JP2014106842A - Collation device and collation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a collation device and a collation method capable of more preventing misrecognition.SOLUTION: A first determination part 17 is configured to, each time a collation score is calculated, compare the collation score with a first threshold, and to, when the collation score exceeds the first threshold, determine that registration data corresponding to the collation score are coincident with the collation data. A second determination part 18 is configured to, when the collation score exceeding the first threshold does not exist, compare the maximum collation score among the calculated collation scores with a second threshold, and to, when the maximum collation score exceeds the second threshold, determine that registration data corresponding to the maximum collation score are coincident with the collation data. Thus, it is possible to more prevent misrecognition than the case of using one threshold.

Description

  The present invention relates to a collation apparatus and a collation method for collating registered data registered in advance with inputted collation data.

  In recent years, in areas where personal authentication is required, such as entry / exit management for computer rooms and critical machine rooms, and access control for computer terminals and financial terminals, fingerprint authentication and face cards have been used instead of personal identification numbers and ID cards. Biometric authentication such as authentication is being adopted. In a collation apparatus employing such biometrics authentication, face and fingerprint image data (hereinafter referred to as “registration data”) registered in advance and a face or fingerprint image input from a camera or sensor, etc. Identity verification is performed by collating data (hereinafter referred to as “collation data”) (see, for example, Patent Documents 1 and 2).

  In such a collation device, the registered data is compared with the collation data, the degree of similarity is digitized, the numeric value is compared with a threshold value, and if the numeric value exceeds the threshold value, the person is identified. doing. For example, in the case of Patent Documents 1 and 2, collation data is subjected to two-dimensional discrete Fourier transform (hereinafter referred to as “DFT”) to create collation Fourier image data, and the same processing as the collation Fourier image data is performed. The registered Fourier image data of the registered data created by applying is synthesized, and after performing amplitude suppression processing on the synthesized Fourier image data, DFT is performed, and appears in the synthesized Fourier image data subjected to DFT The upper n pixels having higher correlation component intensity are extracted from the correlation component area, and the average of the correlation component intensities of the extracted n pixels is compared with a threshold value as a correlation value.

Japanese Patent Application Laid-Open No. 09-022406 JP-A-10-124667 JP 2001-266154 A

  However, in the collation apparatus as described above, only one threshold value is set, and if the numerical value indicating the degree of similarity such as a correlation value exceeds the one threshold value, the person is recognized as the person. As a result, even if someone else was mistaken, it was often misidentified that the person was identified.

  Then, this invention aims at providing the collation apparatus and collation method which can prevent misidentification more.

  In order to solve the above-described problems, a collation apparatus according to the present invention sequentially includes a first storage unit that stores a plurality of registration data in advance, an input unit that receives collation data, collation data, and each registration data. A calculation unit for calculating the similarity by comparison, a first threshold relating to the similarity, and a second storage unit for storing a second threshold smaller than the first threshold When the similarity is calculated by the calculation unit, the similarity is compared with the first threshold value. When the similarity exceeds the first threshold value, the first threshold value is exceeded. A first determination unit that determines that the registered data corresponding to the similarity and the collation data match, and if there is no similarity exceeding the first threshold, all the similarities calculated by the calculation unit The maximum similarity is compared with the second threshold, and the maximum similarity is Exceeds threshold, it is characterized in that the maximum and registration data corresponding to the similarity with the collation data and a second determination unit determines that match.

  In the verification device, the second threshold value may be an average value of all similarities calculated by the calculation unit.

  In the above collation device, the second determination unit may make the collation successful when each of the maximum similarities exceeds the second threshold value when there are a plurality of similarities.

  In the verification device, the first storage unit stores a plurality of registration data of the same person, and the second storage unit further includes a third threshold value that is smaller than the second threshold value. The second determination unit stores the maximum similarity and the second threshold value among all the similarities calculated by the calculation unit when there is no similarity exceeding the first threshold. And when the similarity exceeding the first threshold does not exist, a ranking list creation unit that creates a ranking list in which all the similarities calculated by the calculation unit are arranged in descending order; When the determination unit determines that the maximum similarity exceeds the second threshold, a predetermined number of similarities of the same person as the maximum similarity exist within a predetermined rank in the rank list, and The similarities of the same person are compared with the third threshold value, and the similarities are compared with the third threshold. By weight, may further comprise a third judging section judges that the registered data corresponding to the maximum degree of similarity with the collation data match.

