JPH11110544A - Method and device for certifying ic card - Google Patents

Abstract

PROBLEM TO BE SOLVED: To stabilize the rate of collation by measuring a degree of difference found from the difference of respective elements between registered signature data and collation signature data, determining the possibility of collation based on that measured degree of difference and updating the registered signature data, with data for which weighted mean is operated to the permitted collation signature data and registered signature data. SOLUTION: A signature is written on an electronic tablet 11 with an electronic pen 10, and these signature data are sent through a tablet I/F 21 to a work area 30 and temporarily stored as collation signature data. Then, signature data provided from that new signature are collated with the registered signature data read from an IC card and the collated result is displayed on a display device 14. At this time, whether the collated result is successful or not, the possibility of certification is determined based on a cumulative value as the difference degree of an error between the registered signature data and the collation signature data and the registered signature data are updated with the data, for which the weighted mean is calculated to the permitted collation signature data and registered signature data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method and an apparatus for authenticating an IC card owner. More specifically,
The present invention relates to a method and an apparatus for securing a stable collation rate in a method of performing IC card owner authentication using an electronic signature.

[0002]

2. Description of the Related Art There are various methods for personal authentication.
In Japan, personal identification using a seal is still important. In Japan, seals are the mainstream for personal authentication at public institutions. However, in recent years, personal authentication other than seals has been used.

As computer communications have become more popular and more popular, security issues have surfaced. One type of security is personal authentication. Passwords are the most dominant, but this is not the only way to protect confidentiality. Therefore, biometric personal authentication has recently attracted attention as a method for identifying individuals.

[0004] The biometric is the iris of the eye, DN
It refers to biological characteristics such as A, voiceprint, and fingerprint, including handwriting. The biometric personal authentication has the advantage that it is not lost like a seal and is hard to be stolen by others. However, as a practical matter,
In cardholder authentication, methods other than handwriting are currently of little technical and economic utility.

In the case of cards such as bank cards and credit cards, personal authentication is important because it is necessary to authenticate the correct owner. However, the current use of cards places emphasis on confirmation of reliability. In the case of personal authentication, it has been postponed.

[0006] The terminal that can use the card is connected to the host computer of the financial institution via a network, and the host computer checks whether the balance of the deposit, whether the card has been lost or not, and the creditworthiness is checked by the host computer. The result is notified to the terminal. Based on the result,
Transactions such as cash payment and lending are performed.

At present, magnetic cards are the mainstream, and personal authentication is often performed using a password. I instead of a magnetic card
There has been a proposal to use a C card and more rigorously check for unauthorized use of the card by illegal copying or signature.

[0008]

However, there is an advantage that handwriting is difficult to be imitated by others. Moreover, since the handwritten signature depends on the characteristics of the individual, there is an advantage that the individual is hard to forget. On the other hand, however, there is a drawback in that if one wishes to evaluate handwriting with the eyes of a person, considerable experience is required, and that it depends on the eyesight of the person. As we have seen in the prior art, in Japan, personal authentication using a handwritten signature performed when using a credit card is formal,
It is far from personal authentication by signature.

[0009] For the above reasons, mechanization is indispensable in order to perform a stable personal authentication using a handwritten signature without depending on an appraiser. Personal authentication using a handwritten signature has characteristics that it is difficult for others to imitate because personal characteristics appear. However, at the same time, signatures have the characteristic that their handwriting differs depending on the mental state at that time and the environment of the place. When mechanized, the matching result must be stabilized in consideration of this characteristic. Another thing to consider in handwritten signatures is that handwriting changes over time.

An object of the present invention is to develop a method for maintaining a stable verification rate in mechanized handwritten signature verification.

[0011]

In order to solve the above-mentioned problems, the present invention introduces an IC card and an electronic signature, and allows a user to sign on an electronic tablet with an electronic pen, thereby obtaining an IC card owner. Is determined. The method is described below. Hereinafter, the signature data serving as the reference for verification is referred to as “registered signature data”, and the signature data to be verified is referred to as “verification signature data”.

In the authentication of the IC card on which the registered signature data is recorded, a difference obtained from a difference between each element of the registered signature data stored in the IC card and the verification signature data is measured. A determination is made as to whether or not authentication is possible, and the registered signature data is updated with data obtained by averaging the authorized verification signature data and the registered signature data.

Although various types of signature data can be used, the accuracy of authentication can be improved by using a method including time-series data of character handwriting coordinates and writing pressure information. In the following description, description will be made based on a method including time-series data of handwriting coordinates and pen pressure information, but it goes without saying that the present invention is not limited to this method.

