Biometric authentication device, biometric authentication program and method

The present invention relates to a biometric authentication device, a biometric authentication program and a method.

Biometrics is known as a technique for identifying an individual based on human biological characteristics such as veins and facial patterns. In this biometric authentication, first, data indicating the ecological characteristics of the person to be authenticated is registered as a registration template, and authentication data indicating the characteristics of the individual's biometric pattern acquired by a sensor such as a camera at the time of authentication and registration are performed. Identify individuals by comparing with templates. Specifically, the similarity between the authentication data and the registration template is calculated, and when the similarity exceeds a predetermined threshold value, it is determined that the person to be authenticated is the person himself / herself. In such biometric authentication, authentication errors occur at a certain rate. As this authentication error, there is a person refusal error that determines that the person is not the person because the similarity with the registration template is low even though the feature data belongs to the person.

There are two main causes for the above-mentioned authentication error. One is the deviation of the posture of the authentication part with respect to the sensor used for authentication. Since the subject in the authentication is a person, the authentication part is not always input to the sensor in a posture that matches the registration template. The other is external light for the biometric authentication device. Biometric authentication using images may be affected by sunlight from the outside or incandescent lamps indoors. Most authentication devices irradiate the subject with a light source provided for the subject to take a picture, but when the unintended amount of sunlight or incandescent light has a large effect on the image, these external lights cause an authentication error.

<p><patcit num="1"><text>JP-A-2007-257040</text></patcit><patcit num="2"><text>Japanese Unexamined Patent Publication No. 2009-009434</text></patcit><patcit num="2"><text>Japanese Patent Application Laid-Open No. 2005-071009</text></patcit><patcit num="2"><text>Japanese Patent Application Laid-Open No. 02-158841</text></patcit></p>

<p> However, since it is difficult to save the authentication data including the face and veins of the certifier as a log due to security and privacy issues, it is difficult for the administrator of the biometric authentication device to investigate the cause of the authentication error. Is.</p><p> The present invention has been made to solve the above-mentioned problems, and provides a biometric authentication device, a biometric authentication program, and a method capable of investigating the cause of an authentication error without saving the authentication data as a log. The purpose is.</p>

<p> In order to solve the above-mentioned problem, the biometric authentication device includes a collation unit that authenticates the user by collating the biometric information read by the user with the registered biometric information registered in the storage unit in advance. When the user is not authenticated by the collating unit, a high-precision collating unit that collates the biometric information with the registered biometric information with higher accuracy than the collating by the collating unit, and a collating result by the high-precision collating unit. It has a collation result storage unit that records the above in the storage unit.</p>

<p> The cause of the authentication error can be investigated without saving the authentication data as a log.</p>

<figref num="1">It is a figure which shows the hardware configuration of the biometric authentication apparatus which concerns on Embodiment 1. FIG.</figref><figref num="2">It is a figure which shows the functional structure of the biometric authentication apparatus which concerns on Embodiment 1. FIG.</figref><figref num="3">It is a figure which shows the registration template table.</figref><figref num="4">It is a figure which shows the wide area collation log table.</figref><figref num="5">It is a figure which shows the search position table.</figref><figref num="6">It is a figure which shows the operation of the authentication process in Embodiment 1.</figref><figref num="7">It is a figure which shows the information which is added to the error authentication data.</figref><figref num="8">It is a figure which shows the operation of a wide range collation processing.</figref><figref num="9">It is a figure which shows the search range.</figref><figref num="10">It is a figure which shows the priority of the search range.</figref><figref num="11">It is a figure which shows the operation of the message display processing.</figref><figref num="12">It is a figure which shows the message displayed on the display.</figref><figref num="13">It is a figure which shows the functional structure of the biometric authentication apparatus which concerns on Embodiment 2.</figref><figref num="14">It is a figure which shows the partial collation log table.</figref><figref num="15">It is a figure which shows the error table.</figref><figref num="16">It is a figure which shows the error cause table.</figref><figref num="17">It is a figure which shows the illumination intensity table.</figref><figref num="18">It is a figure which shows the outside light error rate table.</figref><figref num="19">It is a figure which shows the operation of the authentication process in Embodiment 2.</figref><figref num="20">It is a flowchart which shows the operation of high precision collation processing.</figref><figref num="21">It is a figure which shows the block divided in the partial collation.</figref><figref num="22">It is a figure which shows the effect of partial collation.</figref><figref num="23">It is a figure which shows the formula which obtains the outside light direction.</figref><figref num="24">It is a figure which shows typically how to obtain the outside light direction.</figref><figref num="25">It is a figure which shows the angle in the outside light direction.</figref><figref num="26">It is a figure which shows the operation of an analysis process.</figref><figref num="27">It is a figure which shows the formula which calculates the error rate by outside light.</figref><figref num="28">It is a figure which shows the formula which calculates the average vector in the outside light direction.</figref><figref num="29">It is a figure which shows the formula which calculates the external light error occurrence user rate.</figref><figref num="30">It is a figure which shows the history table.</figref><figref num="31">It is a figure which shows the formula which calculates the error rate by a posture change.</figref><figref num="32">It is a figure which shows the fluctuation of an error rate by outside light.</figref><figref num="33">It is a figure which shows the formula which calculates the cause unknown error rate.</figref><figref num="34">It is a figure which shows an example of a computer system.</figref>