  The collation method according to the present invention calculates the similarity by sequentially comparing the first storage step for storing a plurality of registration data in advance, the input step for inputting the collation data, and the collation data and each registration data. A calculation step, a second threshold value for storing a first threshold value related to similarity, and a second threshold value that is smaller than the first threshold value, and the calculation step for similarity. The similarity is compared with the first threshold every time the value is calculated, and if the similarity exceeds the first threshold, it is determined that the registration data corresponding to the similarity and the matching data match. If there is no similarity that exceeds the first threshold value and the first determination step, the maximum similarity among all the similarities calculated in the calculation step is compared with the second threshold value. This maximum similarity is the second threshold Exceeds the value, is characterized in that it has a second determination step of determining that the registered data and the verification data match corresponding to this maximum similarity.

  According to the present invention, each time the similarity is calculated, the similarity is compared with the first threshold value. When the similarity exceeds the first threshold value, the first threshold value is set. If it is determined that the registration data corresponding to the degree of similarity exceeding the matching data matches the collation data, and there is no similarity exceeding the first threshold value, the maximum similarity and the second level of the calculated similarities are determined. When the maximum similarity exceeds the second threshold value by comparing with the threshold value, it is determined that the registration data corresponding to the maximum similarity matches the collation data, thereby obtaining one threshold value. It is possible to prevent misunderstanding more than when using.

FIG. 1 is a block diagram showing the configuration of the collation apparatus according to the first embodiment of the present invention. FIG. 2 is a flowchart showing a collation operation by the collation apparatus according to the first embodiment of the present invention. FIG. 3 is a diagram illustrating a configuration example of the matching score list. FIG. 4 is a graph showing the relationship between the matching score and the first and second threshold values. FIG. 5 is a block diagram showing a configuration of a collation apparatus according to the second embodiment of the present invention. FIG. 6 is a flowchart showing a collation operation by the collation apparatus according to the second embodiment of the present invention. FIG. 7 is a diagram showing an example of a rank list in the collation device according to the second embodiment of the present invention.

<First Embodiment>
Hereinafter, a first embodiment of the present invention will be described in detail with reference to the drawings.

[Configuration of verification device]
As shown in FIG. 1, the collation apparatus 1 according to the present embodiment includes an input unit 11, a collation data storage unit 12, a registration data storage unit 13, a collation score calculation unit 14, a collation score storage unit 15, A threshold storage unit 16, a first determination unit 17, a second determination unit 18, and an output unit 19 are provided. Such a collation apparatus 1 performs personal collation using biometric authentication such as fingerprint authentication, face authentication, and vein authentication.

  The input unit 11 is a functional unit to which image data (verification data) for biometric authentication of a person who performs personal authentication is input from the outside. This collation data includes, for example, a fingerprint, a face, a vein, and the like, and can be acquired by a camera or a sensor.

  The collation data storage unit 12 is a functional unit that stores the collation data input to the input unit 11.

  The registration data storage unit 13 is a functional unit that stores image data (registration data) for biometric authentication of a plurality of persons to be collated. In the present embodiment, the registration data storage unit 13 stores reference numbers and registration data in association with each other. The image data for biometric authentication includes, for example, a fingerprint, a face, a vein, and the like, and is acquired in advance by a camera, a sensor, or the like. The registration data storage unit 13 may be associated with personal information for identifying an individual, such as an employee number, together with the reference number and registration data. Thereby, for example, even when a plurality of registration data of the same person are registered, it becomes possible to identify the registration data of the same person.

  The collation score calculation unit 14 sequentially compares the collation data stored in the collation data storage unit 12 and the registration data stored in the registration data storage unit 13 and indicates a collation score indicating the degree of similarity between the two. Is a functional unit for calculating Specifically, the matching score calculation unit 14 searches for a feature area of the matching data by the phase-only correlation method, compares the feature area with the corresponding area of the registered data, and determines the degree of similarity between the matching areas. (For example, refer to Patent Document 3). In the present embodiment, the collation score indicates that the higher the numerical value, the more similar the two are.