[0014]

Embodiments of the present invention will be described. If the user owns the IC card, the signature data serving as the basis for performing the signature verification as an initial state is stored in the IC card.
Register on the card. The method is performed as follows. First, a user who becomes a card holder makes a plurality of signatures on an electronic tablet with an electronic pen, and creates a plurality of signature data. Individual signatures required for authentication are extracted from the signature data to create registration signature data, which is registered in the IC card. The registration signature data created here includes, in addition to the shape (x, y coordinates of the handwriting), writing pressure information (writing pressure p coordinate) as time-series data.

On the other hand, when the user uses the IC card, in order to confirm whether or not the IC card is owned by the user, the signature data registered in the IC card and a new signature on the spot are used. Verify the data on a computer.

If the verification result indicates that the verification is successful, the signer is
Recognized as C card owner, IC card transaction is performed. In the prior art, the IC card is returned to the user assuming that all the processing is completed at this stage. However, in the present invention, taking into account future signature verification,
The registration signature data is updated by performing a weighted average of the verification signature data used for the verification and the registration signature data.

For example, when the registered signature data A and the verification signature data B are used, an appropriate weighting factor w is introduced, and (1-w) A + wB Here, 0 ≦ w ≦ 0.5 is calculated, and the result is calculated. The new registration signature data A is assumed.
Here, the updated A is written in the IC card to be new registration signature data. This is the learning function.

As a hardware configuration for realizing the above-described method, an IC card verification device is prepared in which an IC card reader / writer and software embodying the method of the present invention are incorporated in the IC card handling side. A communication interface (I / F) is attached to the IC card collation device, and a user who has passed the collation is checked for credit by a credit company or other financial institution via a computer network. Only after passing this check will the final transaction take place.

An embodiment of the present invention will be described with reference to the hardware configuration shown in FIG.
However, here, the IC card owner is a customer, the person performing authentication is a clerk of an IC card dealer, the issuer of the IC card is a credit company, and the customer pays for shopping with a credit card (IC card). .

First, the clerk causes the IC card 13 submitted by the customer to be read by the IC card input / output device 12 (IC card reader / writer). The information read from the IC card is I
It is stored in the work area 30 through the C card I / F 22. At the same time, a “signature request” is displayed as a message on the display device 14 through the display I / F 23. According to this message, the clerk requests the customer to sign. The customer signs on the electronic tablet 11 with the electronic pen 10.

The signature data is used as verification signature data,
The information is sent to the work area 30 via the tablet I / F 21 and is temporarily stored. The verification program 28 verifies the registered signature data read from the IC card and the signature data obtained from the new signature, and displays the verification result on the display device 14. The pass / fail of the verification result is passed if the accumulated value (cumulative error) of the error between the registered signature data and the verification signature data is equal to or less than a predetermined value, that is, the user is authenticated.

If the above verification is successful, the customer information is transmitted to the credit company through the communication port 24, and an inquiry about the reliability of the customer is made. As a result, if the answer is "no problem", the transaction is established.

When the transaction is finally completed with the IC card, the registration / update program 29 is activated to create new registration signature data by reflecting the verification signature data in the registration signature data, and to the IC card reader / writer 12. The data is transferred and written to the IC card 13.

FIG. 2 is a flowchart summarizing the above processing flow. In the flowchart, Sn indicates internal processing, U1 indicates a signature by a user (customer), and Dn indicates determination (n is a serial number). If the collation result between the registered signature data A and the collation signature data B is unsuccessful, the transaction is not established by the IC card (process S3), and if the collation result is passed, the credit company is inquired of reliability by communication (process S4). If the answer is "with reliability", the transaction is performed by the IC card (processing S6), and if the answer is "with doubt of reliability", the transaction is not established (processing S5). In the processing S6 when the transaction is completed, the amount at the time of the transaction is transmitted to the credit company by communication. The processes S3 and S5 when the transaction is not established basically do nothing. Of course, this is displayed on the display as a message to inform the clerk.

The above is an outline of the present invention. How to handle signature data will be described in detail. First, a method for registering registration signature data will be described.

[0026] Registered signature data to customer (IC card owner)
In the case of registration, the customer first performs a plurality of signatures to create registration signature data. From the signature data, some are combined and averaged to create a registered signature data candidate. The registered signature data candidate is compared with each signature,
If all the signatures have a matching result equal to or higher than a certain matching rate, the registered signature data candidate is registered in the IC card as formal registered signature data.

At this time, the registration signature data is sent to the IC card read / writer 12 by the registration / update program 29 shown in FIG. In addition, creation of registration signature data at the time of initial registration, electronic tablet
The signature data written by the electronic pen 10 on 11 is processed by the collation program 28. In the case of a person whose signature is stable at the time of registration, the signature registration process usually ends at this stage.