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(Embodiment 1) First, the hardware configuration of the biometric authentication device according to the first embodiment will be described. FIG. 1 is a diagram showing a hardware configuration of the biometric authentication device according to the first embodiment.

The biometric authentication device 1 according to the first embodiment is connected to the door control device 2 to perform palm vein authentication. The palm vein authentication performs identity verification by collating the registration template, which is a pre-registered palm vein pattern, with the authentication data based on the captured vein image. As shown in FIG. 1, the biometric authentication device 1 is a CPU (Central Processing). It is equipped with unit) 10, memory 11, non-volatile memory 12, lighting unit 13, camera 14, input unit 15, display 16, and external IF (Interface) 17 as hardware. The CPU 10 controls the biometric authentication device 1. The memory 11 is a main storage device directly accessed by the CPU 10. The non-volatile memory 12 stores a registration template table, a wide range collation log table, and a search position table, which will be described later. Examples of the non-volatile memory 12 include a hard disk and a flash memory. The illumination unit 13 irradiates the palm, which is the image pickup target of the camera 14, with near infrared rays. Examples of the lighting unit 13 include LEDs. The camera 14 captures a vein image of the palm by receiving near-infrared rays that are illuminated by the illumination unit 13 and reflected from the palm. Camera 14 includes CMOS (Complementary Metal Oxide Semiconductor) image sensor and CCD (Charge Coupled). Device) Image sensor can be mentioned. The input unit 15 is a device for the user of the biometric authentication device 1 to input an ID. Examples of the input unit 15 include a numeric keypad for inputting an ID, a keyboard, a touch panel, a non-contact IC card reader on which the ID is recorded, and the like. The display 16 presents information to the user of the biometric authentication device 1. The external IF17 mediates the transmission and reception of information between the biometric authentication device 1 and the door control unit 2.

Next, the functional configuration of the biometric authentication device according to the first embodiment will be described. FIG. 2 is a diagram showing a functional configuration of the biometric authentication device according to the first embodiment. Further, FIG. 3 is a diagram showing a registration template table. Further, FIG. 4 is a diagram showing a wide-range collation log table. Further, FIG. 5 is a diagram showing a search position table.

As shown in FIG. 2, the biometric authentication device 1 includes an extraction unit 101, a collation unit 102 (collation unit, presentation unit), a notification unit 103, a wide range collation unit 104 (high-precision collation unit), and a log management unit 105 (collation result). It has a storage unit) as a function. The extraction unit 101 extracts the vein pattern, which is a biological feature used for biometric authentication, as authentication data from the vein image captured by the camera 14. The collation unit 102 refers to the registration template table and verifies the identity by collating the registration template with the authentication data. As shown in FIG. 3, the registration template table stores an ID indicating an individual user and a registration template corresponding to the ID in association with each other. Specifically, the collation unit 102 performs a feature comparison process between the ID input by the input unit 15 and the corresponding registration template and the authentication data in the registration template table, and indicates how much the two match. The similarity is calculated, and if the similarity is equal to or higher than the threshold value, it is determined that the user to be authenticated is the person himself / herself. Further, at the time of collation, the collation unit 102 calculates the maximum similarity while giving the authentication data processing such as translation, rotation, reduction / enlargement within the search range specified in advance, and the similarity is maximum. The search position is associated with the ID and recorded in the search position table shown in FIG. Here, the search range indicates the movement range of the authentication data, and the search position indicates the movement position of the authentication data. The notification unit 103 notifies the door control device 2 of the door opening instruction only when the verification unit 102 succeeds in confirming the identity. The wide-range collation unit 104 performs a wide-range collation, which will be described later, on the authentication data for which the verification unit 102 has failed to verify the identity. The log management unit 105 records the collation result by the wide area collation unit 104 in the wide area collation log table. As shown in FIG. 4, the wide-range collation log table records the log number (No), ID, authentication execution time, collation result, and posture change in association with each other. Here, the posture variation indicates the variation position with respect to the reference position of the authentication data whose similarity in the wide range collation is equal to or more than the threshold value by the coordinates.

