Information processing device, biometric authentication device and biometric authentication system

Abstract

To provide an information processing device, a biometric authentication device and a biometric authentication system capable of reducing false rejection while maintaining the risk of false acceptance at a low level, in biometric authentication.SOLUTION: A biometric authentication system comprises a terminal and a server. The terminal and the server are constituted so as to transmit / receive information to / from each other via a network. The terminal, when FNIR corresponding to a FPIR setting value on a curve of a reference map is larger than a FNIR upper limit, selects a narrowing-down method from a plurality of narrowing-down methods for identifying a user candidate of an authentication object from a plurality of registered users, and executes the selected narrowing-down method to narrow down a registration template corresponding to the user candidate of the authentication object from a plurality of registration templates, and sets a threshold value according to the number of the candidates of the narrowed-down registration templates, and biometric authentication is executed by using the set threshold value and the narrowed-down registration template.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, a biometric authentication apparatus, and a biometric authentication system.

Background Art

[0002] As one of biometric authentication technologies, there is 1:N authentication. A biometric authentication system that employs 1:N authentication uses a plurality of registered templates that are biometric authentication information of a plurality of registered users, compares the biometric authentication information of a user to be authenticated with these registered templates, and identifies (specifies) the user to be authenticated from among the plurality of registered users.

[0003] The personal authentication apparatus of Patent Document 1 (hereinafter referred to as the "conventional apparatus") creates a face image for verification of the person to be authenticated, calculates a collation value between the created face image for verification and all the face images registered in the dictionary, and determines whether the maximum collation value of these collation values is equal to or greater than the authentication threshold value. When the maximum collation value is equal to or greater than the authentication threshold value, the conventional apparatus authenticates the person to be authenticated as the registered person corresponding to the face image for which the maximum collation value was calculated. When the maximum collation value is less than the authentication threshold value, the conventional apparatus determines the level of the maximum collation value. The conventional apparatus causes the user to input storage information (group number or user ID number, or both password and ID number) according to the level of the maximum collation value, and performs personal authentication based on the combination of the collation result of the storage information and the biometric collation result.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] The biometric authentication system compares pre-registered biometric authentication information (registered template) with biometric authentication information at the timing of authentication execution, and based on the matching value between the two and a specific threshold value, determines that authentication is successful when it is determined that the two are close enough to meet a predetermined determination criterion.

[0006] However, in a biometric authentication system, due to differences such as the difference in the biometric authentication information acquisition status at the time of registration of the user's biometric authentication information and the biometric authentication information acquisition status at the time of authentication, there may be a case where an incorrect authentication result occurs. At this time, the most problematic in practice is that an unregistered person is erroneously authenticated as a registered person (resulting in acceptance of others by mistake). If acceptance of others by mistake occurs, there is a possibility that the assets and rights of the registered person who has been erroneously authenticated will be used by others. To prevent such a situation, a biometric authentication system generally sets strict judgment criteria (relatively high threshold values) that are less likely to accept others by mistake. However, in settings that severely limit acceptance of others by mistake, the possibility of the authentication of the person himself / herself failing (the possibility of false rejection of the person himself / herself occurring) increases, so it is difficult to balance the two. For this reason, it has been difficult for a biometric authentication system to reduce false rejection of the person himself / herself while keeping the risk of accepting others low. Incidentally, although the conventional device of Patent Document 1 performs personal authentication using two authentication means, biometric authentication and input of memory information, there may be a case where authentication is successful (erroneously determined to be the person himself / herself) based on the collation result of the memory information input by others finally.

[0007] The present invention has been made to solve the above problems. That is, one of the objects of the present invention is to provide an information processing device, a biometric authentication device, and a biometric authentication system that can reduce the possibility of false rejection of the person himself / herself while keeping the risk of accepting others low in biometric authentication.

Means for Solving the Problems

[0008] In order to solve the above problems, an information processing apparatus according to the present invention includes an arithmetic unit that collates biometric authentication information of a user to be authenticated with registered biometric authentication information of a plurality of registered users, and performs biometric authentication to identify the user to be authenticated from among the plurality of registered users. The arithmetic unit acquires biometric authentication information of the user to be authenticated, the number of the registered biometric authentication information, and biometric authentication performance information indicating a relationship between the false acceptance identification rate and the false rejection identification rate of the biometric authentication corresponding to the number of the registered biometric authentication information. When the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information does not satisfy a predetermined criterion, the arithmetic unit collates authentication information related to the biometrics or knowledge of the user to be authenticated with registered authentication information of the plurality of registered users, selects one or a plurality of narrowing-down methods from among a plurality of narrowing-down methods for identifying candidates for the user to be authenticated from among the plurality of registered users, acquires the authentication information of the user to be authenticated according to the selected narrowing-down method, executes the selected narrowing-down method to narrow down the registered biometric authentication information corresponding to the candidates for the user to be authenticated from among the plurality of registered biometric authentication information, sets collation conditions according to the number of the narrowed-down registered biometric authentication information, and performs the biometric authentication using the set collation conditions and the narrowed-down registered biometric authentication information.

[0009] The biometric authentication device of the present invention acquires biometric authentication information of a user to be authenticated, collates the acquired biometric authentication information of the user to be authenticated with the registered biometric authentication information of a plurality of registered users, and executes biometric authentication for identifying the user to be authenticated from among the plurality of registered users. The arithmetic device includes: the arithmetic device acquires the biometric authentication information of the user to be authenticated, the number of the registered biometric authentication information, and biometric authentication performance information indicating the relationship between the false acceptance identification rate and the false rejection identification rate of the biometric authentication corresponding to the number of the registered biometric authentication information; when the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information does not satisfy a predetermined standard, the arithmetic device collates the authentication information related to the biometrics or knowledge of the user to be authenticated with the registered authentication information of the plurality of registered users, selects one or more of the plurality of narrowing-down methods for identifying candidates for the user to be authenticated from among the plurality of registered users, acquires the authentication information of the user to be authenticated according to the selected narrowing-down method, executes the selected narrowing-down method to narrow down the registered biometric authentication information corresponding to the candidates for the user to be authenticated from among the plurality of registered biometric authentication information, sets collation conditions according to the number of the narrowed-down registered biometric authentication information, and executes the biometric authentication using the set collation conditions and the narrowed-down registered biometric authentication information.

[0010] The biometric authentication system of the present invention includes a terminal that acquires biometric authentication information of a user to be authenticated, and an information processing device that exists outside or inside the terminal, acquires the biometric authentication information of the user to be authenticated from the terminal, and collates the acquired biometric authentication information of the user to be authenticated with the registered biometric authentication information of a plurality of registered users, and executes biometric authentication to identify the user to be authenticated from among the plurality of registered users. The information processing device acquires the biometric authentication information of the user to be authenticated, the number of the registered biometric authentication information, and biometric authentication performance information indicating the relationship between the false acceptance identification rate and the false rejection identification rate of the biometric authentication corresponding to the number of the registered biometric authentication information. When the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information does not satisfy a predetermined standard, the information processing device collates the authentication information related to the biometric or knowledge of the user to be authenticated with the registered authentication information of the plurality of registered users, selects one or a plurality of narrowing-down methods from among a plurality of narrowing-down methods for identifying candidates for the user to be authenticated from among the plurality of registered users, acquires the authentication information of the user to be authenticated according to the selected narrowing-down method, executes the selected narrowing-down method to narrow down the registered biometric authentication information corresponding to the candidates for the user to be authenticated from among the plurality of registered biometric authentication information, sets collation conditions according to the number of the narrowed-down registered biometric authentication information, and is configured to execute the biometric authentication using the set collation conditions and the narrowed-down registered biometric authentication information.

Advantages of the Invention

[0011] According to the present invention, in biometric authentication, it is possible to reduce the possibility of false rejection by the user himself while keeping the risk of false acceptance by others low. Note that the effects described here are not necessarily limited, and any of the effects described in the present disclosure may be applicable.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9A

Figure 9B

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

MODE FOR CARRYING OUT THE INVENTION

[0013] Hereinafter, each embodiment of the present invention will be described with reference to the drawings. In all the drawings of the embodiments, the same or corresponding parts may be denoted by the same reference numerals. In the following description, various types of information may be described using expressions such as "table", but the various types of information may be represented by data structures other than these. When describing identification information, expressions such as "identification number" and "name" are used, but these are mutually replaceable with each other, and other expressions may also be used. When describing a process with a functional block as the subject, the subject of the process may be a CPU or a device instead of the functional block.

[0014] <<First Embodiment>> A biometric authentication system according to the first embodiment of the present invention will be described. FIG. 1 is a diagram showing a configuration example of the biometric authentication system according to the first embodiment of the present invention. As shown in FIG. 1, the biometric authentication system according to the first embodiment includes a terminal 100 and a server 200. The terminal 100 and the server 200 are configured to be able to transmit and receive information to and from each other (communicable) via a network NW1. The terminal 100 may be referred to as a "biometric authentication device".