  The matching score storage unit 15 is a functional unit that stores the matching score calculated by the matching score calculation unit 14. For example, as shown in FIG. 3, the collation score storage unit 15 stores a collation score list in which a serial number of a series of registered data and a collation score between the registration data and the collation data are associated with each other. .

  The threshold value storage unit 16 is a functional unit that stores a first threshold value used for processing of the first determination unit 17 and a second threshold value used for processing of the second determination unit 18. . Each of the first threshold value and the second threshold value is a numerical value to be compared with the matching score. Here, the first threshold value is set to a value that can hardly be reached by the person himself / herself, such as a value with a probability of misperception of about 0.01%. On the other hand, the second threshold value is set to a value that is sufficiently high that the person is likely to be the person, such as a value that is lower than the first threshold value, for example, a probability of misperception being about 0.1%. .

  The first determination unit 17 determines that registered data that matches the verification data is based on the verification score calculated by the verification score calculation unit 14 and the first threshold value stored in the threshold value storage unit 16. It is a functional unit that determines whether or not it exists. Specifically, each time the collation score is calculated by the collation score calculation unit 14, the collation score is compared with the first threshold value stored in the threshold value storage unit 16. When the threshold is exceeded, it is determined that the registered data corresponding to the matching score exceeding the first threshold matches the matching data.

  Based on the collation score calculated by the collation score calculation unit 14 and the second threshold value stored in the threshold value storage unit 16, the second determination unit 18 has registered data that matches the collation data. It is a functional unit that determines whether or not it exists. Specifically, when there is no similarity exceeding the first threshold, the maximum matching score among the matching scores calculated by the matching score calculation unit 14 is compared with the second threshold. When the maximum collation score exceeds the second threshold value, it is determined that the registration data corresponding to the maximum collation score matches the collation data.

  The output unit 19 is a functional unit that outputs the determination results of the first determination unit 17 and the second determination unit 18 to the outside.

  Such a collation apparatus 1 includes an arithmetic unit such as a CPU, a storage device such as a memory and an HDD (Hard Disc Drive), and an input that detects input of information from the outside such as a keyboard, a mouse, a pointing device, a button, and a touch panel. A device, an I / F device that transmits and receives various information via a communication line such as the Internet, a LAN (Local Area Network), and a WAN (Wide Area Network), and a display device such as an LCD (Liquid Crystal Display) It consists of a computer and a program installed on this computer. That is, by the cooperation of the hardware device and software, the above hardware resources are controlled by the program, and the above-described input unit 11, registration data storage unit 13, calculation unit 14, collation score storage unit 15, first The determination unit 17, the threshold storage unit 16, and the output unit 19 are realized. Note that the program may be provided in a state of being recorded on a recording medium such as a CD-ROM, a DVD-ROM, or a memory card.

[Verification operation by verification device]
Next, with reference to FIG. 2, the collation operation by the collation apparatus 1 according to the present embodiment will be described.

  First, the input unit 11 receives collation data from the outside (step S1). For example, in the case of entrance / exit management, the input unit 11 receives collation data of a person entering the room from a fingerprint sensor, a camera, a vein sensor, or the like provided at the entrance. The received verification data is stored in the verification data storage unit 12.

When the collation data is stored, the collation score calculation unit 14 sets the value of the reference number n for reading the registration data as 1 (step S2).
The matching score storage unit 15 clears the stored matching score list (step S3). As a result, a new collation score list can be stored in the collation score storage unit 15.

  Next, the collation score calculation unit 14 calls the registration data to which the nth reference number is assigned from the registration data storage unit 13 (step S4), and uses the registration data and the collation data stored in the collation data storage unit 12. The comparison score is calculated by comparison (step S5). The calculated verification score is sent to the first determination unit 17.

  When the collation score is calculated, the first determination unit 17 calls the first threshold value from the threshold value storage unit 16 and the collation calculated by the first threshold value and the collation score calculation unit 14. The score is compared (step S6).