Next, the operation of the authentication process in the first embodiment will be described. FIG. 6 is a diagram showing the operation of the authentication process in the first embodiment. Further, FIG. 7 is a diagram showing information added to the error authentication data.

As shown in FIG. 6, first, the extraction unit 101 determines whether or not an ID has been input to the input unit 15 (S101).

When the ID is input to the input unit 15 (S101, YES), the collation unit 102 executes the message display process described later (S102). Next, the extraction unit 101 causes the camera 14 to capture a vein image, and extracts authentication data from the captured image (S103). Next, the collation unit 102 collates the registration template corresponding to the ID input in the registration template table with the authentication data (S104), and determines whether the similarity is equal to or higher than the threshold value (S105). If there is a search position corresponding to the ID in the search position table, the search range of the authentication data in the collation shall be based on the corresponding search position. If the search position corresponding to the ID does not exist in the search position table, the search range of the search data in the collation is set to a preset range.

In step S105, when the similarity is equal to or greater than the threshold value (S105, YES), the collation unit 102 records the search position giving the maximum similarity in the search position table (S106). Further, the notification unit 103 notifies the door control device 2 of the door opening instruction (S107). After the notification, the extraction unit 101 again determines whether or not the ID has been input to the input unit 15 (S101).

On the other hand, when the similarity is less than the threshold value (S105, NO), the collation unit 102 records the authentication data as error authentication data in the memory 11. Here, the collation unit adds the information shown in FIG. 7 to the authentication data. The error authentication data includes authentication data, ID, authentication execution time, expiration date, and authentication processing information. Here, the expiration date indicates the expiration date for recording the error authentication data in the memory 11. Further, the authentication processing information includes a search range indicating the range searched in the collation, the maximum similarity, and the coordinate position of the authentication data having the maximum similarity. After recording the error authentication data, the extraction unit 101 again determines whether or not the ID has been input to the input unit 15 (S101).

Next, the operation of the wide range collation processing will be described. FIG. 8 is a diagram showing the operation of the wide range collation process. Further, FIG. 9 is a diagram showing a search range. Further, FIG. 10 is a diagram showing the priority of the search range.

As shown in FIG. 8, first, the wide area collation unit 104 determines whether or not the authentication process is being executed (S201).

When the authentication process is not being executed (S201, NO), the wide area collation unit 104 determines whether or not there is error authentication data in the memory 11 (S202).

When there is error authentication data in the memory 11 (S202, YES), the wide range collation unit 104 executes the collation with a wider search range than the collation during the authentication process (S203), as shown in FIG. At this time, the wide range collation unit 104 refers to the authentication processing information added to the authentication data, and excludes the area already collated in the authentication process from the search range in the wide range collation. Further, the wide range collation unit 104 divides the search range into a plurality of blocks as shown in FIG. 10, calculates the distance from the point P where the similarity in the authentication process is maximum to the center of each block, and this distance is calculated. Priority is given to expanding the search range from the closest block. In FIG. 10, the search range is expanded in the order of f, h, and i. In this way, by expanding the search range from the vicinity of the point where the similarity is maximum, it is possible to efficiently perform wide-range collation.

Next, the log management unit 105 records the result of wide-range collation with the expanded search range in the wide-range collation log table together with the posture change (S204). Here, when the similarity in the wide range collation is equal to or greater than the threshold value, the log management unit 105 records or updates the search position corresponding to the ID in the search position table to the coordinate position where the similarity is maximum.

Next, the log management unit 105 deletes the error authentication data that has been subjected to wide-range verification from the memory 11 (S205), and determines whether or not the expired error authentication data is in the memory 11 (S206).

When the expired error authentication data is in the memory 11 (S206, YES), the log management unit 105 deletes the expired error authentication data from the memory 11 (S207). Next, the wide area collation unit 104 again determines whether or not the authentication process is being executed (S201).