[0015] The terminal 100 includes a camera 110, a display 120, and an information processing device 130. These are configured to be able to communicate information with each other via a bus (not shown).

[0016] The camera 110 acquires a captured image of a specific part (for example, face, palm print, iris) of a person for use in authenticating the person present within the imaging range of the camera 110 in front of the terminal 100 by imaging the specific part of the person.

[0017] The display 120 is a display device capable of displaying an image. In this example, the display 120 is a touch panel type display that functions as a display device and also functions as an input device (operation device). Note that the terminal 100 may be provided with an input device (operation device) separate from the display device.

[0018] FIG. 2 is a diagram showing a hardware configuration example of a computer 2000 applied to an information processing apparatus 130 included in the terminal 100. The computer 2000 may be referred to as a "computer". The computer 2000 includes a CPU 2001, a ROM 2002, a RAM 2003, a non-volatile storage device 2004 capable of reading and writing data, a network interface 2005, an input / output interface 2006, and the like. These are communicably connected to each other via a bus 2007.

[0019] The CPU 2001 is an arithmetic unit that realizes various functions by loading various programs (not shown) stored in the ROM 2002 and / or the storage device 2004 into the RAM 2003 and executing the programs loaded into the RAM 2003. As described above, various programs executed by the CPU 2001 are loaded into the RAM 2003, and data used when the CPU 2001 executes various programs is temporarily stored. The ROM 2002 and / or the storage device 2004 are non-volatile storage media, and various programs are stored in the ROM 2002 and / or the storage device 2004.

[0020] The storage device 2004 stores the narrowing-down setting information 136, the narrowing-down method definition information 137, the reference Map (FPIR, FNIR, N) 138, and the post-narrowing threshold setting information 139. Hereinafter, the reference Map (FPIR, FNIR, N) 138 is referred to as the "reference map 138" for convenience. The details of these pieces of information will be described later.

[0021] The network interface 2005 is an interface for connecting the computer 2000 to the network NW1. The input / output interface 2006 is an interface for connecting the computer 2000 to an operating device and a display 120 (display device) capable of displaying images.

[0022] Note that instead of the computer 2000, a hardware device in which part or all of the computer 2000 is constituted by an FPGA (Field Programmable Gate Array) or the like may be used. Such a hardware device may also be referred to as an "arithmetic unit".

[0023] The information processing apparatus 130 includes, as a program, a user information reading unit 131, an authentication control unit 132, an authentication threshold setting unit 133, an authentication processing unit 134, and a result output unit 135.

[0024] The user information reading unit 131 uses biometric authentication technology to acquire the biometric authentication information of the user for authenticating the user from the camera 110 (the image captured by the camera 110). The terminal 100 in this example can perform iris authentication, palmprint authentication, and PIN (Personal Identification Number) authentication using authentication information (biometric authentication information or knowledge information) related to the biometric or knowledge of the user for authenticating the user as needed, as a narrowing-down method for narrowing down the registration template used for the main biometric authentication and as the main biometric authentication, which is face authentication. Note that the "registration template" may be referred to as "registered biometric authentication information".

[0025] The main biometric authentication identifies (specifies) the user to be authenticated from among a plurality of registered users based on the biometric authentication information of the user to be authenticated and the registration templates (authentication information 210) of a plurality of registered users registered in advance for the main biometric authentication. In the main biometric authentication, if the user to be authenticated can be identified (specified) from among a plurality of registered users, the authentication of the user to be authenticated is successful, and if the user to be authenticated cannot be identified (specified) from among a plurality of registered users, the authentication of the user to be authenticated fails.

[0026] The narrowing-down method identifies (specifies) candidates for the user who matches the user to be authenticated from among the registration authentication information of multiple registered users corresponding to the pre-registered narrowing-down method, by determining the registration authentication information that matches (or is highly likely to match) the biometric authentication information of the user. The narrowing-down method is an authentication performed to identify candidates for the user to be authenticated from among the registration authentication information of multiple registered users corresponding to the narrowing-down method, and is performed to narrow down the registration templates (the number of registration templates) used for the main biometric authentication. Note that in the narrowing-down method, unlike in the main biometric authentication, the success or failure of the authentication of the user to be authenticated is not determined. Also, in the narrowing-down method, not only biometric authentication but also authentication using the user's knowledge information (for example, PIN code, etc.) can be adopted. Note that as the narrowing-down method, from the viewpoint of further reducing the complexity of user utilization, it is preferable that the narrowing-down method does not require the user to input data again for the narrowing-down method, and biometric authentication information for the narrowing-down method can be obtained using the data for obtaining the biometric authentication information of the user for the main biometric authentication.

[0027] When performing face authentication, the user information reading unit 131 acquires the face image of the user captured by the camera 110. When performing iris authentication, the user information reading unit 131 acquires the eye image including the iris of the user captured by the camera 110. When performing palmprint authentication, the user information reading unit 131 acquires the hand image including the palmprint of the user captured by the camera 110. When performing PIN authentication, the user information reading unit 131 acquires the PIN code input via the touch panel (display 120).

[0028] The authentication control unit 132 acquires the authentication information of the user to be authenticated used for authentication from the user information reading unit 131, and performs authentication control to authenticate the user by 1:N authentication using the authentication threshold setting unit 133 and the authentication processing unit 134. The details of this authentication control will be described later.

[0029] The authentication threshold setting unit 133 sets a threshold value (i.e., the matching condition), which is a reference value for determining the degree of matching between the user's biometric authentication information and the registered template (registered template). The degree of matching is a numerical value for evaluating the similarity between the input user's biometric authentication information and the registered template stored in the authentication information 210. The higher the degree of matching, the higher the degree of coincidence between the input biometric authentication information and the registered template.

[0030] The authentication processing unit 134 performs user authentication by 1:N authentication. That is, the authentication processing unit 134 compares the biometric authentication information of the user to be authenticated with the registered templates of a plurality of registered users, and identifies the user to be authenticated from among the plurality of registered users corresponding to the plurality of registered templates.

[0031] For example, the authentication processing unit 134 calculates the degree of matching between the input biometric authentication information and each of the plurality of registered templates, compares the degree of matching with the threshold value, which is the reference value of the degree of matching, and based on the comparison result, identifies the user to be authenticated from among the plurality of registered users.

[0032] If the number of candidates for the authentication result with a degree of matching equal to or higher than the threshold value is 1 and the user to be authenticated can be identified from among the plurality of registered users, the authentication processing unit 134 determines that the user authentication is successful. If the authentication processing unit 134 cannot identify the user to be authenticated from among the plurality of registered users, such as when there is finally no candidate with a degree of matching equal to or higher than the threshold value or when the candidate with a degree of matching equal to or higher than the threshold value does not finally settle down to 1 person (when there are multiple candidates), the authentication processing unit 134 determines that the user authentication fails (rejects the authentication).

[0033] The result output unit 135 obtains the determination result from the authentication processing unit 134 and outputs the determination result to a device inside the terminal 100, such as the display 120. Note that the result output unit 135 may output the determination result to a device outside the terminal 100.

[0034] Next, the above-mentioned filtering setting information 136, filtering method definition information 137, reference map 138, and post-filtering threshold setting information 139 will be described. FIG. 3 is a diagram for explaining the filtering setting information 136. As shown in FIG. 3, the filtering setting information 136 includes an item column 301 and a data column 302. In the filtering setting information 136, an item and a value (data) corresponding to the item are stored in the same row in a mutually associated manner. The filtering setting information 136 is information used to set, for the terminal 100, the criteria for determining whether to perform filtering of the registration template and the criteria for selecting the filtering method. Note that "filtering of the registration template" means narrowing down (identifying one or more registration templates corresponding to candidates for the user to be authenticated) the registration templates corresponding to candidates for the user to be authenticated from among a plurality of registration templates. The "filtering method" is the authentication executed to perform filtering of the registration template.

[0035] In the item column 301, as items, in order from the upper row, an FPIR setting value, an FNIR upper limit value, and filtering selection criteria information are stored. In the data column 302, the value (data) of the corresponding item is stored.

[0036] In the data column 302 corresponding to the FPIR setting value in the item column 301, the setting value of FPIR (False-positive identification-error rate) is stored. FPIR is an index indicating the probability that a target that should not be recognized is erroneously recognized (the probability of false acceptance of others), and is also called the "false acceptance identification rate". It can be said that the smaller the FPIR, the lower the risk of false acceptance of others and the higher the reliability of the system. Note that the "FPIR setting value" may be called the "reference false acceptance identification rate".