When the collation score based on the nth registration data is larger than the first threshold (step S6: YES), the first determination unit 17 determines that the nth registration data and the collation data match, A determination result to be verified is output (step S13). This determination result is output from the output unit 19 to the outside.
Here, a high value is set for the first threshold value so that only the person in question is allowed to collate. Therefore, when the collation score exceeding the first threshold is calculated, it is certain that the person corresponding to the collation data is the person corresponding to the registration data from which the collation score is derived. 1 determines that the collation is OK, and the collation operation is terminated at this point. Thereby, since it is not necessary to compare all the collation scores with the first threshold value, the collation operation is quickly performed.

  On the other hand, when the n-th collation score is smaller than the first threshold (step S6: NO), the first determination unit 17 stores the collation score together with the reference number in the collation score storage unit 15 (step S6). S7). As a result, as shown in FIG. 3, a collation score list in which the reference number n of the registration data is associated with the collation score is stored in the collation score storage unit 15.

  When the collation score is stored, the first determination unit 17 confirms whether or not the determined registration data of the reference number n is the last registration data stored in the registration data storage unit 13 (step S8). .

  If it is not the last registered data (step S8: NO), the first determination unit 17 returns to the process of step S4. Thereby, when there is no collation score exceeding the first threshold value, collation scores are calculated for all the registered data, and a collation score list as shown in FIG. 3 is generated.

  On the other hand, in the case of the last registered data (step S8: YES), the second determination unit 18 extracts the maximum collation score from the collation score list stored in the collation score storage unit 15 (step S10). Subsequently, the second threshold value is extracted from the threshold value storage unit 16, and it is determined whether or not the maximum matching score exceeds the second threshold value (step S11).

When the maximum collation score exceeds the second threshold (step S11: YES), the second determination unit 18 determines that the registration data corresponding to the maximum collation score matches the collation data, A collation result for collation OK is output (step S13). This determination result is output to the outside by the output unit 19.
If the matching score is a sufficiently high value even if it does not exceed the first threshold value, the person of the registered data corresponding to the matching score often corresponds to the person of the matching data. Therefore, in the present embodiment, when the second threshold value lower than the first threshold value is set and the maximum matching score exceeds the second threshold value, the maximum matching score is set. It is determined that the person of the registered data corresponding to is equivalent to the person of the collation data, and the collation is OK.

For example, in the collation score list as shown in FIG. 3, when the first threshold value is set to “2” and the second threshold value is set to “1”, each collation score is assigned to the first and second collation scores. The relationship with the threshold value is as shown in FIG. That is, there is no collation score exceeding the first threshold, and the largest collation score exceeding the second threshold is the collation score having the reference number of 15. In such a case, it is determined that the person corresponding to the registration data from which the collation score of the reference number 15 is derived corresponds to the person of the collation data and is collated OK.
In FIG. 4, the horizontal axis indicates the reference number, and the vertical axis indicates the matching score.

  On the other hand, when the maximum collation score does not exceed the second threshold value (step S11: NO), the second determination unit 17 determines that there is no registration data corresponding to the collation data and determines that collation is NG ( Step S12). This determination result is output to the outside by the output unit 19.

  As described above, according to the present embodiment, each time the matching score is calculated, the matching score is compared with the first threshold value, and when the matching score exceeds the first threshold value, If it is determined that the registration data corresponding to the matching score exceeding the first threshold matches the matching data, and there is no matching score exceeding the first threshold, When the maximum matching score is compared with the second threshold value and the maximum matching score exceeds the second threshold value, the registered data corresponding to the matching score exceeding the second threshold value and By determining that the matching data matches, it is possible to prevent misperception more than when using one threshold value.

  In this embodiment, the case where the determination is made based on the first and second threshold values set in advance has been described as an example. However, the threshold value used for the determination is the first and second threshold values. Without being limited, various values may be used.

For example, you may make it use the average value of a collation score. In this case, the second threshold value in the present embodiment may be the average value of the matching scores. In addition, a matching score that is a predetermined value or more than the average value may be determined as matching OK. Usually, since the collation score of the person is higher than the collation score of another person, the possibility of being a collation NG can be reduced even if the person is the person due to the conditions of the registered image such as the lighting environment.
Moreover, you may make it use the standard deviation of a collation score list. In this case, a verification score that is out of the standard deviation may be set as verification OK.

