JP2000099720A - Personal authentication method and personal authentication device - Google Patents

Abstract

(57) [Summary] [Problem] To provide a personal authentication method and a personal authentication device which can register and collate even a user who is conventionally rejected. Upon registration, biometric data for creating dictionary data is input, dictionary data is created from the input biometric data, and an evaluation value representing the quality of the dictionary data is calculated. Then, the evaluation value of the input biometric data is calculated, the method of creating dictionary data is changed in accordance with the evaluation value, and the created dictionary data is registered and stored in the storage means as registration data together with the evaluation value. At the time of collation, biometric data for collation is input, registered data to be collated with the inputted biometric data is read out from the storage means, and dictionary data in the read registered data is collated with the inputted biometric data. At the time of this collation, the collation method is changed according to the evaluation value of the dictionary data in the registered data read from the storage means.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a personal authentication method and a personal authentication apparatus for performing personal authentication as to whether one is a person or another person using biometric data such as a fingerprint or a finger image of an individual to be authenticated. .

[0002]

2. Description of the Related Art There is no conventional personal authentication apparatus using biometric data such as fingerprints which specifically deals with individual differences of users (subjects) who do not have sufficient features. Users who do not have sufficient features such as those that are rejected at the time of registration and collation, and that the identity confirmation rate and the exclusion rate of others are reduced and the usability is deteriorated, but registration and collation are permitted. ing.

[0003]

In a conventional personal authentication device using biometric data such as a fingerprint, a user who cannot input biometric data having a good feature amount does not take individual differences into account. It is rejected at the time of collation, and it becomes impossible to use the personal authentication device.

[0004] There is also a personal authentication device that performs registration processing or collation processing of biometric data without rejecting a user who cannot input sufficiently good feature information. The result of personal authentication for such a user by the personal authentication device is as follows:
Because the registration data or the input data at the time of collation is not good, the personal identification rate and the exclusion rate of others fall, and the personal authentication device becomes inconvenient to use.

[0005] Further, since the user is not notified of the cause of the decrease in the personal identification rate, leaving the impression that the performance of the personal authentication device itself is low leads to a decrease in the reliability of the personal authentication device.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a personal authentication method and a personal authentication device which enable registration and collation even for a user who is conventionally rejected.

It is another object of the present invention to provide a personal authentication method and a personal authentication device that can deepen the understanding of the result of the verification at the time of verification and increase the reliability of the personal authentication device.

[0008]

According to the personal authentication method of the present invention, at the time of registration, a step of inputting biometric data of a person to be authenticated for creating dictionary data, and a step of inputting the biometric data for creating dictionary data. Creating dictionary data and calculating an evaluation value representing the quality of the created dictionary data; and, when creating the dictionary data, calculating an evaluation value of the inputted biometric data for creating dictionary data;
Changing the method of creating the dictionary data according to the calculated evaluation value of the biometric data, and registering and storing the created dictionary data together with the calculated evaluation value of the dictionary data as registration data in storage means. When,
Inputting the biometric data of the individual to be authenticated at the time of collation, reading the registered data to be collated with the inputted biometric data for collation from the storage means, Collating the dictionary data in and the inputted biometric data for collation,
When the dictionary data is compared with the biometric data for matching, a step of changing the matching method according to the evaluation value of the dictionary data in the registered data read from the storage means.

[0009] The personal authentication method of the present invention includes the steps of:
Inputting the biometric data of the individual to be authenticated for creating dictionary data, creating dictionary data from the inputted biometric data for creating dictionary data, and calculating an evaluation value representing the quality of the created dictionary data Calculating the dictionary data, and calculating an evaluation value of the input biometric data for creating the dictionary data, and changing a method of generating the dictionary data according to the calculated evaluation value of the biometric data. Registering and storing the created dictionary data in the storage unit as registration data together with the calculated evaluation value of the dictionary data, and inputting biometric data of a person to be authenticated for verification at the time of verification.
Reading from the storage means the registered data to be collated with the input biometric data for collation, and collating the dictionary data in the read registered data with the inputted biometric data for collation. When matching the dictionary data with the biometric data for matching, the number of matches is changed according to the evaluation value of the dictionary data in the registered data read from the storage means, and the matching result at each matching is determined by a predetermined condition. And determining that the dictionary data and the biometric data for matching are the same when the condition is satisfied.

[0010] In addition, the personal authentication method of the present invention, at the time of registration,
Inputting the biometric data of the individual to be authenticated for creating dictionary data, creating dictionary data from the inputted biometric data for creating dictionary data, and calculating an evaluation value representing the quality of the created dictionary data Calculating the dictionary data, and calculating an evaluation value of the input biometric data for creating the dictionary data, and changing a method of generating the dictionary data according to the calculated evaluation value of the biometric data. Registering and storing the created dictionary data in the storage unit as registration data together with the calculated evaluation value of the dictionary data, and inputting biometric data of a person to be authenticated for verification at the time of verification.
Reading the registered data to be compared with the inputted biometric data for collation from the storage means, collating dictionary data in the read registered data with the inputted biometric data for collation, Judging the identity between the dictionary data and the biometric data for collation by comparing the collation result with a predetermined threshold value; and reading from the storage means when collating the dictionary data with the biometric data for collation. Changing a threshold value for judging the identity between the dictionary data and the biometric data for collation based on the evaluation value of the dictionary data in the registered data thus obtained.