On the other hand, when there is no expired error authentication data in the memory 11 (S207, NO), the wide range collation unit 104 again determines whether the authentication process is being executed (S201).

Further, in step S202, when there is no error authentication data in the memory 11 (S202, NO), the wide range collation unit 104 again determines whether or not the authentication process is being executed (S201).

Further, in step S201, when the authentication process is being executed (S201, YES), the wide range collation unit 104 again determines whether or not the authentication process is being executed (S201).

As described above, the biometric authentication device 1 according to the first embodiment performs a wide range collation with respect to the error authentication data, and records the result and the posture change as a log. As a result, the administrator of the biometric authentication device 1 can investigate the cause of the authentication error without leaving the authentication data having a problem in terms of security and privacy beyond the expiration date.

Next, the operation of the message display process will be described. FIG. 11 is a diagram showing the operation of the message display process. Further, FIG. 12 is a diagram showing a message displayed on the display.

As shown in FIG. 11, first, the collation unit 102 determines whether or not there is a wide-range collation log table corresponding to the input ID (S301).

When there is a wide-range collation log table corresponding to the input ID (S301, YES), the collation unit 102 determines whether the average of the posture fluctuations in the wide-range collation log table corresponding to the ID is equal to or more than a predetermined value (S302). ..

When the posture variation is equal to or greater than a predetermined value (S302, YES), the collating unit 102 determines whether or not the standard deviation of the attitude variation corresponding to the ID is less than the predetermined value (S303). Based on this judgment, it is judged whether or not there is a certain tendency in the posture change of the user indicated by the ID.

When the standard deviation of the posture change is less than the predetermined value, as shown in FIG. 12, the collating unit 102 sends a message to the display 16 indicating the movement of the hand in the direction opposite to the posture change direction in which the average is equal to or more than the predetermined value. Display (S304). The reminder for the movement of the user's hand may be presented by any means. For example, the user may be moved by voice. Further, when the magnitude of the posture change exceeds a predetermined range, the user may be asked to re-register the registration template. In addition, the re-registration of the registration template of the corresponding ID may be sent to the administrator of the biometric authentication device 1 by e-mail or the like via the connected network.

In this way, by presenting the message based on the posture change in the wide-range collation log table to the user, it is possible to reduce the authentication error caused by the displacement of the hand position at the time of authentication.

(Embodiment 2) In the first embodiment described above, the biometric authentication device performs palm vein authentication, but in the second embodiment, the biometric authentication device performs face authentication as biometric authentication. Hereinafter, the biometric authentication device according to the second embodiment will be described.

First, the hardware configuration of the biometric authentication device 1 according to the second embodiment will be described as being different from the first embodiment. The biometric authentication device 1 according to the second embodiment is connected to the door control device 2 to perform face authentication. Face recognition performs identity verification by collating a pre-registered pattern (registration template) showing facial features with authentication data based on an captured face image. In addition to the registration template table, wide range collation log table, and search position table, the non-volatile memory 12 includes a partial collation log table, a lighting intensity adjustment table, an external light error table for each user, an error factor table, and an external light error, which will be described later. Remember the rate table. The illumination unit 13 irradiates the face of the user to be imaged by the camera 14 with white light. An LED can be mentioned as the lighting unit 13. The camera 14 captures a face irradiated with white light by the illumination unit 13.

Next, the functional configuration of the biometric authentication device according to the second embodiment will be described. FIG. 13 is a diagram showing a functional configuration of the biometric authentication device according to the second embodiment. Further, FIG. 14 is a diagram showing a partial collation log table. Further, FIG. 15 is a diagram showing an error table. Further, FIG. 16 is a diagram showing an error factor table. Further, FIG. 17 is a diagram showing a lighting intensity table. Further, FIG. 18 is a diagram showing an external light error rate table.