  Further, for example, when there are a plurality of maximum matching scores such as a plurality of matching data of the same person, if these matching scores exceed the second threshold value, it is determined that the matching is successful. Also good. Thereby, for example, even when a plurality of images of the person are registered, the misidentification rate can be reduced. In this case, it goes without saying that the present invention can also be applied when an average value is used as the second threshold value.

<Second Embodiment>
Next, a second embodiment of the present invention will be described. The present embodiment is different from the first embodiment described above in the processing after the second determination unit 18, that is, the processing in the case where none of the collation scores exceeds the first threshold value. Is. Therefore, in the present embodiment, the same configuration and operation as those of the first embodiment are denoted by the same names and reference numerals, and description thereof will be omitted as appropriate. In addition, configurations and operations that are partially different from those of the first embodiment will be described with “′” added to the reference numerals.

  As shown in FIG. 5, the collation device 1 ′ according to the present embodiment includes an input unit 11, a collation data storage unit 12, a registration data storage unit 13, a collation score calculation unit 14, and a collation score storage unit 15. , A threshold storage unit 16 ′, a first determination unit 17, a second determination unit 18 ′, an output unit 19 ′, a rank list creation unit 21, and a third determination unit 22. Yes.

  The threshold value storage unit 16 ′ includes a first threshold value used for processing by the first determination unit 17, a second threshold value used for processing by the second determination unit 18, and a third threshold value. It is a functional unit that stores a third threshold value used for the processing of the determination unit 22. Each of the first threshold value and the second threshold value is a numerical value to be compared with the matching score. Here, the first threshold value is set to a value that can hardly be reached by the person himself / herself, such as a value with a probability of misperception of about 0.01%. The second threshold value is lower than the first threshold value, and is set to a value that is sufficiently high that the person is likely to be the person, such as a value having a probability of misperception of about 0.1%. The third threshold value is set to a value that is lower than the second threshold value, but is sufficiently high to be the person himself / herself.

  The ranking list creation unit 21 is a functional unit that creates a ranking list in which registered data is arranged in the order of the matching score values stored in the matching score storage unit 15. In the present embodiment, the registration data is arranged in the order of the collation score value. In addition, the registration data is given an identification number assigned with the same number to the same person together with the reference number.

  The second determination unit 18 ′ is a registration that matches the verification data based on the verification score calculated by the verification score calculation unit 14 and the second threshold value stored in the threshold value storage unit 16 ′. It is a functional unit that determines whether or not data exists. Specifically, when there is no similarity exceeding the first threshold value, the highest collation score among the collation scores calculated by the collation score calculation unit 14, that is, created by the rank list creation unit 21. The first collation score in the ranking list is compared with the second threshold value, and when the first collation score exceeds the second threshold value, the process of the third determination unit 22 is performed. . When the first matching score is equal to or lower than the second threshold, the second determination unit 18 'determines that the matching is NG.

  When the first matching score exceeds the second threshold, the third determining unit 22 is based on the other matching score and the second threshold stored in the threshold storage unit 16 ′. , A functional unit that determines whether there is registered data that matches the verification data. Specifically, when there are two or more matching scores of the same person in the ranking list with the first matching score, the comparison score of the same person is compared with the third threshold value. When the collation score exceeds the third threshold value, it is determined that the registration data corresponding to the first collation score matches the collation data.

[Verification operation by verification device]
Next, with reference to FIG. 6, the collation operation | movement by the collation apparatus 2 which concerns on this Embodiment is demonstrated.

  First, after the processing of steps S1 to S9 in the first embodiment described with reference to FIG. 2 is performed, the registration data of the serial number n determined by the first determination unit 17 is the registration data storage unit 13. In the case of the last registered data stored in (Step S8: YES), the rank list creating unit 21 creates a rank list. As a result, a ranking list as shown in FIG. 7 is created. In this ranking list, the registration data is arranged in the order of the matching score values. Here, the registration data is given a reference number and an identification number assigned the same number to the same person. For example, the registration data “15-003” ranked first is data of the same person as the registration data with the identification number “003” because the reference number is “15” and the identification number is “003”.