[0037] In the data column 302 corresponding to the FNIR upper limit value in the item column 301, the upper limit value of FNIR (False-negative identification-error rate) is stored. FNIR is an indicator that shows the probability that the target to be recognized is misrecognized (the probability of false rejection by the user), and is also called the "false rejection identification rate". The smaller the FNIR, the lower the possibility that the system fails to recognize the original user. Note that the FNIR upper limit value may be called the "upper limit false rejection identification rate".

[0038] In the data column 302 corresponding to the narrowing selection criterion information in the item column 301, the information serving as the criterion when determining (selecting) the narrowing method is stored. In this example, "priority for reducing user complexity" is stored in the data column 302 corresponding to this narrowing selection criterion information. "Priority for reducing user complexity" indicates that the narrowing method with a lower user complexity is preferentially selected. Note that in addition to "priority for reducing user complexity", the data column 302 corresponding to the narrowing selection criterion information in the item column 301 can store "priority for narrowing accuracy", "priority for authentication time", "priority for data volume", or a combination of two or more of these.

[0039] "Priority for narrowing accuracy" is information indicating that when selecting a narrowing method, a method with better narrowing accuracy (an indicator showing the standard of how many candidate numbers can be narrowed down during narrowing) is preferentially selected. "Priority for authentication time" is information indicating that when selecting a narrowing method, a method with better (shorter) authentication time (an indicator showing the standard of the time required for authentication) is preferentially selected. "Priority for data volume" is information indicating that when selecting a narrowing method, a method with better (smaller) required data volume is preferentially selected.

[0040] FIG. 4 is a diagram for explaining the narrowing-down method definition information 137. As shown in FIG. 4, the narrowing-down method definition information 137 includes, as columns (columns) for storing information (values), an ID 401, a method name 402, a narrowing-down accuracy 403, a user complexity 404, an authentication time 405, a required data amount 406, and an availability 407. In the narrowing-down method definition information 137, information corresponding to each column related to the narrowing-down method and the characteristics of the narrowing-down method is associated with each other and stored as row-unit information (records).

[0041] Specifically, the ID 401 stores an identification number that is identification information of the narrowing-down method. The method name 402 stores a method (narrowing-down method) for narrowing down the registration template candidates for the user to be authenticated.

[0042] The narrowing-down accuracy 403 stores information (numerical value) indicating a standard of the narrowing-down accuracy of the corresponding narrowing-down method. For example, 1 / 1000 indicates that when the corresponding narrowing-down method is executed, it is highly likely that the registration template candidates for the user to be authenticated can be narrowed down to about 1 / 1000. When the narrowing-down accuracy is expressed as "uniquely determined" as in the example of FIG. 4, the narrowing-down accuracy is the highest (excellent). When the narrowing-down accuracy is expressed as "a fraction with a numerator of 1" as in the example of FIG. 4, the higher the value of the denominator, the higher (excellent) the narrowing-down accuracy.

[0043] The user complexity 404 stores a parameter (numerical value) indicating the degree of complexity and effort felt by the user when executing the narrowing-down method. In this example, the smaller the numerical value, the better the complexity (the lower the complexity).

[0044] At the authentication time 405, a parameter (time (seconds)) indicating a guideline of the time required for authentication when the user uses the corresponding narrowing-down method is stored. In the required data amount 406, the data amount required to execute the corresponding narrowing-down method is stored. The smaller the data amount, the better the required data amount. In the availability 407, information (yes or no) indicating whether or not the terminal 100 can use the corresponding narrowing-down method is stored.

[0045] FIG. 5 is a diagram for explaining the reference map 138. As shown in FIG. 5, the reference map 138 is a graph in which curves showing the relationship between the false positive identification rate (FPIR) on the horizontal axis and the false negative identification rate (FNIR) on the vertical axis for each number of registered templates (N = 1 to Nmax (Nmax is an arbitrary integer (for example, the number of registered templates used for main biometric authentication (in this example, 10000)))) are plotted in the xy coordinate space. In FIG. 5, for the sake of convenience of explanation, only the curves showing the relationship between FPIR and FNIR for N = 100, N = 1000, and N = 10000 are illustrated. The curve showing the relationship between FPIR and FNIR may be referred to as "biometric authentication performance information".

[0046] FIG. 6 is a diagram for explaining the post-narrowing threshold setting information 139. As shown in FIG. 6, the post-narrowing threshold setting information 139 includes an item column 601 and a data column group 602 including a plurality of data columns.

[0047] In the post-narrowing threshold setting information 139, an item and a value (data) corresponding to the item are stored in association with each other in the same row. Further, in the same column of the post-narrowing threshold setting information 139, the value of the number of candidates and the value of the authentication threshold are stored in association with each other.

[0048] In the data column group 602 corresponding to the number of candidates in item column 601, the number of candidates (1 to Nmax) of the registration template is stored. In the data column group 602 corresponding to the authentication threshold of item column 601, the value of the authentication threshold corresponding to the number of candidates of the registration template is stored. Note that this value of the authentication threshold is a value of the authentication threshold (a value corresponding to a point on the curve) such that the set value (FPIR = 0.0001) of the narrowing-down setting information 136 on the curve of the reference map 138 corresponding to the number of candidates of the registration template is calculated and stored in advance. The threshold stores a value of a relatively strict threshold (a threshold that keeps the FPIR at a low value (= 0.0001)), and by using this threshold for the main biometric authentication, the performance (FPIR) of the main biometric authentication can be kept low.

[0049] In the server 200 of FIG. 1, the computer 2000 shown in FIG. 2 is applied. The CPU 2001 of the server 200 realizes various functions by executing the program stored in the ROM 2002. The server 200 may be composed of a plurality of computers 2000, and is not limited to a physical computer 2000, but may be a virtual computer 2000. The computer 2000 may be computing resources and storage resources provided by the cloud, and the cloud may provide the functions provided by the server 200. The server 200 stores (stores, holds) the authentication information 210 in which the registration template or the registration authentication code is associated with the user information (for example, the user ID for identifying the user) in the storage device 2004.

[0050] FIG. 7 is a diagram for explaining the authentication information 210. As shown in FIG. 7, the authentication information 210 includes, as columns (columns) for storing information (values), ID 201, narrowing-down information 1 (iris information) 202, narrowing-down information 2 (palm print) 203, narrowing-down information 3 (unique PIN) 204, and an encrypted template 205. In the authentication information 210, the information corresponding to each column used for authentication is associated with each other and stored as row unit information (records).

[0051] Specifically, the ID 201 stores an identification number which is the identification information of the registered user. The narrowing-down information 1 (iris information) 202 stores the iris information of the corresponding registered user as a registration template for iris authentication. The narrowing-down information 2 (palm print) 203 stores the palm print information of the corresponding registered user as a registration template for palm print authentication. The narrowing-down information 3 (unique PIN) 204 stores the PIN information of the corresponding user as PIN information for PIN authentication. The encrypted template 205 stores an encrypted template obtained by encrypting the features of the face image of the corresponding user.

[0052] Note that the face authentication performed by the terminal 100 may use a known PBI (Public Biometrics Infrastructure). Since PBI is well-known, to briefly explain, in order to use biometric authentication information like a private key, a PBI template (also called a "public template") that is difficult to restore to the original biometric authentication information is generated by one-way converting the biometric authentication information embedded with the private key. It is an individual authentication infrastructure that realizes authentication, signature, and encryption by utilizing the PBI template. This face authentication involves the camera 110 of the terminal 100 capturing the user's face to obtain a face image (face information), and obtaining data indicating the features of the face (hereinafter referred to as "face feature information" or "biometric private key") extracted (generated) based on the face image, and performing authentication using the face feature information like a private key. In this face authentication, at the time of registration, the terminal 100 generates (extracts) face feature information (biometric private key) based on the face image captured by the camera 110 of the user to be registered, creates a private key and a public key by an existing encryption technology, embeds the private key in the face feature information (biometric private key) and performs one-way conversion, and registers the generated information (PBI template) and the public key in the server 200. The PBI template and the public key may be referred to as "biometric public key" for convenience. Note that after registration, the biometric private key and the biometric public key are discarded from the terminal 100. In this face authentication, at the time of authentication, the terminal 100 obtains a face image by capturing the face of the user to be authenticated with the camera 110, generates a biometric private key based on the face image, and the server 200 performs authentication using the biometric private key and the registered biometric public key. The biometric private key (face feature information) is discarded from the terminal 100 after authentication. When the face authentication performed by the terminal 100 uses a known PBI, the value (data) stored in the encryption template 205 may be the registered PBI template.

[0053] <Overview> The terminal 100 obtains from the server 200 the number of registered templates of the authentication information 210 corresponding to the main face authentication. Based on the index of the performance of the main biometric authentication (face authentication in this example) corresponding to the obtained number of registered templates, when the FPIR is the reference value (FPIR setting value) of the narrowing-down setting information 136, the terminal 100 determines whether the FNIR is equal to or less than the FNIR upper limit value of the narrowing-down setting information.