[0011] In addition, the personal authentication device of the present invention, at the time of registration,
First biometric data input means for inputting biometric data of a person to be authenticated for dictionary data creation, and a dictionary for creating dictionary data from the biometric data for dictionary data creation input by the first biometric data input means Data creation means, dictionary data evaluation means for calculating an evaluation value representing the quality of the dictionary data created by the dictionary data creation means, and biometric data for dictionary data creation input by the first biometric data input means Biological data evaluation means for calculating the evaluation value of, and a dictionary data creation method change means for changing the dictionary data creation method in the dictionary data creation means according to the evaluation value of the biological data calculated by this biological data evaluation means, The dictionary data created by the dictionary data creating means is used to evaluate the dictionary data calculated by the dictionary data evaluating means. A registration data storage means for registering and storing both as registration data, a second biometric data input means for inputting biometric data of an individual to be verified for verification, and a second biometric data input means for inputting the biometric data. Registration data reading means for reading registered data to be compared with biometric data for verification from the registered data storage means; dictionary data in the registered data read by the registered data reading means; and the second biometric data input means A collating unit for collating the biometric data for collation input by the collating unit, and when collating the dictionary data with the biometric data for collation by the collating unit, the dictionary data in the registration data read out from the registration data storage unit. A matching method changing means for changing the matching method according to the evaluation value.

[0012] Further, the personal authentication device of the present invention, at the time of registration,
First biometric data input means for inputting biometric data of a person to be authenticated for dictionary data creation, and a dictionary for creating dictionary data from the biometric data for dictionary data creation input by the first biometric data input means Data creation means, dictionary data evaluation means for calculating an evaluation value representing the quality of the dictionary data created by the dictionary data creation means, and biometric data for dictionary data creation input by the first biometric data input means Biological data evaluation means for calculating the evaluation value of, and a dictionary data creation method change means for changing the dictionary data creation method in the dictionary data creation means according to the evaluation value of the biological data calculated by this biological data evaluation means, The dictionary data created by the dictionary data creating means is used to evaluate the dictionary data calculated by the dictionary data evaluating means. A registration data storage means for registering and storing both as registration data, a second biometric data input means for inputting biometric data of an individual to be verified for verification, and a second biometric data input means for inputting the biometric data. Registration data reading means for reading registered data to be compared with biometric data for verification from the registered data storage means; dictionary data in the registered data read by the registered data reading means; and the second biometric data input means A collating unit for collating the biometric data for collation input by the collating unit, and when collating the dictionary data with the biometric data for collation by the collating unit, the dictionary data in the registration data read out from the registration data storage unit. The number of matchings is changed according to the evaluation value, and when the matching result at each matching satisfies a predetermined condition, the dictionary data and the biometric data for matching are used. And and means for determining that it is identical.

Further, the personal authentication apparatus of the present invention, at the time of registration, first biometric data input means for inputting biometric data of an individual to be authenticated for creating dictionary data, and input by the first biometric data input means. Dictionary data creation means for creating dictionary data from the created biometric data for creating dictionary data; dictionary data evaluation means for calculating an evaluation value representing the quality of the dictionary data created by the dictionary data creation means; A biometric data evaluation means for calculating an evaluation value of biometric data for creating dictionary data input by the biometric data input means, and the biometric data evaluation means for calculating the evaluation value of the biometric data calculated by the biometric data evaluation means. A dictionary data creation method change unit for changing a dictionary data creation method; and the dictionary data created by the dictionary data creation unit. Registration data storage means for registering and storing as dictionary data together with the evaluation value of the dictionary data calculated by the certificate data evaluation means, and second biometric data input means for inputting biometric data of a person to be authenticated for verification at the time of verification. Registration data reading means for reading, from the registration data storage means, registration data to be matched with the matching biometric data input by the second biometric data input means;
By comparing the dictionary data in the registered data read by the registered data reading means with the matching biometric data input by the second biometric data input means, and comparing the matching result with a predetermined threshold value A matching unit that determines the identity between the dictionary data and the biometric data for matching, and when the matching unit matches the dictionary data with the biometric data for matching, the registration data read out from the registration data storage unit. Threshold value changing means for changing a threshold value for judging the identity between the dictionary data and the biometric data for matching in the matching means based on the evaluation value of the dictionary data.

According to the present invention, not only the method of creating dictionary data is changed according to the evaluation value representing the quality of the biometric data inputted at the time of registration, but also the evaluation value representing the quality of the dictionary data is held as information of the registration data. In addition, by changing the collation method such as the number of times of collation or the processing method based on the evaluation value indicating the quality of the dictionary data, the authentication target having biometric data that would otherwise be rejected and cannot use the personal authentication device. It is possible to perform registration and collation even for a person.

Also, by presenting the evaluation value of the registered data of the person to be authenticated at the time of registration, it is possible to deepen the understanding of the collation result at the time of collation and increase the reliability of the personal authentication device.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration of a personal authentication device according to an embodiment of the present invention. The personal authentication device using biometric data according to the present embodiment is roughly divided into a biometric data input device (sensor unit) 1 as biometric data input means for inputting biometric data such as a fingerprint, and a biometric data input device. A host computer 2 that controls the apparatus 1 to perform communication with the user, registration processing, collation processing, and the like, and an input device 3 such as a keyboard for inputting user (authenticatee) information.
And an output device 4 for displaying a communication screen such as a liquid crystal screen or a collation result, for example.

Hereinafter, each part will be described in detail. The host computer 2 includes a CPU (Central Processing Unit) 21 for performing various controls, an interface unit 22 for notifying a user of an instruction at each stage, a result of registration and collation, and the like, A registration unit 23 for registering data, a collation unit 24 for collating registered biometric data with registered biometric data, and a registration data storage unit 25 as a storage unit for registering and storing the registration data. ing.

The user communicates with the apparatus via the interface section 22 of the host computer 2. This communication includes, for example, at the time of registration, at the time of inputting various types of user information (user information), at the timing of data input timing by the biometric data input device 1, at the time of verification, notification of a verification result, and the like. There is.

