increased by changing from one-factor authentication by face authentication to two-factor authentication by face authentication and IC card authentication. On the other hand, since it is a problem that a specific person who has entered the first section cannot enter the second section by any chance, the threshold values of the gate G3 and the gate G4 may be lowered to improve convenience. Further, it is assumed that the third section is, for example, a server room, and the gate G5 is normally electrically closed in order to prevent the intrusion of unauthorized persons. At this time, when an authorized person enters the second section through the gate G3 or the gate G4, the closure of the gate G5 may be electrically released to accept the face recognition of the gate G5. In addition, various rules for controlling the thresholds of a plurality of gates can be considered.

In addition, the first embodiment applies the present invention to a case where it is desired to increase the security toward the inner section. However, on the contrary, the present invention can be applied to the case where the inner section is desired to be more convenient.

The present invention can be used for an entry / exit monitoring system, an attendance management system, etc., and is a case where the threshold value used for collation is appropriately and automatically corrected to improve convenience and safety without incurring a burden on the user. Suitable for.

Some or all of the above embodiments may also be described, but not limited to:
[Appendix 1]
A face image acquisition unit that acquires a face image of a user passing through a first gate installed at a first boundary in a plurality of sections, and a face image acquisition unit.
A collation unit that performs face recognition based on a threshold value for the face image,
An operating state acquisition unit that acquires the operating state of the second gate installed at a second boundary different from the first boundary, and
A threshold value changing unit that changes the threshold value based on the operating state,
Face recognition device including.
[Appendix 1-1]
The plurality of partitions have a plurality of security levels.
The face recognition device according to Appendix 1.
[Appendix 2]
The operating state acquisition unit acquires whether or not the second gate is open so that it can pass freely as one operating state of the second gate.
The face recognition device according to Appendix 1.
[Appendix 3]
The operating state acquisition unit acquires whether or not the second gate is closed so that it cannot pass through, as one operating state of the second gate.
The face recognition device according to Appendix 1 or 2.
[Appendix 4]
The operating state acquisition unit acquires whether or not the personal authentication performed at the second gate is multi-factor authentication as one operating state of the second gate.
The face recognition device according to any one of Appendix 1 to 3.
[Appendix 5]
The operating state acquisition unit acquires whether or not the personal authentication performed at the second gate is one-factor authentication as one operating state of the first gate.
The face recognition device according to any one of Appendix 1 to 4.
[Appendix 6]
The operating state acquisition unit acquires, as the operating state of the second gate, whether or not the person has been successfully authenticated to the specific person at the second gate.
The face recognition device according to any one of Appendix 1 to 5.
[Appendix 7]
The operating state acquisition unit acquires the number of successful authentications that occurred at the second gate in a certain period as the operating state of the second gate.
The face recognition device according to any one of Appendix 1 to 6.
[Appendix 8]
The first gate and the second gate are gates for entering and exiting different compartments of a facility partitioned into a plurality of compartments according to a plurality of security levels.
The face recognition device according to any one of Appendix 1 to 7.
[Appendix 9]
The first gate and the second gate are gates for entering and exiting the same section of a facility partitioned into a plurality of sections according to a plurality of security levels.
The face recognition device according to any one of Appendix 1 to 7.
[Appendix 10]
The threshold value changing unit changes the threshold value based on the operating state and a preset matching threshold value changing rule.
The face recognition device according to any one of Appendix 1 to 9.
[Appendix 11]
The plurality of compartments include a first compartment and a second compartment that can only be entered from the first compartment.
The first gate is installed at the entrance from the first compartment to the second compartment.
The face recognition device according to any one of Appendix 1 to 10.
[Appendix 12]
The plurality of compartments include a first compartment and a second compartment that can only be entered from the first compartment.
The first gate is installed at the entrance from the first section to the second section, and
The collation unit detects visitors in the first section based on the result of personal authentication of the gate installed at the entrance to the first section among the plurality of gates, and a reference face image for each registered user. The face recognition is performed by selecting the reference face image of the visitor from the above, calculating the similarity between the face image and the selected reference face image, and comparing it with the threshold value.
The face recognition device according to any one of Appendix 1 to 10.
[Appendix 13]
Acquire the face image of the user passing through the first gate installed at the first boundary in the plurality of sections, and obtain the face image of the user.
Face recognition is performed on the face image based on the threshold value.
Acquire the operating state of the second gate installed at the second boundary different from the first boundary.
The threshold is changed based on the operating state.
Face recognition method.
[Appendix 13]
The plurality of partitions have a plurality of security levels.
The face recognition method according to Appendix 12.
[Appendix 14]
In the acquisition of the operating state, whether or not the second gate is open so that it can pass freely is acquired as one operating state of the second gate.
The face recognition method according to Appendix 13.
[Appendix 15]
In the acquisition of the operating state, whether or not the second gate is closed so as not to pass is acquired as one operating state of the second gate.
The face recognition method according to Appendix 13 or 14.
[Appendix 16]
In the acquisition of the operating state, whether or not the personal authentication performed at the second gate is multi-factor authentication is acquired as one operating state of the second gate.
The face recognition method according to any one of Appendix 13 to 15.
[Appendix 17]
In the acquisition of the operating state, whether or not the personal authentication performed at the second gate is one-factor authentication is acquired as one operating state of the first gate.
The face recognition method according to any one of Appendix 13 to 16.
[Appendix 18]
In the acquisition of the operating state, the presence or absence of successful personal authentication for the specific person at the second gate is acquired as the operating state of the second gate.
The face recognition method according to any one of Appendix 13 to 17.
[Appendix 19]
In the acquisition of the operating state, the number of successful authentications that occurred at the second gate in a certain period is acquired as the operating state of the second gate.
The face recognition method according to any one of Appendix 13 to 18.
[Appendix 20]
The first gate and the second gate are gates for entering and exiting different compartment areas of a facility partitioned into a plurality of compartments according to a plurality of security levels.
The face recognition method according to any one of Appendix 13 to 19.
[Appendix 21]
The first gate and the second gate are gates for entering and exiting the same compartment area of a facility partitioned into a plurality of compartments according to a plurality of security levels.
The face recognition method according to any one of Appendix 13 to 19.
[Appendix 22]
In the change of the threshold value, the threshold value is changed based on the operation state and the matching threshold value change rule set in advance.
The face recognition method according to any one of Appendix 13 to 21.
[Appendix 23]
The plurality of compartments include a first compartment and a second compartment that can only be entered from the first compartment.
The first gate is installed at the entrance from the first compartment to the second compartment.
The face recognition method according to any one of Appendix 13 to 22.
[Appendix 24]
The plurality of compartments include a first compartment and a second compartment that can only be entered from the first compartment.
The first gate is installed at the entrance from the first section to the second section, and
The collation unit detects visitors in the first section based on the result of personal authentication of the gate installed at the entrance to the first section among the plurality of gates, and a reference face image for each registered user. The face recognition is performed by selecting the reference face image of the visitor from the above, calculating the similarity between the face image and the selected reference face image, and comparing it with the threshold value.
The face recognition method according to any one of Appendix 13 to 22.
[Appendix 25]
Computer,
A face image acquisition unit that acquires a face image of a user passing through a first gate installed at a first boundary in a plurality of sections, and a face image acquisition unit.
A collation unit that performs face recognition based on a threshold value for the face image,
An operating state acquisition unit that acquires the operating state of the second gate installed at a second boundary different from the first boundary, and
A threshold value changing unit that changes the threshold value based on the operating state,
A program that works.
[Appendix 26]
In the face recognition, the degree of similarity between the face image and the reference face image is calculated by collating the face feature amount extracted from the face image with the face feature amount extracted from the reference face image for each registered user. To do
The face recognition device, face recognition method, and program according to any one of Appendix 1 to 25.