As shown in FIG. 13, the biometric authentication device 1 according to the second embodiment has an extraction unit 101a, a collation unit 102a (collation unit), a notification unit 103a, a wide range collation unit 104a (high-precision collation unit), and a log management unit 105a. It has a (collation result storage unit), a partial collation unit 106 (high-precision collation unit), and an analysis unit 107 as functions. The extraction unit 101a causes the camera 14 to capture a face image and extracts authentication data from the face image. Further, the extraction unit 101a adjusts the illumination intensity by the illumination unit 13 based on the illumination intensity adjustment table at the time of imaging. The collation unit 102a refers to the registration template table, verifies the identity by collating the registration template with the authentication data, and updates the error table based on the result. As shown in FIG. 15, the error table records the combination of the ID and the time zone, the number of collations, and the number of errors in association with each other. The notification unit 103a notifies the door control device 2 of the door opening instruction only when the collation unit 102a succeeds in confirming the identity. The wide-range collation unit 104a performs wide-range collation on the authentication data for which the identity verification by the collation unit 102a has failed. The partial collation unit 106 performs partial collation, which will be described later, with respect to the authentication data for which the identity verification by the collation unit 102a has failed. The log management unit 105a records the collation result by the wide area collation unit 104a in the wide area collation log table, and records the collation result by the partial collation unit 106 in the partial collation log table. As shown in FIG. 14, the partial collation log table records the log number (No), ID, authentication execution time, collation result, collation block, and external light direction in association with each other. The collation block and the external light direction will be described later. The analysis unit 107 performs analysis based on at least one of an error table, a wide range collation log table, and a partial collation log table, and updates the illumination intensity adjustment table, the error cause table, and the external light error rate table. Further, as shown in FIG. 16, the error factor table records the combination of the ID and the time zone and the number of times of each of the causes of the authentication error in association with each other. In addition, the lighting intensity adjustment table As shown in FIG. 17, the time zone and the illumination intensity of the illumination unit 13 are recorded in association with each other. The illumination intensity is indicated by, for example, the duty ratio of PWM (Pulse Width Modulation). Further, as shown in FIG. 18, the external light error rate table records the time zone, the number of external light errors, the total number of collations, the error rate due to external light, and the external light direction in association with each other. The error rate due to external light indicates the ratio of the number of errors caused by external light to the total number of collations of all users who have authenticated in the corresponding time zone. Further, the external light direction indicates an average value of all the external light directions calculated by partial collation with respect to the error authentication data in the corresponding time zone.

Next, the authentication process according to the second embodiment will be described. FIG. 19 is a diagram showing the operation of the authentication process according to the second embodiment.

As shown in FIG. 19, first, the extraction unit 101a determines whether or not an ID has been input to the input unit 15 (S401).

When the ID is input to the input unit 15 (S401, YES), the extraction unit 101a adjusts the illumination unit 13 to the illumination intensity corresponding to the current time zone in the illumination intensity adjustment table (S402). Next, the extraction unit 101a causes the camera 14 to capture a face image and extracts authentication data from the captured image (S403). Next, the collation unit 102a collates the registration template corresponding to the ID entered in the registration template table with the authentication data (S404), and determines whether the similarity is equal to or higher than the threshold value (S405). Here, the collation unit 102a refers to the search position table as in the collation unit 102 in the first embodiment.

When the similarity is less than the threshold value (S405, NO), the collating unit 102a records the authentication data as error authentication data in the memory 11 (S406) and updates the error table (S407). In this case, the collation unit 102a increments the total number of collations and the number of errors corresponding to the IDs. After updating the error table, the extraction unit 101a again determines whether or not the ID has been input to the input unit 15 (S401).

On the other hand, when the similarity is equal to or higher than the threshold value (S405, YES), the collation unit 102a records the search position giving the maximum similarity in the search position table (S408). Further, the notification unit 103a notifies the door control device 2 of the door opening instruction (S409). Next, the collation unit 102a updates the error table (S407). In this case, the collation unit 102a increments only the total number of collations corresponding to the IDs. After updating the error table, the extraction unit 101a again determines whether or not the ID has been input to the input unit 15 (S401).

Next, the operation of the high-precision collation processing will be described. This high-precision collation process is a process of performing partial collation on the error authentication data in addition to the wide-range collation described in the first embodiment. FIG. 20 is a flowchart showing the operation of the high-precision collation process. Further, FIG. 21 is a diagram showing blocks divided in partial collation. Further, FIG. 22 is a diagram showing the effect of partial collation. Further, FIG. 23 is a diagram showing an equation for obtaining the external light direction. Further, FIG. 24 is a diagram schematically showing how to obtain the external light direction. Further, FIG. 25 is a diagram showing an angle in the external light direction.

As shown in FIG. 20, first, the wide area collation unit 104a determines whether or not the authentication process is being executed (S501).

When the authentication process is not being executed (S501, NO), the wide area collation unit 104a determines whether or not there is error authentication data in the memory 11 (S502).

When there is error authentication data in the memory 11 (S502, YES), the wide range collation unit 104a executes the collation with a wider search range than the collation at the time of the authentication process (S503). After executing the wide-range collation, the log management unit 105a records the result of the wide-range collation with the expanded search range in the wide-range collation log table together with the posture change (S504).