The personal authentication device according to the present embodiment includes, for example, a biometric data input device 1 for inputting biometric data used for personal authentication such as a fingerprint, and a registration and collation process separately from the biometric data input device 1. However, the device may have a CPU in the biometric data input device 1 and perform a registration process and a collation process including communication with a user.

The user inputs necessary user information in accordance with an instruction via the interface unit 22, and inputs biometric data using the biometric data input device 1. here,
The biometric data input device 1 may be, for example, a device for inputting biometric data (for example, a fingerprint or the like) by optical means, or a device for electrically measuring the electric capacity of a living body. Alternatively, the input device may use input means for inputting biometric data by other means.

The registration unit 23 includes an input biometric data buffer 231 for temporarily storing input biometric data for creating dictionary data, an input data evaluation unit 232 for evaluating the input biometric data, and an input data evaluation unit 232. An input data determination unit 233 for classifying the biometric data input from the evaluation result, a first data buffer (data buffer A) 234 for recording the input data sorted according to the evaluation result for creating a dictionary, and a second data buffer (Data buffer B) 235, dictionary data creating unit 23 that extracts dictionary data from extracted biometric data to create dictionary data
6. A dictionary data evaluation unit 237 that performs final evaluation of the created dictionary data, and a registration data creation unit 238 that creates registration data from the dictionary data, the evaluation values of the dictionary data, user information, and the like.

Here, the input biometric data buffer 231,
The first data buffer 234 and the second data buffer 235 are memory devices in the host computer 2 and temporarily record input biometric data.

The registration data created by the registration data creation unit 238 of the registration unit 23 is stored in the registration data storage unit 25. The registered data storage unit 25 stores registered data 1, 2,..., N for each user.

The collating unit 24 includes the input biometric data,
The collation with the registered data stored in the registered data storage unit 25 is performed, and the result of the comparison is presented to the user by the output device 4 via the interface unit 22.

The collation unit 24 includes an input biometric data buffer 24 for temporarily storing inputted biometric data for collation.
1. A registration data buffer 242 for temporarily storing arbitrary registration data called from the registration data storage unit 25; a dictionary data buffer 243 for temporarily storing dictionary data extracted from the registration data; Input data evaluation unit 24 for evaluating whether or not the data is valid
4. Input biometric data and dictionary data buffer 243
A data collating unit 245 that collates the dictionary data stored in the data collating unit, and determines whether the biometric data input from the collation result in the data collating unit 245 and the collated dictionary data belong to the same person. Matching result determination unit 246
It is constituted by.

Here, the input biometric data buffer 241,
The registration data buffer 242 and the dictionary data buffer 243 are memory devices in the host computer 2 as in the case of the registration unit 23, and temporarily record input biometric data.

The input biometric data buffer 231 of the registration unit 23 and the input biometric data buffer 241 of the matching unit 24
Need not be provided separately, and the same input data buffer may be shared, and the input data evaluation unit 232 of the registration unit 23 and the input data evaluation unit 244 of the collation unit 24 may be shared. Good.

The dictionary data buffer 2 of the collating unit 24
The reference numeral 43 may temporarily store one dictionary data, or may temporarily store a plurality of dictionary data when necessary.

In the present embodiment, each input data input by the user from the biometric data input device 1 is classified and defined as follows. Invalid data: input data that is not recognized as legitimate biometric data, such as a lack of necessary information on a part of the biometrics.

Valid data: input data recognized as valid biometric data. Good data: input data in which the feature amount included in the data is more than a predetermined value in the valid data.

Semi-good data: input data in which the feature amount included in the valid data is less than a predetermined value. That is, all input data by the user is classified into invalid data and valid data, and the valid data is further classified into good data and semi-good data.

Here, irrespective of the number of valid data or invalid data, the number of times of input of biometric data by the user during one registration process is n, the number of good data in valid data is p, and the number of half good data is Let the number be q.

Next, registration processing in the above configuration will be described with reference to the flowcharts shown in FIGS. Note that the processing described below is mainly performed by the CP
It is executed under the control of U21.

When the apparatus is started, a GUI (Graphical User Interface) screen is displayed on the output device 4 so that the user can select whether to perform registration processing or verification processing. Alternatively, operation buttons are displayed. Here, when performing the registration process, the user selects, for example, a corresponding menu or operation button.

When the registration process is selected, the interface unit 22 indicates that the user has selected the registration process.
Is notified to the CPU 21 via. The CPU 21 prompts the user to input user information (user information), and the user inputs the user information (S1). Here, the user information is information capable of uniquely specifying the registered data, for example, an ID number (unique identification information) prepared for each user or a user name such as a user name. It is something. In this case, the ID number prepared for each user is not input by the user, but an unused ID number.
The number may be automatically issued to save the user's trouble.

When the input of the user information is completed, the CPU 2
1 resets the number of inputs (n), the number of good data (p), and the number of semi-good data (q) of biometric data to “0”, respectively (S2).

Next, the CPU 21 checks the number (n) of times of input of biometric data by the user (S3), and if the number of inputs (n) is less than the predetermined number (Th1), proceeds to the next processing step. . In this embodiment, in order to reduce the burden on the user, the number of times of input of the biometric data by the user is checked here. The input of the biometric data by the user may be repeated until the required number of input data for the creation is obtained.

If it is determined in step S3 that the number of times of input of biometric data by the user is less than the predetermined number, the CPU
21 checks the number (p) of good input data already existing (S4), and if the number of good data (p) is less than the predetermined number (Th2), proceeds to the next processing step.

In step S4, if the number (p) of good data is less than the predetermined number (Th2), the CPU
21 prompts the user to input biometric data via the interface unit 22 (S5). When the input of the biometric data in step 5 is performed, the number of inputs (n) is incremented (S6).