[0054] Here, the index of the performance of the main biometric authentication can be represented by a curve showing the relationship between the FPIR (false acceptance identification rate) and the FNIR (false rejection identification rate) of the reference map 138 shown in FIG. 5. This curve indicates that the FPIR and the FNIR are in a trade-off relationship.

[0055] At the terminal 100, according to the points on the curve of the reference map 138, a threshold for verification corresponding to the performance of the biometric authentication represented by the FPIR value and the FNIR value of that point can be defined. In other words, by changing the threshold for verification, the position (FPIR value and FNIR value) of the point on the curve corresponding to that threshold can be changed (that is, the performance of the main biometric authentication can be changed). Also, the curve changes according to the number of registered templates (the number of candidates for registered templates) of the main biometric authentication. Specifically, when the FPIR is fixed, the FNIR becomes smaller as the number of registered templates corresponding to the candidates of the user to be authenticated decreases. When the FNIR is fixed, the FPIR becomes smaller as the number of registered templates (the number of candidates for registered templates) decreases.

[0056] First, the terminal 100 identifies the curve corresponding to this number of registered templates based on the number of registered templates used for the main biometric authentication.

[0057] The terminal 100 applies to the curve that specifies the FPIR setting value of the narrowing-down setting information, and obtains the FNIR value corresponding to the FPIR setting value. That is, it obtains the values of the points on the curve indicating the biometric authentication performance of the terminal 100 (= FPIR setting value, FNIR value). These obtained values are the FPIR value and FNIR value of the terminal 100 at this time (that is, the FPIR value and FNIR value indicating the authentication performance (the authentication performance planned to be set) of the terminal 100 at this time).

[0058] The terminal 100 determines whether the FNIR value indicating the authentication performance planned to be set is less than or equal to the upper limit value. When the FNIR value planned to be set is less than or equal to the FNIR upper limit value, the terminal 100 determines that the FNIR also meets the standard, and performs main biometric authentication using the threshold value for verification (the threshold value corresponding to the FPIR value and FNIR value planned to be set). Note that the threshold value corresponding to the FPIR value and FNIR value can be calculated by referring to a table (not shown) that associates the number of registered templates with the threshold value. As this threshold value, the threshold value corresponding to the point where the FPIR becomes the FPIR setting value on the curve corresponding to the number of registered templates is set.

[0059] On the other hand, when the FNIR value planned to be set by the terminal 100 is greater than the FNIR upper limit value, since the FNIR does not meet the standard, it is necessary to change the threshold value for verification (the threshold value corresponding to the FPIR value and FNIR value planned to be set). If the terminal 100 performs main biometric authentication using the current threshold value as it is, although the FPIR meets the standard, the FNIR may not meet the standard, and the reliability of user authentication may decrease (the FNIR is relatively high).

[0060] Therefore, the terminal 100 executes a narrowing-down method to narrow down the registered templates corresponding to the candidates of the user to be authenticated (the number of registered template candidates) from among the registered templates of multiple registered users for main biometric authentication. Which narrowing-down method to use at this time is determined based on the narrowing-down selection criterion information in the narrowing-down setting information 136. In this example, the narrowing-down selection criterion information is "priority for reducing user complexity".

[0061] In this case, the terminal 100 selects, from among a plurality of narrowing-down methods in the narrowing-down method definition information 137, the method with the smallest user complexity 404, and selects that method as the narrowing-down method to be executed. In the case of this example, iris authentication with the smallest user complexity 404 is selected and executed. By executing the selected narrowing-down method, the terminal 100 narrows down the registration template corresponding to the candidate user from among the registration templates of a plurality of registered users for the main biometric authentication.

[0062] The terminal 100 sets, as the threshold for verification, the threshold corresponding to the point corresponding to FPIR 0.0001 on the curve corresponding to the number of narrowed-down registration templates (the number of registration template candidates) in the reference map 138. This threshold is set such that the FNIR value corresponding to that value is smaller than the FNIR value corresponding to the threshold corresponding to the number of registration templates before narrowing down.

[0063] In this example, the narrowing-down post-threshold setting information 139 stores the number of registration templates and the threshold corresponding to the FPIR reference value (FPIR 0.0001) corresponding to the number of registration templates. The terminal 100 refers to the narrowing-down post-threshold setting information 139, obtains the threshold corresponding to the number of narrowed-down registration templates, and sets that threshold as the threshold to be used for the main biometric authentication. By narrowing down the number of registration template candidates by executing the narrowing-down method and setting the threshold corresponding to the point on the curve corresponding to the reference setting value (PAIR setting value), the FPIR of the authentication performance when performing the main biometric authentication can be made to meet the reference (while reducing the risk of false acceptance by others), and the FNIR of the authentication performance can be reduced.

[0064] After executing the narrowing-down method, the terminal 100 performs the main biometric authentication using the set verification threshold and the narrowed-down registration templates. The performance of the main biometric authentication at this time meets the FPIR setting value (reference) as described above and is lower than the FNIR upper limit value.

[0065] As described above, the biometric authentication system according to the first embodiment can reduce the false rejection of the user (the possibility that the authentication fails even though the user is the genuine one) while keeping the risk of misaccepting others low. Furthermore, since the biometric authentication system according to the first embodiment selects and executes a method with a low level of user complexity as the narrowing-down method, it can also ensure the ease of authentication.

[0066] <Specific operation> FIG. 8 is a flowchart showing the processing flow executed by the authentication control unit 132. When the authentication control unit 132 acquires a face image for authentication (biometric authentication information from the face image) of the user to be authenticated, it starts processing from step 800 in FIG. 8 and sequentially executes the processes of step 805 and step 810 described below, and then proceeds to step 815.

[0067] Step 805: The authentication control unit 132 reads and acquires the narrowing-down setting information 136.

[0068] Step 810: The authentication control unit 132 acquires the number of registered templates to be used for face authentication, which is the main biometric authentication, from the authentication information 210 of the server 200.

[0069] When the authentication control unit 132 proceeds to step 815, it determines whether it is outside the narrowing-down setting. Specifically, the authentication control unit 132 uses the curve corresponding to the number of registered templates acquired in step 810 of the reference map 138. The authentication control unit 132 determines whether it is outside the narrowing-down setting based on the curve and the narrowing-down setting information 136.

[0070] More specifically, the authentication control unit 132 acquires the curve corresponding to the number of registered templates from the reference map 138. The authentication control unit 132 acquires the FNIR value corresponding to the FPIR setting value of the narrowing-down setting information 136 on the curve. The authentication control unit 132 compares the FNIR value with the FNIR upper limit value of the narrowing-down setting information 136.

[0071] When the comparison result is that the FNIR value is greater than the FNIR upper limit value, since the FNIR is greater than the reference (and does not meet the reference), the possibility of false rejection by the user increases. Therefore, in this case, the authentication control unit 132 determines that it is outside the narrowing-down setting.

[0072] When the comparison result is that the FNIR value is less than or equal to the FNIR upper limit value, since the FNIR is less than or equal to the reference (and meets the reference), the possibility of false rejection by the user is small. Therefore, in this case, the authentication control unit 132 determines that it is within the narrowing-down setting.

[0073] As shown in FIG. 9A, for example, assume that the number of registered templates is N = 10000. In this case, the authentication control unit 132 acquires the curve corresponding to N = 10000 from the reference map 138. The authentication control unit 132 acquires the FNIR value (= 0.2) corresponding to the FPIR setting value (= 0.0001) of the narrowing-down setting information 136 on the curve. The authentication control unit 132 compares the FNIR value (= 0.2) with the FNIR upper limit value (= 0.1) of the narrowing-down setting information 136. Since the comparison result is that the FNIR value (= 0.2) is greater than the FNIR upper limit value (= 0.1), the authentication control unit 132 determines that it is outside the narrowing-down setting.

[0074] As shown in FIG. 9B, for example, assume that the number of registered templates is N = 100. In this case, the authentication control unit 132 acquires the curve corresponding to N = 100. The authentication control unit 132 acquires the FNIR value (= 0.01) corresponding to the FPIR setting value (= 0.0001) of the narrowing-down setting information 136 on the curve. The authentication control unit 132 compares the FNIR value (= 0.01) with the FNIR upper limit value (= 0.1) of the narrowing-down setting information 136. Since the comparison result is that the FNIR value (= 0.01) is less than or equal to the FNIR upper limit value (= 0.1), the authentication control unit 132 determines that it is within the narrowing-down setting.

[0075] When it is outside the filtering setting, the authentication control unit 132 determines "YES" in step 815, sequentially executes the processes of steps 820 to 840 described below, and then proceeds to step 895 to temporarily end this processing flow.