If at least one signature has a matching rate or less that satisfies the acceptance criterion, the signature is further requested m times. The signature data at this stage is (n +
There are m) pieces of signature data. First, the signature data is extracted from one of them, and is compared with other signatures. This process is performed on all the signatures, a plurality of signatures with the best matching result (more precisely, a plurality of signatures in order from the one with the highest matching rate) are selected, and these are averaged to create a registered signature data candidate.

Further, the registered signature data candidate is collated with other signature data to select signature data that has been successfully collated, and the registered signature data candidate and the selected signature data are put together into an average pattern (average value conversion). ) To create registration signature data and register it in the IC card.

In the present invention, the registration signature data includes position coordinates (information representing a shape, (x, y)) and pen pressure coordinates (pen pressure information, p).
Are taken as time-series data, so that assuming that the registered signature data A and the verification signature data B are: A = ｛x _t , y _t , _pt ｝ _{t = 1} ^NB = ｛x _t ′, y _t ′, _pt '｝ _{T = 1} ^N (N is the total number of sample points), and the accumulated error D (dissimilarity D) is D ² = | AB | ² = Σ (| x _t −x _t ′ | ² + | y _t -y _t the sum of ²⁾ and a (Σ is ^{t = 1~N) | '| 2} + | p t -p t'.

However, the size and inclination of handwritten characters and the spacing between characters are not always stable and constant.
Even if the character size and inclination are normalized for each signature data, the coordinate values of A and B cannot be simply made to correspond as in the above equation. Therefore, as an example of the present invention, [Equation 1]
[Equation 2] is given. [Equation 1] is the degree of difference from the character of the verification signature data B corresponding to the m-th character of the registered signature data A.

[0032]

(Equation 1)

[Equation 2] is the difference in the entire signature, that is, the normalized difference obtained by dividing the sum of the differences of each character by the total number of data points of each character.

[0034]

(Equation 2)

Where L _s (m) and L _e (m) are the start and end points of the m-th character, M is the number of characters, w _p is a weighting factor, τ
(K) is the distortion function at the k-th sample point of the m-th character, and z ^* and p ^* are the normalized position coordinate value and pen pressure coordinate value, respectively. However, w _p is a coefficient that satisfies 0 ≦ w _p ≦ 1, and the influence of the pen pressure information on the authentication determination can be adjusted by this coefficient.

Z ^* is a complex coordinate, that is, z ^* = x ^* + iy ^* i = SQRT (-1) (square root of -1).

The distortion function τ is a function that minimizes the accumulated error of the handwritten signature data, and the position coordinate value and the writing pressure coordinate value of the signature data B corrected by this function are z.
^* _B (τ (k)) and p ^* _B (τ (k)). In other words, the coordinate point {z ^* _A (k), p ^* _A (k)} of the signature data A
署名 z ^* _B (τ (k)), p ^{* of} the signature data B corresponding to
By choosing _B (τ (k))｝, the cumulative error can be minimized.

The number of characters can be obtained from pen pressure information.
An example is shown in FIG. As shown in the figure, characters are separated by setting the point where the pen pressure P becomes 0 as a delimiter. However, when the interval between the stroke (2) and the stroke (3) is short as in the case of the character (2), these two strokes are regarded as one character. Start of the thus coordinate values obtained by separating the character (including brush pressure component) is is L _s, the end point is L _e.

If the cumulative error D obtained as described above is equal to or smaller than the threshold value, the verification is passed, and otherwise, the verification is rejected.

As can be seen from the above example, the registered signature data in the initial state is created with a limited number of signature data. Certainly, the registered signature data shows the individual's personal characteristics, but it is not enough for the number of samples. To compensate for this drawback, the learning function of the present invention, that is, feedback of the verification signature data used for verification to the registered signature data. On the other hand, in the long run,
As a human characteristic, the signature itself changes over time.
This means that fixed registration signature data cannot cope with long-term changes in signatures. This is also the reason that the verification signature data is fed back to the registered signature data.

In the feedback method, a weighting factor w is introduced to obtain (1-w) A + wB, where 0 ≦ w ≦ 0.5, and this calculation result is used as new registration signature data.
This update only happens when the match passes,
The registered signature data does not move in an extremely bad direction. Here, the bad direction means that the cumulative error frequency distribution of the verification signature data based on the registered signature data is degraded in a direction of spreading.