The input biometric data is temporarily stored in the input biometric data buffer 231 of the registration unit 23, and is used by the input data evaluation unit 232 to determine whether it can be used for creating dictionary data. An evaluation value is calculated (S7). This evaluation value is, for example, the number of feature points included in the input data.

The input data determination section 233 converts the evaluation value calculated by the input data evaluation section 232 into a predetermined numerical value (Th
Compared with 3) (S8), if it is equal to or more than a predetermined numerical value (Th3), it is determined that the data is valid as biometric data and can be used for creating dictionary data. Then, the input data is stored in the first data buffer 234 (S15).

Here, when the evaluation value of the input data is equal to or smaller than the predetermined numerical value (Th3), the evaluation value is further compared with a predetermined numerical value (Th4) (S9), and the predetermined numerical value (Th3) is obtained.
4) If the data is above, it is determined that the data is valid as biometric data, but not semi-good data which is not sufficiently good for creating dictionary data (S11), and the input data is converted to the second data. It is registered in the buffer 235 (S12).

In this embodiment, the first data buffer 234 is a buffer for temporarily storing good data, and the second data buffer 235 is for temporarily storing semi-good data. Used as a buffer. The good data number (p) is incremented when the input data is stored in the first data buffer 234 (S16), and the semi-good data number (q) is incremented when the input data is stored in the second data buffer 235. (S13).

When the processing in step S13 or S16 is completed, the registration processing returns to the check on the number of inputs (n) in step S3. While the number of times of input (n) is less than Th1 and the number of good data (p) is less than Th2, the above-described processing in steps S5 to S16 is performed.

When the number of inputs (n) is less than Th1 and the number of good data (p) is greater than Th2,
The process proceeds to step S18. This is a case where the number of input times less than the predetermined number of times is valid as biometric data, and the number of good input data required for dictionary data creation has been collected.

In step S18, the dictionary data creation unit 2
At 36, dictionary data is created from good input data recorded in the first data buffer 234.
Here, as a method of creating the registered data in the dictionary data creating unit 236, for example, a feature point is extracted from each input data and extracted from the input data having the largest average value of similarity with the input data other than itself. The obtained feature amount data may be used as dictionary data, or another method may be used.

When the dictionary data creation unit 236 completes the creation of the dictionary data, the registration data creation unit 238
Dictionary data and user information, and dictionary data evaluation unit 2
The registration data is created from the evaluation value of the dictionary data calculated in 37, and registered in the registration data storage unit 25 (S19).

The evaluation value of the dictionary data calculated by the dictionary data evaluation unit 237 is, for example, the average value of the evaluation values of the biometric data used for the creation of the dictionary data. It may be a numerical value obtained by using the evaluation value of the data, or may be an evaluation result obtained by comparing a feature amount included in the dictionary data itself with a predetermined value.

FIG. 4 shows an example of the format of the registration data. The registration data stored in the registration data storage unit 25 is, for example, as shown in FIG. 4A, a user (user) identification code (for example, an ID number) capable of uniquely specifying a user. And user (user) information such as the name and gender of the user, dictionary information such as which finger is the fingerprint in the case of personal authentication using fingerprints, and registration date and time, , Dictionary data evaluation values, and dictionary data (feature data) used as a dictionary. Alternatively, as shown in FIG. 4B, the configuration may be such that the registration data of each individual has a plurality of dictionary data.

If the number of inputs (n) is equal to or greater than Th1 in step S3, the registration process proceeds to step S17. In step S17, when the number of times of input of biometric data (n) by the user becomes equal to or greater than Th1, the number of good data (p) that is valid as biometric data and is good input data is converted into dictionary data. It is checked whether there is a necessary number Th2 necessary to create the data. If the required number Th2 exists, the dictionary data creation unit 236 determines whether or not there is good biometric data recorded in the first data buffer 234. Is created (S18).

In step S18, the dictionary data creation unit 236
When the creation of the dictionary data is completed, the registration data creation unit 238 creates the registration data from the dictionary data and the user information, the evaluation value of the dictionary data calculated by the dictionary data evaluation unit 237, and the like. Registration in the storage unit 25 is performed (S19).

When the number of inputs (n) is equal to or greater than Th1 and the number of good data (p), which is the number of valid and valid input data as biometric data, is less than Th2, that is, the number of good input data is This is the case where the required number of dictionary data could not be collected within a predetermined number of input times.In the conventional personal authentication device using biometric data, registration of dictionary data was refused, and Will not authorize use with the personal authentication device, or
This is a case where the matching rate is significantly reduced only for the user even after registration.

In such a case, the personal authentication device using biometric data according to the present embodiment checks the number of good data (p), which is the number of valid and good input data as biometric data ( S20) If there is even one good data number, the biometric data of both the first data buffer 234 storing good data and the second data buffer 235 storing semi-good data are used. Then, the dictionary data creation unit 236 creates dictionary data (S21).

If no good data exists, the number (q) of semi-good data is checked (S22).
Number of semi-good data required to create dictionary data (Th5)
If the data exists, the dictionary data creation unit 236 creates dictionary data using the biometric data stored in the second data buffer 235 (S23).

Here, step S18 and step S21
The method of creating dictionary data in step S23 is different from that of step S23. For example, one dictionary data is created in step S18 where all the biometric data used for creating dictionary data is good data. In step S21 in which the biometric data to be used is composed of good data and semi-good data, for example, two dictionaries, one dictionary data created from good data and one dictionary data created from semi-good data In step S23 where the data is created and the biometric data used to create the dictionary data is all composed of semi-good data, for example, three dictionary data are created.

In addition to changing the number of dictionary data to be created, the feature extracted from the biometric data input as dictionary data may be changed in each case.