[0076] Step 820: The authentication control unit 132 selects a filtering means (filtering method) based on the filtering selection criteria information and the filtering method definition information 137 of the filtering setting information 136. Specifically, the authentication control unit 132 acquires the filtering selection criteria information from the filtering setting information 136. In this example, since the filtering selection criteria information is "priority for reducing user complexity", the authentication control unit 132 selects the iris (photo) with ID 401 having the smallest user complexity 404 among the methods for which the availability 407 of the filtering method definition information 137 is "available".

[0077] Step 825: The authentication control unit 132 filters out the registration templates (the number of registration template candidates) corresponding to the candidate users to be authenticated from among the registration templates of multiple registered users by executing the selected filtering method. In this example, the authentication control unit 132 acquires the iris information of the user to be authenticated from the already acquired face image. The authentication control unit 132 collates the iris information with the registration templates of the authentication information 210 (the multiple registration templates stored in the filtering information 1 (iris information)), compares the collation degree with the threshold value, specifies the registration templates with a collation degree equal to or higher than the threshold value as the candidates for the user to be authenticated, and specifies the main biometric registration template corresponding to the specified user candidates, thereby filtering out the number of registration template candidates.

[0078] Step 830: The authentication control unit 132 sets a threshold value based on the candidate target number (the number of candidate registered templates) and the threshold value setting information 139 after narrowing down by the authentication threshold value setting unit 133. Specifically speaking, the authentication control unit 132 obtains, by the authentication threshold value setting unit 133, an authentication threshold value corresponding to the number of candidate registered templates narrowed down in step 825 from the threshold value setting information 139 after narrowing down, and sets the authentication threshold value as the threshold value used for the main biometric authentication.

[0079] Step 835: The authentication control unit 132 uses the set threshold value by the authentication processing unit 134 to calculate the degree of matching between the already acquired biometric authentication information (face image) and the registered template for each of the candidates of the registered templates narrowed down as described above by the authentication processing unit 134, and executes the main biometric authentication based on the degree of matching and the set threshold value. When the number of candidates for the authentication result (registered user identified as the user to be authenticated) whose degree of matching is equal to or higher than the threshold value is 1, the authentication processing unit 134 determines that the user authentication is successful. When the number of candidates for the authentication result (registered user identified as the user to be authenticated) whose degree of matching is equal to or higher than the threshold value is not 1, the authentication processing unit 134 determines that the user authentication fails (rejects the authentication).

[0080] Step 840: The authentication control unit 132 outputs the authentication result in step 835 by the result output unit 135.

[0081] In addition, if it is outside the narrowing-down setting in step 815, the authentication control unit 132 determines "NO" in step 815, proceeds to step 835, and as described above, the authentication processing unit 134 calculates the degree of matching between the biometric authentication information (face image) and each of the registered templates, and based on the degree of matching and the threshold value, executes the main biometric authentication. As the threshold value in this case, the threshold value corresponding to the point on the curve corresponding to the number of registered templates before narrowing down (the point where FPIR is the reference value) is used. When the number of candidates for the authentication result with a degree of matching equal to or higher than the threshold value is one, the authentication processing unit 134 determines that the user's authentication has succeeded. When the number of candidates for the authentication result with a degree of matching equal to or higher than the threshold value is not one, the authentication processing unit 134 determines that the user's authentication has failed (rejects the authentication). Thereafter, the authentication control unit 132 proceeds to step 840, and after the result output unit 135 outputs the authentication result in step 835, proceeds to step 895 to temporarily end this processing flow.

[0082] <Effect> As described above, in the biometric authentication system according to the first embodiment of the present invention, when the performance of the main biometric authentication corresponding to the number of registered templates of the main biometric authentication satisfies the reference FPIR and the FNIR does not satisfy the standard, a plurality of registered templates are narrowed down to the registered templates corresponding to the candidates for the registered users to be identified from the user to be authenticated, and the threshold value corresponding to the reference FPIR is set. Thereby, the biometric authentication system sets the performance of the main biometric authentication so that the FNIR decreases while maintaining the reference FPIR. Therefore, the biometric authentication system can reduce the false rejection of the user while keeping the risk of misaccepting others low.

[0083] Furthermore, the biometric authentication system selects a narrowing-down method so as to prioritize reducing the user's complexity, and executes a narrowing-down method (for example, iris authentication that can use the already captured face image also used for main biometric authentication) with excellent ease of authentication in the selected narrowing-down method, so that candidates for registered templates can be narrowed down. Therefore, it is not necessary to newly acquire biometric authentication information (acquire user data for acquiring biometric authentication information) for the narrowing-down method, and the ease of authentication can also be ensured.

[0084] <<First to Fourth Modification Examples>> Hereinafter, each modification example of the biometric authentication system according to the first embodiment of the present invention will be described. In the following first to third modification examples, the narrowing-down method definition information 137 shown in FIG. 10 is used instead of the narrowing-down method definition information 137 in FIG. 4. This narrowing-down method definition information 137 is different from the narrowing-down method definition information 137 shown in FIG. 4 except that the availability 407 of the row with the ID401 value of "5" is changed from "unavailable" to "available", and the availability 407 of the row with the ID401 value of "7" is changed from "available" to "unavailable". In addition, the terminal 100 can also perform finger vein authentication, and the authentication information 210 stores the registered templates of each of a plurality of registered users used for finger vein authentication (not shown, but the authentication information 210 has an added column for narrowing-down information 4 (finger vein authentication), and the registered templates are stored in that column). In addition, in the first to third modification examples, only narrowing-down methods that do not require manual input of the PIN code are available, so that the ease of authentication can be further ensured.

[0085] <<First Modification Example>> The first modification example of the biometric authentication system according to the first embodiment of the present invention will be described. In the first modification example, "narrowing-down accuracy priority" is stored in the data column 302 corresponding to the narrowing-down selection criterion information of the narrowing-down setting information 136.

[0086] In the first modification example, in step 820 of FIG. 8, since the narrowing selection criterion information gives priority to narrowing accuracy, the authentication control unit 132 selects the finger vein with the best narrowing accuracy among the methods in which the availability 407 of the narrowing method definition information 137 in FIG. 10 is "available". In step 825, the authentication control unit 132 acquires the finger vein information of the user to be authenticated, and the authentication processing unit 134 collates the finger vein information with a plurality of registered templates stored in the registration template (narrowing information (finger vein information) (not shown)) of the authentication information 210, compares the degree of collation with a threshold value, identifies the registered user of the registered template whose degree of collation is equal to or higher than the threshold value as a candidate for the user to be authenticated, and identifies the main biometric registration template corresponding to the identified user candidate, thereby narrowing down the number of candidate registered templates. Thereafter, the authentication control unit 132 executes the processes of steps 830 and 835 described above to set a threshold value according to the number of narrowed-down registered templates, and performs main biometric authentication using the set threshold value and the narrowed-down registered templates. According to the first modification example, it is possible to select a narrowing method so as to give priority to narrowing accuracy, and execute the selected narrowing method with excellent narrowing accuracy (for example, finger vein authentication).

[0087] <<Second Modification Example>> A second modification example of the biometric authentication system according to the first embodiment of the present invention will be described. In the second modification example, "authentication time priority" is stored in the data sequence 302 corresponding to the narrowing selection criterion information of the narrowing setting information 136.

[0088] In the first modification example, in step 820 of FIG. 8, since the narrowing selection criterion information gives priority to authentication time, the authentication control unit 132 gives priority to the iris (photo) with the shortest authentication time and the finger vein with the smaller ID401 among the methods in which the availability 407 of the narrowing method definition information 137 is "available", and selects the iris (photo).

[0089] The authentication control unit 132 obtains the iris information of the user to be authenticated from the face image in step 825, and the authentication processing unit 134 compares the iris information with the registration templates of the authentication information 210 (a plurality of registration templates stored in the narrowing information 1 (iris information)), compares the matching degree with the threshold value, and identifies the registered user of the registration template whose matching degree is equal to or higher than the threshold value as a candidate for the user to be authenticated, and identifies the main biometric registration template corresponding to the identified user candidate, thereby narrowing down the number of candidate registration templates. Thereafter, the authentication control unit 132 sets a threshold value according to the number of the narrowed-down registration templates by executing the processes of the above-described steps 830 and 835, and performs the main biometric authentication using the set threshold value and the narrowed-down registration templates. According to the second modification example, it is possible to select a narrowing-down method so as to prioritize the authentication time and execute the selected narrowing-down method with excellent authentication time (for example, iris authentication). Thereby, the second modification example can shorten the authentication time when the narrowing-down method is executed as compared with the first embodiment and other modification examples.

[0090] <<Third Modification Example>> A third modification example of the biometric authentication system according to the first embodiment of the present invention will be described. In the third modification example, "data volume priority" is stored in the data series 302 corresponding to the narrowing-down selection criterion information of the narrowing-down setting information 136.