  When the ranking list is generated, the second determination unit 18 ′ has the highest matching score among the matching scores calculated by the matching score calculation unit 14, that is, 1 in the ranking list created by the ranking list creation unit 21. A rank matching score is extracted (step S10). Subsequently, the second threshold value is extracted from the threshold value storage unit 16 ', and it is determined whether or not the first matching score exceeds the second threshold value (step S11).

  When the first-order collation score does not exceed the second threshold (step S11: NO), the second determination unit 17 determines that there is no registration data corresponding to the collation data and determines that the collation is NG (step). S12). The determination result is output to the outside by the output unit 19 '.

  On the other hand, when the first matching score exceeds the second threshold (step S11: YES), the third determination unit 22 determines that the matching score of the same person as the first matching score is 2 in the ranking list. It is confirmed whether there are two or more in the fifth to fifth positions (step S22).

  Even when the maximum matching score is not sufficiently high to exceed the first threshold, registered data corresponding to the matching score often corresponds to a person in the matching data. Therefore, in this embodiment, when a plurality of pieces of registration data of the same person are registered, the maximum collation score and another collation score of the same person have a certain value, that is, a value equal to or greater than the third threshold value. If so, it is determined that the person of the registered data corresponding to the maximum collation score corresponds to the person of the collation data, and the collation is OK. Thereby, misidentification can be prevented more than the case of using one threshold value.

  For example, in the case of the ranking list shown in FIG. 7, the matching score of the same person as the first-ranked registration data “15-003” is “5-003” in the second place and “14-003” in the fourth place. The collation score of the same person as the first collation score is two or more in the second to fifth places in the rank list.

  In the case where two or more matching scores of the same person as the first matching score do not exist in the second to fifth positions (step S22: NO), the third determination unit 22 assumes that there is no registered data corresponding to the matching data It determines with collation NG (step S12). The determination result is output to the outside by the output unit 19 '.

  On the other hand, when there are two or more matching scores of the same person as the first matching score in 2nd to 5th (step S22: YE), the third determination unit 22 receives the threshold value from the threshold storage unit 16 ′. A threshold value of 3 is extracted, and it is determined whether or not the matching score of the same person as the first matching score exceeds the third threshold value (step S23).

  If the collation score of the same person as the first collation score does not exceed the third threshold value (step S23: NO), the third determination unit 22 collates because there is no registered data corresponding to the collation data. It is determined as NG (step S12). The determination result is output to the outside by the output unit 19 '.

  On the other hand, when the collation score of the same person as the first collation score exceeds the third threshold (step S23: YES), the third determination unit 22 registers the registration data corresponding to the first collation score. And the collation data match, and the collation result is output as collation OK (step S13). The determination result is output to the outside by the output unit 19 '.

  For example, in the case of the ranking list shown in FIG. 7, if the third threshold value is set to “0.8”, the collation score of “5-003” in the second place is “1.2”. The collation score of “14-003” in the fourth place is “0.9”, which exceeds the third threshold value, so that it is determined that collation is OK.

Thus, according to the present embodiment, when it is determined that the first matching score exceeds the second threshold, it is the same as the first matching score in the second to fifth places in the ranking list. If there are two or more person matching scores and these matching scores exceed the third threshold, it is determined that the person in the registered data corresponding to the first similarity matches the person in the matching data. Thus, misperception can be prevented more than when one threshold value is used.
In addition, since it is determined whether or not the first collation score is collation, it is possible to improve collation accuracy.

In the present embodiment, the case where the collation is OK when there are two or more collation scores of the same person as the first collation score in the second to fifth ranks has been described as an example. The order range and the quantity included in this range can be set freely as appropriate.
Needless to say, the present embodiment can also be applied to a case where a plurality of first-order verification scores exist.
Further, the second threshold value may be an average value of all the matching scores calculated by the matching score calculation unit 15.