The number of good data (p) is not sufficient, and
If the number of semi-good data (q) is not sufficient, the registration process ends without creating and registering dictionary data.

In the registration processing according to the present embodiment,
The evaluation is performed on the input data every time the biometric data is input. However, the input data may be evaluated collectively after a predetermined number of data inputs are completed.

Next, a first example of the collation processing will be described with reference to the flowcharts shown in FIGS. Note that the processing described below is mainly executed under the control of the CPU 21.

When the apparatus is activated, a GUI screen is displayed on the output device 4 so that the user can select whether to perform a registration process or a collation process, and a menu or an operation button is displayed. . Here, when performing the matching process, the user selects, for example, a corresponding menu or operation button.

When the collation processing is selected, the interface unit 22 indicates that the collation processing is selected by the user.
Is notified to the CPU 21 via. The CPU 21 prompts the user to input user information (user information), and the user inputs the user information (S31). Here, the user information is information that can uniquely specify registered data used as a dictionary at the time of collation, and is, for example, an ID number prepared for each user.

When the input of the user information is completed, the input user information is correct depending on whether the registered data having the user information input by the user exists in the registered data storage unit 25 or not. It is determined whether or not it is (S32). If it is determined that correct user information has been input, the corresponding registration data is called from the registration data storage unit 25 (S33), and temporarily stored in the registration data buffer 242 provided in the collation unit 24 (S34).

When the registered data is stored in the registered data buffer 242, the CPU 21 extracts only the dictionary data from the registered data and temporarily stores it in the dictionary data buffer 243 (S35). The evaluation value (v1) of the dictionary data is extracted from the registration data stored in 242 (S36).

The evaluation value (v1) of the dictionary data extracted from the registered data is compared with a predetermined threshold value (Th7) (S37). If the evaluation value (v1) is equal to or greater than the threshold value Th7, the corresponding dictionary data is determined to be good. It is determined that the dictionary data is created using only the biometric data, and the collation processing proceeds to step S38. If the biometric data is less than the threshold Th7, the corresponding dictionary data is determined to be good biometric data and semi-good biometric data. It is determined that the data is created from the data or the data is created only from semi-good biometric data, and the collation processing proceeds to step S43.

When collation with dictionary data created only from good biometric data is performed, first, biometric data for collation is input (S 38) and temporarily stored in the input biometric data buffer 241. . First, the input data evaluation unit 244 evaluates whether the input biometric data has sufficient information as biometric data and whether it is valid data (S39). If the evaluation result is less than the predetermined threshold Th4, the input biometric data is determined to be invalid data, and the process returns to input of biometric data in step S38. If the evaluation result of the input data evaluation unit 244 is equal to or greater than the predetermined threshold Th4, the input biometric data is determined to be valid data, and the process proceeds to the collation processing in step S41.

Here, a threshold value T for determining whether the input biometric data is valid data or invalid data is set.
h4 is the same as the threshold used by the input data evaluation unit 232 at the time of registration to evaluate whether the input data is valid data.

The matching process between the biometric data stored in the input biometric data buffer 241 and the dictionary data stored in the dictionary data buffer 243 in step S41 is performed by the data matching unit 245. For example, the feature value extracted from the biometric data This is a process of calculating the similarity between the data and the dictionary data by correlation at each feature point.

When the collation processing is completed, the collation result determination unit 2
At 46, the collation result is determined (S42).
Here, the determination of the collation processing result is performed, for example, by comparing the numerical value of the collation result with a predetermined threshold, and when the value is equal to or greater than the predetermined threshold (OK), the input biometric data and the dictionary in the registration data are compared. The data is determined to be the same (S
53) If it is less than the predetermined threshold (in the case of NG), it is determined that both data are not the same (S5).
4), the collation processing ends.

As described above, when the evaluation value of the dictionary data stored in the registration data buffer 242 is less than the predetermined threshold value Th7 in step S37, the corresponding dictionary data is determined to be good biometric data and semi-good It is determined that the data is created from the biometric data or only from the semi-good biometric data, and the collation processing proceeds to step S43.

The CPU 21 determines in step S43
The number of inputs (n) and the number of times (ok) when the collation result is OK are initialized, and the current number of inputs (n) is checked (S44). If the number of inputs (n) is less than the predetermined threshold value Th8 in step S44, the user inputs biometric data according to the instruction output on the interface screen of the output device 4 (S45).

The input biometric data is stored in step S38.
, The input biometric data buffer 241
The number of inputs (n) is temporarily incremented (S46), and the input data evaluation unit 244 calculates an evaluation value for evaluating whether the data is valid biometric data (S47).

Next, in step S48, the evaluation result of the input data evaluation unit 244 is compared with a predetermined threshold value Th4, and if it is less than the threshold value Th4, it is determined that the input biometric data is invalid data. Then, the process returns to the input of biometric data in step S45. If the evaluation result of the input data evaluation unit 244 is equal to or greater than the predetermined threshold Th4, the input biometric data is determined to be valid data, and the process proceeds to the collation processing in step S49.

In the matching process in step S 49, the biometric data determined to be valid data by the data matching unit 245 is compared with the dictionary data stored in the dictionary data buffer 243. At this time, the same collation processing as in step S41 may be performed, or a different processing from the collation processing in step S41 may be performed.

The collation result in the data collation unit 245 is
The matching result determination unit 246 makes that determination (S5).
0), if it is determined that the input biometric data and the dictionary data are the same (if the collation result is OK), the number of OKs (ok) is incremented (S5).
1) Otherwise, the input of the biometric data, the evaluation of the input data, the collation processing, and the result determination processing are repeated until the number of inputs (n) becomes equal to or greater than the threshold Th8 without changing the number of OKs.