[0091] In the first modification example, at step 820 in FIG. 8, since the narrowing selection criterion information is data volume priority, the authentication control unit 132 selects a finger vein with a small required data volume from among the methods where the availability 407 of the narrowing method definition information 137 is "available". At step 825, the authentication control unit 132 acquires the finger vein information of the user to be authenticated, and the authentication processing unit 134 collates the finger vein information with a plurality of registered templates (narrowed information (finger vein information) (not shown) stored in the registered template of the authentication information 210), compares the collation degree with the threshold value, and identifies the registered user of the registered template whose collation degree is equal to or higher than the threshold value as a candidate for the user to be authenticated, and by identifying the main biometric authentication registered template corresponding to the identified user candidate, the number of candidate registered templates is narrowed down. Thereafter, the authentication control unit 132 executes the processes of steps 830 and 835 described above to set a threshold value according to the number of narrowed registered templates, and performs main biometric authentication using the set threshold value and the narrowed registered templates. According to the third modification example, it is possible to select a narrowing method so as to prioritize the required data volume and execute the selected narrowing method with excellent required data volume (for example, finger vein authentication). Thereby, the third modification example can reduce the required data volume when executing the narrowing method as compared with the first embodiment and other modification examples.

[0092] <<Fourth Modification Example>> A fourth modification example of the biometric authentication system according to the first embodiment of the present invention will be described. In the fourth modification example, "narrowing accuracy priority and complexity priority" is stored in the data sequence 302 corresponding to the narrowing selection criterion information of the narrowing setting information 136. In the following fourth modification example, the narrowing method definition information 137 shown in FIG. 11 is used instead of the narrowing method definition information 137 in FIG. 4. This narrowing method definition information 137 is different from the narrowing method definition information 137 shown in FIG. 4 except that the availability 407 of the row where the value of ID401 is "7" is changed from "available" to "not available".

[0093] In the fourth modification example, at step 820 in FIG. 8, since the narrowing-down selection criterion information is "narrowing-down accuracy priority and complexity priority", the authentication control unit 132 selects the method with the best narrowing-down accuracy among the methods for which the availability 407 of the narrowing-down method definition information 137 is "permitted". If there are multiple selected methods, the method with the best complexity among them is selected. In this example, the iris is selected. The authentication control unit 132 acquires iris information of the user to be authenticated from the face image at step 825, and the authentication processing unit 134 collates the iris information with the registered templates of the authentication information 210 (the multiple registered templates stored in the narrowing-down information 1 (iris information)), compares the collation degree with the threshold value, and identifies the registered user of the registered template whose collation degree is equal to or higher than the threshold value as a candidate for the user to be authenticated. By identifying the main biometric registered template corresponding to the identified user candidate, the number of candidate registered templates is narrowed down. Thereafter, the authentication control unit 132 sets a threshold value according to the number of the narrowed-down registered templates by executing the processes of step 830 and step 835 described above, and performs the main biometric authentication using the set threshold value and the narrowed-down registered templates. According to the fourth modification example, the narrowing-down method can be selected with the narrowing-down accuracy as the first priority and the complexity as the second priority, and the selected narrowing-down method (for example, iris authentication) can be executed.

[0094] Although the first to fourth modification examples have been described above, the narrowing-down selection criterion information is not limited to the above examples, and various information indicating the criteria for selecting the narrowing-down method may be set. In this case, the narrowing-down method is selected according to the selection criteria indicated by the various set narrowing-down selection criterion information.

[0095] <<Second Embodiment>> The biometric authentication system according to the second embodiment of the present invention will be described. The biometric authentication system according to the second embodiment of the present invention differs from the biometric authentication system according to the first embodiment only in the following points. · When the number of candidates for the authentication result is other than one when the first main biometric authentication is executed, the biometric authentication system according to the second embodiment performs resetting of the threshold value or execution (or re - execution) of the narrowing - down method according to the number of candidates, and then performs the main biometric authentication again.

[0096] Hereinafter, the description will focus on this difference.

[0097] FIG. 12 shows a configuration example of the biometric authentication system according to the second embodiment. As shown in FIG. 12, the information processing apparatus 130 of the biometric authentication system according to the second embodiment has the small - number - of - candidates threshold - setting information 140 stored (stored, held) in the storage device 2004.

[0098] FIG. 13 is a diagram for explaining the small - number - of - candidates threshold - setting information 140. As shown in FIG. 13, the small - number - of - candidates threshold - setting information 140 includes an item column 1301 and a data - column group 1302 including a plurality of data columns. In the small - number - of - candidates threshold - setting information 140, an item and a value (data) corresponding to the item are stored in the same row in an associated manner. Further, in the same column of the data - column group 1302 of the small - number - of - candidates threshold - setting information 140, the value of the number of candidates, the value of the FNIR setting value, and the value of the threshold are stored in an associated manner with each other.

[0099] In the data column group 1302 corresponding to the number of candidates in item column 1301, the number of candidates for the registration template is stored. In the data column group 1302 corresponding to the FNIR setting value in item column 1301, the FNIR setting value corresponding to the number of candidates for the registration template is stored. In the data column group 1302 corresponding to the threshold value in item column 1301, the value of the threshold corresponding to the number of candidates for the registration template is stored. Note that in the data column group 1302 corresponding to this threshold value, the performance of the main biometric authentication is expressed by the points on the curve of the reference map 138 corresponding to the number of candidates for the registration template corresponding to the threshold value (= FNIR setting value corresponding to the threshold value, FPIR value corresponding to the FNIR setting value), and values are stored such that the performance is achieved. That is, the threshold setting information 140 for a small number of candidates is information that defines the threshold according to a high FNIR setting value (in other words, an FPIR setting value lower than the FPIR setting value of the narrowing-down method definition information 137), unlike the post-narrowing threshold setting information 139. Since the threshold is defined in this way, when the number of candidates becomes relatively small, the person can be determined with a more rigorous determination.

[0100] For example, when the number of candidates for the registration template corresponding to the user to be authenticated is 2, as shown in FIG. 14, on the curve of the reference map 138 corresponding to the number of candidates for the registration template being 2, the value of the threshold corresponding to the point where FNIR is 0.35 (= FNIR setting value corresponding to the threshold value, FPIR value corresponding to the FNIR setting value (FPIR value lower than the FPIR setting value of the narrowing-down method definition information 137))) (the threshold value such that the performance of the main biometric authentication becomes the performance expressed by the FNIR value and FNIR value of the above point) is stored. When the remaining number of candidates decreases, by increasing the threshold so that the FNIR of the performance of the main biometric authentication becomes relatively high (in other words, the FPIR becomes even lower), it is possible to narrow down one person from a small number of candidates for the registration template by a more rigorous determination. Note that in this example, the FNIR setting value is changed according to the number of candidates, but it may also be a fixed predetermined value that does not change according to the number of candidates.

[0101] <Specific operation> FIG. 15 is a flowchart showing a processing flow executed by the authentication control unit 132. When the authentication control unit 132 acquires a face image for authentication (biometric authentication information from the face image) of a user to be authenticated, it starts processing from step 1500 in FIG. 15 and proceeds to step 1505. When the authentication control unit 132 proceeds to step 1505, it starts processing from step 800 in FIG. 8, executes an appropriate process among the above-described steps 805 to 835, and then proceeds to step 1510 in FIG. 15 via step 895. In the first embodiment, in step 835, when the number of candidates for the authentication result with a matching degree equal to or higher than the threshold is one person, the authentication processing unit 134 determines that the user authentication is successful, and when the number of candidates for the authentication result with a matching degree equal to or higher than the threshold is not one person, it determines that the user authentication has failed. In contrast, in the second embodiment, the authentication processing unit 134 does not determine the success or failure of authentication at the time of step 835, but only calculates the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated).

[0102] When the authentication control unit 132 proceeds to step 1510, it determines whether the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) executed in step 835 of FIG. 8 is zero, one, or two or more.

[0103] When the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) is one person, the authentication control unit 132 determines that the authentication is successful and proceeds to step 1515, and the result output unit 135 outputs the result of authentication passed (authentication successful). Thereafter, the authentication control unit 132 proceeds to step 1595 and temporarily ends this processing flow.

[0104] When the number of candidates for the result of the main biometric authentication (registered users identified as the user to be authenticated) is two or more, the authentication control unit 132 proceeds to step 1520 and refers to the threshold setting information 140 for a small number of candidates based on the remaining number of candidate targets (number of candidates) to reset the threshold. Specifically, the authentication control unit 132 causes the authentication threshold setting unit 133 to reset the threshold corresponding to the number of candidates in the threshold setting information 140 for a small number of candidates as the threshold to be used for the main biometric authentication (step 1525) again. As described above, the threshold in the threshold setting information 140 for a small number of candidates is a threshold defined so that the performance of the main biometric authentication becomes a relatively high FNIR value (in other words, an FPIR lower than the FPIR setting value in the narrowing-down method definition information 137). Therefore, if the main biometric authentication is executed again using this reset threshold, a more rigorous determination (strict identification of the person) can be made for the narrowed-down small number of registered templates.