ST ... Site BU ... Building IR ... Important rooms G1 to g10 ... Gate 100 ... Face recognition device 111 ... Interface unit 112 ... Interface unit 113 ... Communication interface unit 114 ... Operation input unit 115 ... Screen display unit 120 ... Storage unit 121 ... Program 122 ... Reference face image data 123 ... Collation threshold 124 ... Gate operation state data 125 ... Collation threshold change rule 130 ... Arithmetic processing unit 131 ... Input unit 132 ... Face image acquisition unit 133 ... Collation unit 134 ... Output unit 135 ... Operation state detection Unit 136 ... Threshold changing unit 141 ... User 142 ... Camera unit 200 ... Face authentication device 201 ... Face image acquisition unit 202 ... Collation unit 203 ... Operating state acquisition unit 204 ... Threshold changing unit

Claims (6)

A face image acquisition unit that acquires a face image of a user passing through a first gate installed at a first boundary that divides adjacent sections, and a face image acquisition unit.
A collation unit that performs face recognition based on a threshold value for the face image,
An operating state acquisition unit that acquires the operating state of the second gate installed at the first boundary, and
A threshold value changing unit that changes the threshold value based on the operating state,
Face recognition device including. The face recognition device according to claim 1, wherein the operating state acquisition unit acquires whether or not the second gate is closed so that it cannot pass through, as one operating state of the second gate. Acquire the face image of the user passing through the first gate installed at the first boundary that divides the adjacent sections.
Face recognition is performed on the face image based on the threshold value.
Acquire the operating state of the second gate installed at the first boundary,
The threshold is changed based on the operating state.
Face recognition method. The face according to claim 3, wherein in the acquisition of the operating state of the second gate, whether or not the second gate is closed, which is impassable, is acquired as one operating state of the second gate. Authentication method. On the computer
The process of acquiring the face image of the user passing through the first gate installed at the first boundary that divides the adjacent sections, and
The process of performing face recognition based on the threshold value for the face image and
The process of acquiring the operating state of the second gate installed at the first boundary and
The process of changing the threshold value based on the operating state, and
A program to make you do. The program according to claim 5, wherein in the acquisition of the operating state of the second gate, whether or not the second gate is closed, which is impassable, is acquired as one operating state of the second gate. ..

Images