Next, the partial collation unit 106 executes partial collation on the error authentication data (S505). Here, partial collation will be described. In the partial collation, as shown in FIG. 21, the authentication data is divided into a plurality of blocks, and each block is collated with the registration template. In the second embodiment, the authentication data is divided into four blocks a, b, c, and d. By this partial collation, as shown in FIG. 22, even if the authentication data is such that an image suitable for authentication of a part of the area is missing due to external light or the like, the verification is performed by the block in which the missing occurs. Can be done.

In this partial collation, the partial collation unit 106 calculates the external light direction indicating the incident direction of the external light in the captured image. This outside light direction is the direction with respect to the center of the error authentication data at the location having the lowest similarity in the error authentication data. The direction of external light to be obtained is vector A, the number of divided blocks is i, and the vector from the center of the error authentication data to block i is vector a._i, W the similarity of each block_iThen, the vector A indicating the direction of external light is obtained by the formula shown in FIG. According to this equation, as shown in FIG. 24, the external light direction is obtained based on the weight indicated by the thickness of the arrow from the center of the error authentication data to the center of each block. Here, the weight is a value represented by the reciprocal of the similarity of each block. In FIG. 24, the lightness and darkness of each block indicates the degree of similarity, the darker the degree, the lower the degree of similarity, and the brighter the degree, the higher the degree of similarity. As shown in FIG. 25, the external light direction calculated in this way is represented by a clockwise angle with respect to the center of the error authentication data in each table. Further, the partial collation unit 106 calculates the average brightness of the block having the lowest similarity in the partial collation, and records the value in the non-volatile memory 12 in association with the ID and the time zone in which the collation was performed.

After executing the partial collation, the log management unit 105a records the result of the wide-range collation with the expanded search range in the wide-range collation log table together with the posture change (S504). Here, when there is a block whose similarity is equal to or higher than the threshold value in the partial collation, the log management unit 105a searches for the block whose similarity is equal to or higher than the threshold value and the external light direction calculated by the above method in association with the ID. Record in the position table.

Next, the log management unit 105a deletes the error authentication data that has been subjected to wide-range collation and partial collation from the memory 11 (S507), and determines whether or not the expired error authentication data is in the memory 11 (. S508)

When the expired error authentication data is in the memory 11 (S508, YES), the log management unit 105a deletes the expired error authentication data from the memory 11 (S509). Next, the wide range collation unit 104a again determines whether or not the authentication process is being executed (S501).

On the other hand, when there is no expired error authentication data in the memory 11 (S508, NO), the wide range collation unit 104a determines again whether the authentication process is being executed (S501).

Further, in step S502, when there is no error authentication data in the memory 11 (S502, NO), the wide range collation unit 104a again determines whether or not the authentication process is being executed (S501).

Further, in step S501, when the authentication process is being executed (S501, YES), the wide range collation unit 104a again determines whether or not the authentication process is being executed (S501).

As described above, the biometric authentication device 1 according to the second embodiment performs partial collation with respect to the error authentication data, and records the result and the external light direction as a log. As a result, the administrator of the biometric authentication device 1 can investigate the cause of the authentication error without leaving the authentication data having a problem in terms of security and privacy beyond the expiration date. In addition, by performing wide-range collation and partial collation on the same error authentication data, the administrator can identify the cause of the authentication error from any of attitude change, outside light, attitude change, and outside light. Can be done.

Next, the operation of the analysis process will be described. This analysis process is a process of performing analysis based on the results of wide-area collation and partial collation. FIG. 26 is a diagram showing the operation of the analysis process. FIG. 27 is a diagram showing an equation for calculating the error rate due to external light. FIG. 28 is a diagram showing an equation for calculating the average vector in the external light direction.

As shown in FIG. 26, first, the analysis unit 107 determines whether or not a preset predetermined time has elapsed (S601).

When the predetermined time has elapsed (S601, YES), the analysis unit 107 determines whether or not the authentication process is being executed (S602).

When the authentication process is not being executed (S602, NO), the analysis unit 107 determines whether or not the high-precision collation process is being executed (S603).

When the high-precision collation process is not being executed (S603, NO), the analysis unit 107 updates the error cause table based on the error table, the wide range collation table, and the partial collation log table (S604).