[0105] Thereafter, the authentication control unit 132 proceeds to step 1525 and executes the main biometric authentication again for the candidates of the registered users (candidates for registered templates) in the result of the main biometric authentication executed in step 1505 (step 835 in FIG. 8). Thereafter, the authentication control unit 132 proceeds to step 1530 and determines whether the number of candidates for the result of the main biometric authentication (registered users identified as the user to be authenticated) executed in step 1525 is one person or otherwise.

[0106] When the number of candidates for the result of the main biometric authentication (registered users identified as the user to be authenticated) executed in step 1525 is otherwise, the authentication control unit 132 determines that the authentication has failed and proceeds to step 1535, and the result output unit 135 outputs the result of authentication failure. Thereafter, the authentication control unit 132 proceeds to step 1595 and temporarily ends this processing flow.

[0107] If the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) is one, the authentication control unit 132 determines that the authentication has passed (authentication succeeded) and proceeds to step 1515, and the result output unit 135 outputs the result of the authentication passing (authentication succeeded). Thereafter, the authentication control unit 132 proceeds to step 1595 and temporarily ends this processing flow.

[0108] If the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) is zero in step 1510 described above, the authentication control unit 132 proceeds to step 1540 and determines whether or not narrowing down has been executed up to the previous step (before the current time point) (that is, whether or not the narrowing-down method in step 825 of FIG. 8 has been executed).

[0109] If narrowing down has not been executed up to the previous step, the authentication control unit 132 determines "NO" in step 1540 and proceeds to step 1545. When the authentication control unit 132 proceeds to step 1545, it executes the same processing as steps 820 to 835 of FIG. 8 to narrow down the registered templates, re-set the threshold value, and execute biometric authentication, and then proceeds to step 1555. As a result, the registered templates are narrowed down, and biometric authentication is executed (step 835) with the threshold value corresponding to the FNIR smaller than the FNIR of the performance of the main biometric authentication executed previously and the narrowed-down registered templates.

[0110] If narrowing down has been executed up to the previous step, the authentication control unit 132 determines "YES" in step 1540 and proceeds to step 1550. After referring to the narrowing-down method definition information 137 and selecting a narrowing-down method with a narrowing-down accuracy smaller than the narrowing-down accuracy of the previous narrowing-down method, it executes the same processing as steps 825 to 835 of FIG. 8 and then proceeds to step 1555. As a result, the registered templates are narrowed down, and the main biometric authentication is executed again (step 835) with the narrowed-down registered templates and the threshold value corresponding to the FNIR smaller than the previous one.

[0111] When the authentication control unit 132 proceeds to step 1555, it determines whether the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) executed in the previous step (step 1545 or step 1550) is one person and which of the others it is.

[0112] If the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) executed in the previous step is one person, the authentication control unit 132 determines that authentication has passed (authentication is successful) and proceeds to step 1515, and the result output unit 135 outputs the result of authentication passing (authentication being successful). Thereafter, the authentication control unit 132 proceeds to step 1595 and temporarily ends this processing flow.

[0113] If the number of candidates for the result of the main biometric authentication (registered user identified as the user to be authenticated) executed in the previous step is other than one person, the authentication control unit 132 determines that authentication has failed and proceeds to step 1560, and the result output unit 135 outputs the result of authentication failure. Thereafter, the authentication control unit 132 proceeds to step 1595 and temporarily ends this processing flow.

[0114] <Effect> As described above, the biometric authentication system according to the second embodiment of the present invention, similar to the first embodiment, when the performance of biometric authentication corresponding to the number of registered templates of the main biometric authentication satisfies the standard FPIR and the FNIR does not satisfy the standard, after narrowing down a plurality of registered templates to the registered templates corresponding to the candidates for the registered user identified as the user to be authenticated, a threshold value corresponding to the standard FPIR is set. Thereby, the biometric authentication system sets the performance of biometric authentication so that the FNIR decreases while maintaining the standard FPIR. Therefore, the biometric authentication system can reduce the false rejection of the user while keeping the risk of false acceptance of others low.

[0115] Furthermore, when authentication fails after the main biometric authentication is performed once (when the candidates cannot be narrowed down to one person, or when there are no candidates), the biometric authentication system according to the second embodiment re-sets the threshold value or executes (or re-executes) the narrowing-down method according to the number of candidates in the biometric authentication result, and then performs the main biometric authentication again. As a result, the biometric authentication system can change the threshold value according to the number of candidates in the biometric authentication result, reduce the false rejection of the user while keeping the false acceptance risk of others low.

[0116] <<Other Modification Examples>> The present invention is not limited to the above-described embodiments and modification examples, and various modification examples can be adopted within the scope of the present invention. Furthermore, the above-described embodiments and modification examples can be combined with each other as long as they do not deviate from the scope of the present invention.

[0117] In each of the above-described embodiments and modification examples, for example, as shown in FIG. 16, the terminal 100 and the information processing apparatus 130 may have a separate configuration. The computer 1600 included in the terminal 100 has the same configuration as the computer 2000 shown in FIG. 2. In this case, as shown in FIG. 17, the server 200 may be omitted, and instead of the server 200, the information processing apparatus 130 may provide the functions of the server 200. In this case, the terminal 100 in FIGS. 16 and 17 functions as a biometric authentication information acquisition device that acquires the user's biometric authentication information. The terminal 100 acquires the user's biometric authentication information and transmits it to the information processing apparatus 130 via a network (not shown). The information processing apparatus 130 performs biometric authentication using the biometric authentication information as described above.

[0118] In each of the above embodiments and each modification example, although the main biometric authentication is face authentication, the main biometric authentication may be biometric authentication other than face authentication, and face authentication may be included in the narrowing-down method. In each of the above embodiments, although one narrowing-down method is selected, a combination of a plurality of narrowing-down methods may be selected and executed. Typically, as the narrowing-down method, one with authentication performance different from the performance of the main biometric authentication (the performance defined by the narrowing-down setting information 136) (lower or higher performance) may be used. In the second embodiment, when the number of candidates for the result of biometric authentication is equal to or less than a predetermined threshold number of people, the threshold setting information 140 for a small number of candidates may be used, and when the number of candidates for the result of biometric authentication is more than a predetermined threshold, it may be determined as authentication failure.

[0119] In each of the above embodiments and each modification example, the post-narrowing threshold setting information 139 may be omitted, and the relationship between the values of the points on the curve of the reference map 138 (FPIR value, FNIR value) and the threshold corresponding to the performance of the main biometric authentication represented by the values of the points on the curve (FPIR value, FNIR value) may be obtained in advance. In this case, from the number of registered templates and the reference map 138, the values of the points on the curve (FPIR value, FNIR value) corresponding to the FPIR set value are specified, and the specified point values (FPIR value, FNIR value) are applied to the above relationship to calculate the threshold, and the threshold may be set as the threshold used for the main biometric authentication.

Explanation of Signs

[0120] 100…Terminal, 110…Camera, 120…Display, 130…Information processing device, 131…User information reading unit, 132…Authentication control unit, 133…Authentication threshold setting unit, 134…Authentication processing unit, 135…Result output unit, 136…Narrowing-down setting information, 137…Narrowing-down method definition information, 138…Reference Map (FPIR, FNIR, N), 139…Post-narrowing threshold setting information

Claims

1. An information processing apparatus including an arithmetic unit that collates biometric authentication information of a user to be authenticated with registered biometric authentication information of a plurality of registered users and performs biometric authentication to identify the user to be authenticated from among the plurality of registered users, wherein the arithmetic unit, acquires the biometric authentication information of the user to be authenticated, the number of the registered biometric authentication information, and biometric authentication performance information indicating the relationship between the false acceptance identification rate and the false rejection identification rate of the biometric authentication corresponding to the number of the registered biometric authentication information, and when the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information does not satisfy a predetermined criterion, collates authentication information related to the biometrics or knowledge of the user to be authenticated with the registered authentication information of the plurality of registered users, selects one or more narrowing-down methods from among a plurality of narrowing-down methods for identifying candidates for the user to be authenticated from among the plurality of registered users, acquires the authentication information of the user to be authenticated according to the selected narrowing-down method, and narrows down the registered biometric authentication information corresponding to the candidates for the user to be authenticated from among the plurality of registered biometric authentication information by executing the selected narrowing-down method, sets collation conditions according to the number of the narrowed-down registered biometric authentication information, and performs the biometric authentication using the set collation conditions and the narrowed-down registered biometric authentication information. An information processing apparatus configured as described above.

2. The information processing apparatus according to claim 1, wherein the arithmetic unit performs the biometric authentication based on a degree of collation, which is an index for evaluating the similarity between the biometric authentication information of the user and the registered biometric authentication information, and a threshold for collation, and sets a threshold according to the number of the narrowed-down registered biometric authentication information as the collation condition. An information processing apparatus configured as described above.