Moreover, although this Embodiment demonstrated as an example the case where it determines whether two or more collation scores of the same person as the 1st collation score exist in 2nd-5th in a ranking list, The order range and the number of matching scores of the same person included in this range are not limited to this, and can be set as appropriate.
For example, if there are five or more matching scores of the same person as the first matching score within the second to tenth positions in the ranking list and these matching scores are equal to or higher than the third threshold, It may be determined that the person in the registered data corresponding to the similarity matches the person in the collation data. At this time, the third threshold value may be “0.34”.
Also, for example, there are three or more matching scores of the same person as the first matching score within the 2nd to 10th places in the ranking list, and one of these is a threshold value of 0, 38 or more, and 2 If the threshold value is 0.32 or more, it may be determined that the person in the registered data corresponding to the first similarity matches the person in the collation data.

In the first and second embodiments, the case where the number of threshold values is two in the first embodiment and three in the second embodiment is described as an example. The quantity is not limited to this, and can be set freely as appropriate.
Also, the value of each threshold can be set freely as appropriate.

  The present invention can be applied to various verification devices that perform identity verification.

  DESCRIPTION OF SYMBOLS 1, 2 ... Collation apparatus, 11 ... Input part, 12 ... Collation data memory | storage part, 13 ... Registration data memory | storage part, 14 ... Collation score calculation part, 15 ... Collation score memory | storage part, 16, 16 '... Threshold memory | storage part , 17 ... 1st determination part, 18, 18 '... 2nd determination part, 19, 19' ... Output part, 21 ... Ranking list creation part, 22 ... 3rd determination part.

Claims (5)

A first storage unit storing a plurality of registration data in advance;
An input unit for input of collation data;
A calculation unit that sequentially compares the verification data with each of the registered data to calculate a similarity;
A second storage unit that stores a first threshold value related to the similarity and a second threshold value that is smaller than the first threshold value;
Each time the similarity is calculated by the calculation unit, the similarity is compared with the first threshold value. When the similarity exceeds the first threshold value, the first threshold value is calculated. A first determination unit that determines that the registration data corresponding to the similarity exceeding the reference and the verification data are verified;
When the similarity exceeding the first threshold does not exist, the maximum similarity among all the similarities calculated by the calculation unit is compared with the second threshold, and this A second determination unit that determines that the registration data corresponding to the maximum similarity and the matching data match when the maximum similarity exceeds the second threshold; Verification device. The verification device according to claim 1,
The second threshold value includes an average value of all the similarities calculated by the calculation unit. In the collation device according to claim 1 or 2,
When there are a plurality of the maximum similarities, the second determination unit determines that the verification is successful if they exceed a second threshold value. In the collation device according to any one of claims 1 to 3,
The first storage unit stores a plurality of registration data of the same person,
The second storage unit further stores a third threshold value that is smaller than the second threshold value,
The second determination unit, when there is no similarity that exceeds the first threshold, out of all the similarities calculated by the calculation unit and the second similarity Compare with threshold,
A ranking list creating unit that creates a ranking list in which all the similarities calculated by the calculating unit are arranged in a descending order when the similarity exceeding the first threshold does not exist;
When the second determination unit determines that the maximum similarity exceeds the second threshold, the maximum similarity and the similarity of the same person within a predetermined rank in the rank list Are present in a predetermined quantity, and each similarity degree of the same person is compared with the third threshold value. When each similarity degree exceeds the third threshold value, the maximum similarity degree A verification apparatus further comprising: a third determination unit that determines that the registration data corresponding to and the verification data match. A first storage step for storing a plurality of registration data in advance;
An input step where the verification data is entered;
A calculation step of sequentially comparing the verification data and each of the registration data to calculate a similarity;
A second storage step of storing a first threshold value related to the similarity and a second threshold value smaller than the first threshold value;
Each time the similarity is calculated by the calculating step, the similarity is compared with the first threshold, and when the similarity exceeds the first threshold, the first threshold A first determination step of determining that registration data corresponding to a degree of similarity exceeding the verification data matches the verification data;
If the similarity exceeding the first threshold does not exist, the maximum similarity among all the similarities calculated by the calculating step is compared with the second threshold, and this And a second determination step of determining that the registered data corresponding to the maximum similarity and the verification data match when the maximum similarity exceeds the second threshold value. Method.

Images