In step S44, if the number of biometric data inputs (n) by the user is equal to or greater than the threshold Th8, the process proceeds to step S52, a check process of the number of OKs (ok) up to that point. If the number of OKs (ok) is equal to or greater than the predetermined threshold Th9, there is a sufficient number of matches between the input data and the dictionary data even though the evaluation value of the dictionary data is less than the predetermined threshold (Th7). Therefore, it is determined that the finally input biometric data is the same as the dictionary data in the designated registration data (S53). Also, the number of OKs (ok) is equal to a predetermined threshold Th.
If it is less than 9, it is determined that the finally input biometric data and the dictionary data in the designated registration data are not the same (S54). Then, the collation processing ends.

That is, by increasing the number of times that the comparison result between good data and semi-good data or dictionary data created from only semi-good data is increased, the reliability is improved and a final verification result that can be believed is obtained. Can be.

Next, a second example of the collation processing will be described with reference to the flowcharts shown in FIGS. Note that the processing described below is mainly executed under the control of the CPU 21.

In the first example of the collation processing shown in FIGS. 5 and 6, the number of times of collation is changed according to the evaluation value of dictionary data in the registered data to be collated with the input biometric data. The second of the collation processing shown in FIGS.
In the example, the matching method, specifically, changing a threshold value for determining whether or not the input biometric data and the dictionary data are the same is changed.

When this device is started, a GUI screen is displayed on the output device 4 so that the user can select whether to perform a registration process or a collation process, and a menu or an operation button is displayed. . Here, when performing the matching process, the user selects, for example, a corresponding menu or operation button.

When the collation processing is selected, the interface section 22 indicates that the collation processing is selected by the user.
Is notified to the CPU 21 via. The CPU 21 prompts the user to input user information (user information), and the user inputs the user information (S61). Here, the user information is information that can uniquely specify registered data used as a dictionary at the time of collation, and is, for example, an ID number prepared for each user.

When the input of the user information is completed, the input user information is correct depending on whether or not the registered data having the user information input by the user exists in the registered data storage unit 25. It is determined whether or not (S62). If it is determined that correct user information has been input, the corresponding registration data is called from the registration data storage unit 25 (S63), and is temporarily stored in the registration data buffer 242 provided in the matching unit 24 (S64).

When the registered data is stored in the registered data buffer 242, the CPU 21 extracts only the dictionary data from the registered data and temporarily stores the dictionary data in the dictionary data buffer 243 (S65). The evaluation value (v1) of the dictionary data is extracted from the registration data stored in 242 (S66).

The evaluation value (v1) of the dictionary data extracted from the registered data is compared with a predetermined threshold value (Th7) (S37). If the evaluation value (v1) is equal to or greater than the threshold value Th7, the corresponding dictionary data is determined to be good. It is determined that the data is created using only the unique biometric data, and the threshold for determination at the time of matching is set to Th1.
If the value is set to 0 (S68) and is less than the threshold Th7, the corresponding dictionary data is created from good biometric data and semi-good biometric data, or is created from only semi-good biometric data. Then, the threshold for determination at the time of comparison is set to Th11 (S69).

Here, the threshold values Th10 and Th11 may be common numerical values for each user. The threshold 10 may use a common numerical value, and the threshold Th11 may use a predetermined numerical value for each user. in this case,
The value of the threshold value Th11 is obtained by comparing the biometric data input for creating dictionary data with the created dictionary data at the time of creating dictionary data in the registration process for each user, and comparing the input biometric data with the dictionary data. Are registered in advance as a part of the registration data as a threshold value that can be determined to be the same.

Alternatively, after the dictionary data is created at the time of the registration process for each user, the user is requested to input the biometric data again, and the biometric data input here is compared with the created dictionary data. May be performed, and a threshold value that can determine that the input biometric data and the dictionary data are the same may be obtained.

When the setting of the judgment threshold value at the time of collation is completed, the CPU 21 requests the user to input the biometric data, and the user inputs the biometric data (S 70). Is memorized. First, the input biometric data is input data evaluation unit 244.
In step S71, it is evaluated whether the information has sufficient information as biometric data and whether the data is valid data (S71). If the evaluation result of the input data evaluation unit 244 is less than the predetermined threshold Th4, It is determined that the input biometric data is invalid data, and the process returns to input of biometric data in step S70.

If the evaluation result of the input data evaluation section 244 is equal to or greater than the predetermined threshold Th4, the input biometric data is determined to be valid data, and the process proceeds to the collation processing in step S73. .

Here, a threshold value T for determining whether the input biometric data is valid data or invalid data.
h4 is the same as the threshold used by the input data evaluation unit 232 at the time of registration to evaluate whether the input data is valid data.

The matching process between the biometric data stored in the input biometric data buffer 241 and the dictionary data stored in the dictionary data buffer 243 in step S73 is performed by the data matching unit 245. For example, the feature value extracted from the biometric data This is a process of calculating the similarity between the data and the dictionary data by correlation at each feature point.

When the collation processing is completed, the collation result determination unit 2
At 46, the collation result is determined (S74).
Here, the determination of the collation processing result is performed, for example, by comparing the numerical value of the collation result with a predetermined threshold, and when the value is equal to or greater than the predetermined threshold (OK), the input biometric data and the dictionary in the registration data are compared. The data is determined to be the same (S
75) If it is less than the predetermined threshold (NG), it is determined that the two data are not the same (S7).
6), the collation processing ends.

As described above, according to the above-described embodiment, not only the method of creating dictionary data is changed according to the evaluation value representing the quality of biometric data input at the time of registration, but also the evaluation value representing the quality of dictionary data is changed. Retain as information of registration data,
Furthermore, by changing the collation method such as the number of times of collation or the processing method based on the evaluation value of the dictionary data, it is possible to register even users who have biometric data that would otherwise be rejected and cannot use the personal authentication device. And collation can be performed.