3. The information processing apparatus according to claim 2, wherein when the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information satisfies the predetermined criterion, the arithmetic unit does not execute the narrowing-down method, sets a predetermined threshold, and performs the biometric authentication using the set predetermined threshold and the plurality of registered biometric authentication information. An information processing apparatus configured as described above.

4. The information processing apparatus according to claim 2, wherein a plurality of the biometric authentication performance information corresponding to each of the number of the plurality of registered biometric authentication information, ​ ​ ​ ​ The narrowing-down setting information including the reference false acceptance identification rate, the upper limit false rejection identification rate, and the narrowing-down selection criterion information serving as a criterion for selecting the narrowing-down method, The narrowing-down method definition information including a plurality of the narrowing-down methods and information indicating characteristics of each narrowing-down method, It has a storage device in which they are stored, The arithmetic unit, acquires the upper limit false rejection identification rate, and determines whether or not the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information is equal to or less than the upper limit false rejection identification rate, thereby determining whether or not the false rejection identification rate satisfies the predetermined criterion, acquires the narrowing-down method definition information and the narrowing-down selection criterion information, selects the narrowing-down method from among the plurality of the narrowing-down methods included in the narrowing-down method definition information by a selection method according to the narrowing-down selection criterion information, configured as, an information processing apparatus.

5. In the information processing apparatus according to claim 4, the narrowing-down method definition information includes, for each of the plurality of the narrowing-down methods, information indicating characteristics of the narrowing-down method, including information indicating narrowing-down accuracy, user complexity, authentication time, and required data amount, the narrowing-down selection criterion information includes information indicating prioritizing a narrowing-down method with high narrowing-down accuracy, information indicating prioritizing a narrowing-down method with low user complexity, information indicating prioritizing a narrowing-down method with short authentication time, or information indicating prioritizing a narrowing-down method with small required data amount, or a combination of at least two of these, an information processing apparatus.

6. In the information processing apparatus according to claim 5, the narrowing-down selection criterion information is information indicating prioritizing a narrowing-down method with low user complexity, the arithmetic unit, prioritizes and selects the narrowing-down method with low user complexity from among the plurality of the narrowing-down methods, configured as, an information processing apparatus.

7. In the information processing apparatus according to claim 5, the narrowing-down selection criterion information is a combination of information indicating prioritizing a narrowing-down method with high narrowing-down accuracy and information indicating prioritizing a narrowing-down method with low user complexity, the arithmetic unit, selects, as a first priority, the narrowing-down method with excellent narrowing-down accuracy from among the plurality of the narrowing-down methods, and as a second priority, the narrowing-down method with low user complexity, configured as, an information processing apparatus.

8. In the information processing apparatus according to claim 1, the biometric authentication is face authentication, an information processing apparatus.

9. In the information processing apparatus according to claim 2, the arithmetic unit sets, as a threshold value corresponding to the number of the registered biometric authentication information after narrowing down, a threshold value such that the performance of the biometric authentication corresponding to the number of the registered biometric authentication information after narrowing down becomes the reference false acceptance discrimination rate, is configured as an information processing apparatus.

10. In the information processing apparatus according to claim 3, the arithmetic unit sets, as the predetermined threshold value, a threshold value such that the performance of the biometric authentication corresponding to the number of the registered biometric authentication information becomes the reference false acceptance discrimination rate, is configured as an information processing apparatus.

11. In the information processing apparatus according to claim 4, the arithmetic unit when the biometric authentication is executed, if the number of candidates for the registered user identified as the user to be authenticated is 0, the narrowing-down method is selected again from the plurality of narrowing-down methods, and by executing the selected narrowing-down method, the registered biometric authentication information corresponding to the candidate for the user to be authenticated is narrowed down from among the plurality of registered biometric authentication information, a threshold value corresponding to the number of the registered biometric authentication information after narrowing down is set, and the biometric authentication is executed again using the set threshold value and the registered biometric authentication information after narrowing down, if the number of candidates for the registered user identified as the user to be authenticated is 2 or more, a threshold value corresponding to the number of candidates for the registered user identified as the user to be authenticated is set, and the biometric authentication is executed again using the set threshold value and the registered biometric authentication information corresponding to the candidate for the user to be authenticated identified as the user to be authenticated, is configured as an information processing apparatus.

12. In the information processing apparatus according to claim 11, the arithmetic unit if the number of candidates for the registered user identified as the user to be authenticated is 2 or more, sets, as a threshold value corresponding to the number of candidates for the registered user identified as the user to be authenticated, a threshold value such that the performance of the biometric authentication to be executed again becomes a predetermined false rejection discrimination rate, is configured as an information processing apparatus.

13. In the information processing apparatus according to claim 4, the arithmetic unit if the number of candidates for the registered user identified as the user to be authenticated is 0, determines whether the narrowing-down method has been executed before the current time, If the narrowing-down method has been executed before the current time, select and execute the narrowing-down method that has a better narrowing-down accuracy than the executed narrowing-down method. By executing the selected narrowing-down method, narrow down the registered biometric authentication information corresponding to the candidate of the user to be authenticated from among the plurality of registered biometric authentication information, set a threshold according to the number of the narrowed-down registered biometric authentication information, and use the set threshold and the narrowed-down registered biometric authentication information to execute the biometric authentication again. If the narrowing-down method has not been executed before the current time, select the narrowing-down method from among the plurality of narrowing-down methods by a selection method according to the narrowing-down selection criterion information, and by executing the selected narrowing-down method, narrow down the registered biometric authentication information corresponding to the candidate of the user to be authenticated from among the plurality of registered biometric authentication information, set a threshold according to the number of the narrowed-down registered biometric authentication information, and use the set threshold and the narrowed-down registered biometric authentication information to execute the biometric authentication again. When the number of candidates of the registered users identified as the user to be authenticated is two or more, set a threshold according to the number of candidates of the registered users identified as the user to be authenticated, and use the set threshold and the registered biometric authentication information corresponding to the candidates of the registered users identified as the user to be authenticated to execute the biometric authentication again. configured as information processing device.

14. A biometric authentication device including an arithmetic unit that acquires biometric authentication information of a user to be authenticated, collates the acquired biometric authentication information of the user to be authenticated with the registered biometric authentication information of a plurality of registered users, and executes biometric authentication to identify the user to be authenticated from among the plurality of registered users. The arithmetic unit acquires the biometric authentication information of the user to be authenticated, the number of the registered biometric authentication information, and biometric authentication performance information indicating the relationship between the false acceptance identification rate and the false rejection identification rate of the biometric authentication corresponding to the number of the registered biometric authentication information. and when the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information does not satisfy a predetermined standard, collate the authentication information related to the biometrics or knowledge of the user to be authenticated with the registered authentication information of the plurality of registered users, and select one or more of the plurality of narrowing-down methods for identifying candidates of the user to be authenticated from among the plurality of registered users. Obtain the authentication information of the user to be authenticated according to the selected narrowing-down method, and by executing the selected narrowing-down method, narrow down the registered biometric authentication information corresponding to the candidates of the user to be authenticated from among the plurality of registered biometric authentication information, Set a collation condition according to the number of the narrowed-down registered biometric authentication information, and execute the biometric authentication using the set collation condition and the narrowed-down registered biometric authentication information, A biometric authentication device configured as described above.

15. A terminal that obtains biometric authentication information of a user to be authenticated, Existence outside or inside the terminal, obtains biometric authentication information of the user to be authenticated from the terminal, collates the obtained biometric authentication information of the user to be authenticated with the registered biometric authentication information of a plurality of registered users, and executes biometric authentication for identifying the user to be authenticated from among the plurality of registered users. An information processing device, A biometric authentication system having: The information processing device: Obtains the biometric authentication information of the user to be authenticated, the number of the registered biometric authentication information, and biometric authentication performance information indicating the relationship between the false acceptance identification rate and the false rejection identification rate of the biometric authentication corresponding to the number of the registered biometric authentication information, And obtains, When the false rejection identification rate corresponding to the reference false acceptance identification rate of the biometric authentication performance information corresponding to the number of the registered biometric authentication information does not satisfy a predetermined standard, Collate the authentication information related to the biometrics or knowledge of the user to be authenticated with the registered authentication information of the plurality of registered users, and select one or more of the plurality of narrowing-down methods for identifying the candidates of the user to be authenticated from among the plurality of registered users, Obtain the authentication information of the user to be authenticated according to the selected narrowing-down method, and by executing the selected narrowing-down method, narrow down the registered biometric authentication information corresponding to the candidates of the user to be authenticated from among the plurality of registered biometric authentication information, Set a collation condition according to the number of the narrowed-down registered biometric authentication information, and execute the biometric authentication using the set collation condition and the narrowed-down registered biometric authentication information, Configured as follows, A biometric authentication system.

Images