Next, the analysis unit 107 calculates the error rate due to external light for each time zone based on the error factor table (S605). The error rate due to external light is t for the time zone and e for the error rate due to external light in the time zone t._ErrL(t), the total number of collations in the time zone t is C (t), and the number of errors due to external light in the time zone t is C._ErrLIn the case of (t), it is calculated by the formula shown in FIG. 27. Here, t is a variable indicating 00:00 to 23:00 by the values 0 to 23.

Next, the analysis unit 107 calculates the average vector in the external light direction based on the partial collation log table (S606). The average vector is t for the time zone, vector &lt;A&gt; for the average vector in the time zone t, i for the variable indicating each external light error in the time zone t, and vector A for the vector in the external light error i._i, C the number of errors due to external light in time zone t_ErrLIn the case of (t), it is calculated by the formula shown in FIG. 28. After calculating the average vector, the analysis unit 107 updates the external light error rate table based on these calculated values (S607).

Next, the analysis unit 107 illuminates the error factor table based on the average brightness value associated with the user ID whose number of external light errors is equal to or greater than a predetermined number and which is not recorded in the illumination intensity table. Calculate the strength (S608). This average luminance value is a value calculated by the partial collation unit 106 in the above-mentioned partial collation. Here, the analysis unit 107 calculates the illumination intensity based on the correspondence between the average brightness value and the appropriate illumination intensity. In this correspondence, the face is imaged so that the predetermined block i contains external light, and the illumination intensity by the illumination unit 13 is set so that the average brightness value of the block i of the authentication data based on this face image becomes an appropriate value. It shall be required in advance by adjusting. After calculating the illumination intensity, the analysis unit 107 updates the illumination intensity table for the time zone associated with the average brightness value based on the calculated illumination intensity (S609). After the update, the analysis unit 107 again determines whether or not the predetermined time has elapsed (S601).

Further, in step S603, when the high-precision collation is being executed (S603, YES), the analysis unit 107 again determines whether or not the authentication process is being executed (S602).

Further, in step S602, when the authentication process is being executed (S602, YES), the analysis unit 107 again determines whether or not the predetermined time has elapsed (S601).

Further, in step S601, when the predetermined time has not elapsed (S601, NO), the analysis unit 107 determines again whether or not the predetermined time has elapsed (S601).

In this way, the biometric authentication device 1 can reduce the burden on the administrator of the biometric authentication device 1 by performing the analysis based on the error table, the wide range collation log table, and the partial collation log table. For example, when the preset error rate is exceeded, the biometric authentication device 1 notifies the administrator of an e-mail via the network, so that the administrator needs to periodically monitor the log data of the biometric authentication device 1. Absent.

In the above-mentioned analysis process, the external light error rate was calculated for each time zone, but the analysis unit 107 calculates the percentage of users who have an error due to external light in a certain time zone (external light error occurrence user rate). You may calculate. FIG. 29 is a diagram showing an equation for calculating the external light error occurrence user rate. The number of users who have an external light error is the time zone t, and the percentage of users who have an error due to external light in the time zone t is e'._ErrL(t), N (t) is the number of users collated in the time zone t, and N is the number of users in which an error occurred due to external light in the time zone t._ErrLIn the case of (t), it is calculated by the formula shown in FIG. 29. The external light error occurrence user rate is particularly effective when a specific user causes a large number of errors and pushes up the overall external light error rate. For example, when the external light error occurrence user rate corresponding to the time zone when the external light error rate is high is high, the administrator can infer that the external light error has occurred not only for a specific user. That is, by referring to the external light error occurrence user rate, the administrator can determine how reliable the external light error rate is.

Further, the analysis unit 107 may create a history table showing the history of the external light error rate focusing on a certain time zone t. FIG. 30 is a diagram showing a history table. As shown in FIG. 30, the history table records the date, the number of external light errors on a daily basis, the total number of collations on a daily basis, the error rate due to external light on a daily basis, and the average in the external light direction on a daily basis. .. In addition, a history table is created for each time zone. The analysis unit 107 updates the history table of the time zone in which the collation was performed for each collation. Further, when displaying the external light error rate table and the history table, the history table of each time zone shall be linked to the corresponding time zone in the external light error rate table. By creating such a history table, the administrator can know the change in the external light error rate, and it can be inferred that the environment of the biometric authentication device 1 has changed.