In the above-described embodiment, the number of times of input of biometric data by the user at the time of verification is restricted in consideration of the burden on the user. May be provided, and the input may be repeated until the required number of data is obtained.

In the registration processing shown in FIGS. 2 and 3 or the collation processing shown in FIGS. 5, 6, 7 and 8, steps S9, FIGS. In step S40, FIG. 7 and FIG. 8, the user repeats the input of the biometric data until valid biometric data is input in step S72. Here, the user is enabled within a predetermined number of times. If the biometric data cannot be input, the registration process or the collation process may be forcibly terminated.

Further, in the personal authentication device according to the present invention, the evaluation value representing the quality of the dictionary data calculated at the time of registration is output to the output device and presented to the user. Can be confirmed, and an understanding of the collation result at the time of collation can be obtained. Further, by presenting the personal identification rate not only based on the evaluation value of the dictionary data but also, for example, based on the relationship between the evaluation value of the dictionary data obtained empirically and the matching result, the user's understanding can be further obtained. . Specifically, for example, a user confirmation screen as shown in FIG. 9 is displayed on the output device 4 to confirm the user.

Further, a predetermined number of valid biometric data is input at the time of registration, and the registration method or both registration and collation are determined based on the composition of the quality of the biometric data (the number of good data and the number of semi-good data). By allowing the user to select a method, an understanding of the collation result at the time of collation may be obtained. In this case, the collation method selected by the user is stored as a part of the information of the registration data, and the collation is performed according to the stored collation method when the collation is performed by the user. Specifically, for example, a registration / collation method selection screen as shown in FIG.
, The user's selection is accepted.

[0096]

As described above in detail, according to the present invention, there is provided a personal authentication method and a personal authentication apparatus which can register and collate even a user who is conventionally rejected. it can.

Further, according to the present invention, it is possible to provide a personal authentication method and a personal authentication device that can deepen the understanding of the verification result at the time of verification and increase the reliability of the personal verification device.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a personal authentication device according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a registration process.

FIG. 3 is a flowchart illustrating a registration process.

FIG. 4 is a diagram showing a configuration example of registration data.

FIG. 5 is a flowchart illustrating a first example of a matching process.

FIG. 6 is a flowchart illustrating a first example of a matching process;

FIG. 7 is a flowchart for explaining a second example of the collation processing.

FIG. 8 is a flowchart for explaining a second example of the collation processing.

FIG. 9 is a view showing an example of a user confirmation screen at the time of registration.

FIG. 10 is a diagram showing an example of a registration / collation method selection screen at the time of registration.

[Explanation of symbols]

1 ... biological data input device, 2 ... host computer, 3 ... input device, 4 ... output device, 21 ... CP
U, 23 ... registration unit, 24 ... collation unit, 25 ... registered data storage unit, 231 ... input biometric data buffer, 23
2 ... input data evaluation section, 233 ... input data determination section, 234 ... first data buffer, 235 ... second data buffer, 236 ... dictionary data creation section, 237 ...
... Dictionary data evaluation unit, 238 ... Registered data creation unit, 2
41 ... input biometric data buffer, 242 ... registered data buffer, 243 ... dictionary data buffer, 244 ...
... input data evaluation section, 245 ... data collation section, 246
... Matching result determination unit.

Claims (10)