FIG. 4 shows the cumulative error frequency distribution.
The accumulated error of the verification signature data B to be fed back to the registered signature data is equal to or less than the threshold value and is 0.5 (50
%) The influence of B is suppressed by the following weight coefficient w. For example, when reflecting with 25% weight, w =
Since it is 0.25, new registration signature data A 'is obtained as A' = 0.75A + 0.25B. If A ′ is replaced with the new registered signature data A, the influence of B on A is as small as 0.25 and B is within the threshold value of the original A.
(= A ′) does not extremely change with B. In addition, not only can the value representing the signature characteristic of the user be gently approached, but also the signature characteristic that changes over time can be reflected.

As described above, the learning function of the present invention has a function of improving the initial registration signature data in the short term, and correcting an individual signature characteristic that changes with time in the long term. As a result, a stable matching rate can be always maintained.

[0044]

The advantages of personal authentication (authentication of IC card owner) using a handwritten signature are that the signature is difficult to imitate to another person, and that it is stolen by another person like a seal or password because it depends on the characteristics of the individual. It is rarely carried around, and it is unnecessary to carry around. However, handwriting changes slowly over time. Big changes are immediately noticeable, while small changes are hard to notice. What passed the signature verification at first may suddenly have a bad verification rate from a certain point. It depends on the characteristics of human beings, which are changing every day.

Another consideration when using a handwritten signature is the stability of the registered signature data itself. As described in the embodiment of the present invention, the registered signature data is created based on several signatures for a person who can make a stable signature, and ten or more signatures for an unstable person. Humans have different handwriting depending on their environment and mental state. The registered signature data created in the initial state is under a certain situation, and is likely to be biased signature data. To solve this problem, signature data created under various circumstances must be reflected in the registered signature data.

The learning function of the present invention solves the above problem. The method for realizing the learning function of the present invention is as follows.
Since the collation signature data when the IC card is actually used is used as a basis, the signed data under different environments and mental states is reflected in the registered signature data. Moreover, since only the signature data that has already been verified against the registered signature data and authenticated is the subject of update, the signature data of another person can be replaced with the registered signature data of the other person without dynamically changing the registered signature data. Absent. Therefore, the learning function of the present invention is useful for satisfying the necessary conditions for obtaining a stable matching rate. That is, in the short term, it is the reinforcement of the initial signature registration signature data,
In the long term, it absorbs changes in signature data resulting from human characteristics that change over time, which helps to secure a stable collation rate.

[Brief description of the drawings]

FIG. 1 is a diagram showing a hardware configuration required when performing IC card owner authentication by handwritten character collation according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a digital signature matching process including a learning function according to the embodiment of the present invention.

FIG. 3 is a diagram for explaining a method of separating character information from pen pressure information in the embodiment of the present invention.

FIG. 4 is a cumulative error frequency distribution diagram for explaining conditions of verification signature data used for updating registration signature data in the embodiment of the present invention.

[Explanation of symbols]

Reference Signs List 10 electronic pen 11 electronic tablet 12 IC card reader / writer (IC card input / output device) 13 IC card 14 display device 20 handwriting character recognition device 21 tablet I / F (I / F: interface) 22 IC card I / F 23 Display I / F 24 Communication port 25 CPU 26 Memory group 27 OS 28 Verification Program 29 Registration / update program (Registration program) 30 Work area

Claims (6)

[Claims] 1. In authentication of an IC card on which registered signature data is recorded, (1) measuring a difference obtained from a difference between each element of registered signature data stored on the IC card and verification signature data; An IC card authentication method comprising: determining whether or not authentication is possible based on the measured dissimilarity, and updating the registered signature data with data obtained by averaging the permitted verification signature data and the registered signature data. 2. The IC card authentication method according to claim 1, wherein said signature data includes time series data of character handwriting coordinates and writing pressure information. 3. A means for measuring a degree of difference obtained from a difference between each element of registered signature data and verification signature data stored in an IC card in authentication of an IC card on which registered signature data is recorded, 2) An IC card comprising means for performing authentication determination based on the measured dissimilarity and updating the registered signature data with data obtained by performing a weighted average of the authorized verification signature data and the registered signature data. Authentication device. 4. The IC card authentication device according to claim 3, wherein said signature data includes time series data of character handwriting coordinates and writing pressure information. 5. In authentication of an IC card on which registered signature data is recorded, (1) measuring a degree of difference obtained from a difference between each element of registered signature data stored on the IC card and verification signature data; A) a medium for recording a program for updating registration signature data with a weighted average of authorized verification signature data and registration signature data by making a determination as to whether authentication is possible based on the measured degree of difference. 6. The medium according to claim 5, wherein said signature data includes time series data of character handwriting coordinates and writing pressure information.