Further, the analysis unit 107 may calculate the error rate due to the posture change for each user. FIG. 31 is a diagram showing an equation for calculating the error rate due to posture fluctuation. Further, FIG. 32 is a diagram showing fluctuations in the error rate due to external light. Time zone is t, error rate due to posture change of a specific user is e_A, C (t) is the total number of collations of a specific user in the time zone t, and C is the total number of errors due to the attitude of a specific user in the time zone t._ErrAIn the case of (t), the error rate due to attitude change is calculated by the formula shown in FIG. 31. By calculating the error rate due to posture change for each user in this way, it is possible to extract users with few errors due to posture change. Since the error due to the posture change is an error peculiar to the user, the influence of the error due to the posture change can be reduced by calculating the error rate due to the external light only for the user having a low error rate due to the posture change. The analysis unit 107 may calculate the error rate due to external light for the top 5% of users in ascending order of error rate due to posture change. By plotting the error rate due to external light obtained in this manner for each time zone, the administrator can obtain fluctuations in the external light error rate as shown in FIG. 32. In FIG. 32, the vertical axis shows the error rate due to external light, and the horizontal axis shows the time. This fluctuation in the external light error rate allows the administrator to more accurately estimate the time zone in which the external light affects the collation.

In addition, the analysis unit 107 may calculate an error rate (error rate of unknown cause) due to factors other than external light and attitude fluctuation. FIG. 33 is a diagram showing an equation for calculating the cause unknown error rate. Time zone t, total number of collations in time zone t is C (t), total number of errors in time zone t is C_Err(t), C the total number of errors due to attitude change in time zone t_ErrA(t), C the total number of errors due to external light in the time zone t_ErrLWhen (t) is set, the cause unknown error rate e in the time zone t^*_AL(t) is calculated by the formula shown in FIG. 33. When the unexplained error rate is 0 or more, the biometric authentication device 1 notifies the administrator by e-mail with a higher priority than the error caused by other factors. This allows the administrator to respond quickly to an error of unknown cause.

As mentioned above, by performing more accurate collations than normal collations, such as wide-area collations and partial collations, and recording the results, the administrator can determine the cause of the error without saving the error authentication data. Can be guessed. Moreover, the above-mentioned two embodiments may be carried out in combination.

The above-mentioned processing by the biometric authentication device 1 can be applied to the following computer system by connecting a device for biometric authentication. FIG. 34 is a diagram showing an example of a computer system. The computer system 900 shown in FIG. 34 includes a main body 901 having a built-in CPU, disk drive, etc., a display 902 that displays images according to instructions from the main body 901, and a keyboard 903 for inputting various information to the computer system 900. It has a mouse 904 that specifies an arbitrary position on the display screen 902a of the display 902, and a communication device 905 that accesses an external database or the like and downloads a program or the like stored in another computer system. The communication device 905 may be a network communication card, a modem, or the like.

A program for executing each of the above steps in a computer system as described above can be provided as a biometric authentication program. This program can be executed by the computer system constituting the biometric authentication device by storing it in a recording medium that can be read by the computer system. The program that executes each of the above steps is stored in a portable recording medium such as a disk 910, or downloaded from a recording medium 906 of another computer system by the communication device 905. Further, a biometric authentication program (biometric authentication software) that gives the computer system 900 at least a biometric authentication function is input to the computer system 900 and compiled. This program operates the computer system 900 as a biometric authentication device having a biometric authentication function. Further, this program may be stored in a computer-readable recording medium such as a disk 910. Here, as the recording medium that can be read by the computer system 900, an internal storage device such as ROM or RAM internally mounted on the computer, a portable storage device such as a disk 910, a flexible disk, a DVD disk, an optical magnetic disk, or an IC card. It includes a medium, a database that holds a computer program, or another computer system, and various recording media that can be accessed by the database or a computer system connected via a communication means such as a communication device 905.

The present invention can be practiced in various other forms without departing from its spirit or key features. Therefore, the above embodiments are merely exemplary in all respects and should not be construed in a limited way. The scope of the present invention is shown by the scope of claims, and is not bound by the text of the specification. Moreover, all modifications, various improvements, substitutions and modifications that fall within the equivalent scope of the claims are all within the scope of the present invention.

1 Biometric device, 2 door control device, 10 CPU, 11 memory, 12 non-volatile memory, 13 lighting unit, 14 camera, 15 input unit, 16 display, 17 external IF, 101,101a extraction unit, 102,102a collation unit, 103,103 a Notification unit, 104,104a Wide area collation unit, 105,105a Log management unit, 106 partial collation unit, 107 analysis unit.