[Claims] At the time of registration, a step of inputting biometric data of an individual to be authenticated for creating dictionary data, creating dictionary data from the inputted biometric data for creating dictionary data, and creating the dictionary data Calculating an evaluation value representing the quality of the dictionary data; and, at the time of creating the dictionary data, calculating an evaluation value of the input biometric data for creating dictionary data, and according to the calculated evaluation value of the biometric data, Changing the data creation method; registering and storing the created dictionary data in the storage unit as registration data together with the calculated dictionary data evaluation value; Inputting biometric data; reading registered data to be collated with the inputted biometric data for collation from the storage means; Collating the dictionary data in the read registration data with the inputted biometric data for collation; and when collating the dictionary data with the biometric data for collation, the dictionary data in the registration data read from the storage means. Changing the collation method according to the evaluation value of the dictionary data of the personal authentication method. 2. At the time of registration, a step of inputting biometric data of an individual to be authenticated for creating dictionary data; creating dictionary data from the inputted biometric data for creating dictionary data; Calculating an evaluation value representing the quality of the dictionary data; and, at the time of creating the dictionary data, calculating an evaluation value of the input biometric data for creating dictionary data, and according to the calculated evaluation value of the biometric data, Changing the data creation method; registering and storing the created dictionary data in the storage unit as registration data together with the calculated dictionary data evaluation value; Inputting biometric data; reading registered data to be collated with the inputted biometric data for collation from the storage means; Collating the dictionary data in the read registration data with the inputted biometric data for collation; and when collating the dictionary data with the biometric data for collation, the dictionary data in the registration data read from the storage means. Changing the number of times of matching according to the evaluation value of the dictionary data, and determining that the dictionary data and the biometric data for matching are the same when the matching result at each matching satisfies a predetermined condition; A personal authentication method, comprising: 3. At the time of registration, a step of inputting biometric data of an individual to be authenticated for creating dictionary data; creating dictionary data from the inputted biometric data for creating dictionary data; Calculating an evaluation value representing the quality of the dictionary data; and, at the time of creating the dictionary data, calculating an evaluation value of the input biometric data for creating dictionary data, and according to the calculated evaluation value of the biometric data, Changing the data creation method; registering and storing the created dictionary data in the storage unit as registration data together with the calculated dictionary data evaluation value; Inputting biometric data; reading registered data to be collated with the inputted biometric data for collation from the storage means; The dictionary data in the read registration data is compared with the input biometric data for collation, and the matching result is compared with a predetermined threshold value to determine the identity between the dictionary data and the biometric data for collation. Judging, at the time of matching the dictionary data with the biometric data for matching, the identity of the dictionary data and the biometric data for matching is determined by the evaluation value of the dictionary data in the registration data read from the storage means. Changing a threshold value for determination; and a personal authentication method. 4. When creating the dictionary data, an evaluation value of the input biometric data for creating dictionary data is calculated, and the input evaluation data is used to convert the input biometric data for creating dictionary data into good data. , Semi-good data, and invalid data, and the method of creating the dictionary data is changed according to the configuration of good data, semi-good data, and invalid data of the number of biometric data required for creating dictionary data. The personal authentication method according to any one of claims 1 to 3, wherein 5. The method according to claim 1, wherein when the dictionary data is created, the calculated evaluation value of the dictionary data is presented to the person to be authenticated, and the person to be authenticated confirms the evaluation value. The personal authentication method described in one. 6. A method according to claim 1, wherein when the dictionary data is created, an identification ratio at the time of collation that can be inferred from the calculated evaluation value of the dictionary data is presented to the authenticated person, and the authenticated person is confirmed. The personal authentication method according to claim 1. 7. At the time of registration of the registration data, the input biometric data for creating dictionary data is evaluated, and the evaluation result is presented to a person to be authenticated. 4. The personal authentication method according to claim 1, wherein a collation method at the time of collation is determined based on the selection result. 8. A first biometric data input means for inputting biometric data of an individual to be authenticated for dictionary data creation at the time of registration, and a biometric data for dictionary data creation input by the first biometric data input means. Dictionary data creation means for creating dictionary data from the data; dictionary data evaluation means for calculating an evaluation value representing the quality of the dictionary data created by the dictionary data creation means; and input by the first biometric data input means. Biometric data evaluation means for calculating an evaluation value of biometric data for creating dictionary data, and a method of creating dictionary data in the dictionary data creation means is changed according to the evaluation value of the biometric data calculated by the biometric data evaluation means. A dictionary data creation method changing unit, and the dictionary data created by the dictionary data creation unit is calculated by the dictionary data evaluation unit. Registration data storage means for registering and storing the evaluation data of the dictionary data as registration data, second biometric data input means for inputting biometric data of an individual to be verified for verification, and second biometric data input means. Registration data reading means for reading, from the registration data storage means, registration data to be compared with the biometric data for verification input by the data input means; dictionary data in the registration data read by the registration data reading means; Collating means for collating the biometric data for collation inputted by the second biometric data input means; and when collating the dictionary data with the biometric data for collation by the collating means, the collation data read from the registered data storage means. Collation method changing means for changing the collation method according to the evaluation value of the dictionary data in the registered data; Place. 9. A first biometric data input means for inputting biometric data of an individual to be authenticated for dictionary data creation at the time of registration; and a biometric data for dictionary data creation input by the first biometric data input means. Dictionary data creation means for creating dictionary data from the data; dictionary data evaluation means for calculating an evaluation value representing the quality of the dictionary data created by the dictionary data creation means; and input by the first biometric data input means. Biometric data evaluation means for calculating an evaluation value of biometric data for creating dictionary data, and changing a method of creating dictionary data in the dictionary data creation means in accordance with the evaluation value of the biometric data calculated by the biometric data evaluation means. A dictionary data creation method changing unit, and the dictionary data created by the dictionary data creation unit is calculated by the dictionary data evaluation unit. Registration data storage means for registering and storing the evaluation data of the dictionary data as registration data, second biometric data input means for inputting biometric data of an individual to be verified for verification, Registration data reading means for reading, from the registration data storage means, registration data to be compared with the biometric data for verification input by the data input means; dictionary data in the registration data read by the registration data reading means; Collating means for collating the biometric data for collation input by the second biometric data input means; and when collating the dictionary data with the biometric data for collation by the collating means, the collation data read from the registered data storage means. The number of matchings is changed according to the evaluation value of the dictionary data in the registration data, and when the matching result at each matching satisfies a predetermined condition, the dictionary data is changed. Data and personal authentication apparatus characterized by the matching of the biometric data is provided means for determining that the same, the. 10. A first biometric data input means for inputting biometric data of an individual to be authenticated for dictionary data creation at the time of registration, and a biometric data for dictionary data creation input by the first biometric data input means. Dictionary data creation means for creating dictionary data from the data; dictionary data evaluation means for calculating an evaluation value representing the quality of the dictionary data created by the dictionary data creation means; and input by the first biometric data input means. Biometric data evaluation means for calculating an evaluation value of biometric data for creating dictionary data, and a method of creating dictionary data in the dictionary data creation means is changed according to the evaluation value of the biometric data calculated by the biometric data evaluation means. Dictionary data creation method changing means, and calculating the dictionary data created by the dictionary data creation means by the dictionary data evaluation means. Registration data storage means for registering and storing the evaluation data of the dictionary data as registration data, second biometric data input means for inputting biometric data of an individual to be verified for verification, and second biometric data input means. Registration data reading means for reading, from the registration data storage means, registration data to be compared with the biometric data for verification input by the data input means; dictionary data in the registration data read by the registration data reading means; Collation for collating the biometric data for collation input by the second biometric data input means, and comparing the collation result with a predetermined threshold value to determine the identity between the dictionary data and the biometric data for collation. Means for comparing the dictionary data with the biometric data for matching in the matching means, the dictionary in the registered data read from the registered data storage means. Personal authentication apparatus characterized by the evaluation value of the data anda threshold value changing means for changing a threshold value for determining the identity of the dictionary data and the verification of the biometric data in the collating means.