KR100787116B1 - Method of generating changeable fingerprint templates which is invariant to translation and rotation of an input finger image and verification system thereof - Google Patents

Abstract

The present invention relates to a method for generating a variable fingerprint template independent of movement and rotation of an input fingerprint image and a verification system using the same. An invariant value with little change in movement and rotation of a fingerprint image from image information around a feature point extracted from an input fingerprint image By using the extracted invariant value as the input of the change function, the change amount of the feature point is calculated, the feature point is changed by the change amount calculated based on the direction of the feature point, and the changed fingerprint template is generated. Personal authentication is possible.

The present invention authenticates an individual with the changed feature information by changing the feature information of the original fingerprint, so that the original fingerprint template cannot be restored even if the variable fingerprint template and the conversion method are known. Since the fingerprint information can be canceled and a new modified feature point can be generated, the problem of privacy infringement and protection of personal fingerprint information that can occur when the fingerprint information of the individual is stolen can be solved.

In particular, the present invention does not require alignment between input fingerprint images and uses a variable fingerprint template having the same format as the existing fingerprint template, so that personal authentication is possible even when applied to a fingerprint system using an existing matching method without changing the matching method. Do.

Fingerprint recognition, fingerprint image, fingerprint template

Description

METHODS OF GENERATING CHANGEABLE FINGERPRINT TEMPLATES WHICH IS INVARIANT TO TRANSLATION AND ROTATION OF AN INPUT FINGER IMAGE AND VERIFICATION SYSTEM THEREOF}

1 is a block diagram of a verification system using a variable fingerprint template generation method independent of movement and rotation of an input fingerprint image according to the present invention.

Figure 2 is a flow chart illustrating a variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image in accordance with the present invention.

3 is an embodiment showing a feature of the input fingerprint image, the fingerprint feature point extracted from the image and the modified variable fingerprint template according to the present invention.

Figure 4 is a graph showing a feature point change function according to the present invention.

5 is a graph schematically illustrating a change in position of a feature point according to the present invention.

6 is an embodiment showing a process of obtaining an invariant value with little change in movement and rotation of an input fingerprint image according to the present invention.

7 is a view illustrating a process of obtaining an invariant value having little change in movement and rotation of an input fingerprint image by using feature direction and neighbor direction information according to the present invention.

8 is a graph showing a probability density function for generating a change function according to the present invention.

<Description of Symbols for Major Parts of Drawings>

11: fingerprint sensor 12: fingerprint image processing unit

13: fingerprint feature extraction unit 14: change function storage unit

15: change function generator 16: fingerprint template change unit

17: fingerprint information storage unit 18: matching unit

The present invention relates to fingerprint recognition, and more particularly, it is not necessary to align the position between the input fingerprint images, and thus it is possible to change the fingerprint template regardless of the movement and rotation of the fingerprint image and to perform personal authentication using the changed fingerprint template. The present invention relates to a variable fingerprint template generation method independent of movement and rotation of an input fingerprint image and a verification system using the same.

Fingerprint recognition is a method of authenticating individuals using fingerprint information unique to each individual and is the most widely used biometric method because of its excellent stability and accuracy.

In addition, the biggest reason that such fingerprint information can be used as a means of personal authentication is because of the uniqueness (Uniqueness) that varies from person to person and the invariance (Permanence) does not change significantly over time.

However, the uniqueness and invariability of such fingerprint information may cause serious privacy infringement when the personal fingerprint information is stolen or stolen.

In fact, in the case of identity card or password, which is a traditional personal authentication method, it is possible to solve the problem by reissuing a new ID card or password during theft or theft, but in the case of fingerprint information, it is impossible to regenerate new fingerprint information.

In particular, with the recent activation of personal authentication methods using biometrics, personal fingerprint information can be shared with investigative agencies such as criminal investigations, banks or other commercial companies using the Internet, which can lead to theft of personal fingerprint information. There is a problem that can be.

On the other hand, as a method of converting fingerprint information used for personal authentication in order to cope with the theft or theft of the fingerprint information, "biological information conversion method and system for security and privacy enhancement process" according to US Pat. (System and Method for Distorting a Biometric for Transactions with Enhanced Security and Privacy).

The U. S. Patent No. 6836554 discloses a method of generating a modified fingerprint pattern by dividing an input fingerprint image into small rectangular blocks and irregularly mixing the blocks, and using a morphing transformation that distorts the input fingerprint image. Method of generating a modified fingerprint pattern, variable fingerprint template using the position of the fingerprint feature point as the input (definition range) of the curve using the quadratic or higher curve and the output (range) of the corresponding curve as the feature point position of the modified fingerprint Disclosed is a method of generating a.

However, the conventional method disclosed in the above-described US Patent No. 6836554 generates a modified feature point by using an absolute position of the input fingerprint feature point, so that the position of the input fingerprint image is not always constantly input, There is a problem that is difficult to apply to the actual system that the movement and rotation of the fingerprint image according to the position.

In addition, the conventional method disclosed in the above-described US Patent No. 6836554 requires alignment to match positions between the fingerprint images in order to generate the same variable fingerprint template from the fingerprint images obtained from the same person, and to solve this problem, before the change, Although a method of aligning an input fingerprint image using a center point and a triangular point is disclosed, it is often impossible to extract both a center point and a delta from an input fingerprint image.

Therefore, as described above, the conventional method which presupposes alignment between fingerprint images is difficult to apply in practical terms.

The present invention is to solve the conventional problems as described above, an object of the present invention is to extract an invariant value with little change in the movement and rotation of the fingerprint image from the image information around the feature point extracted from the input fingerprint image, Calculate the change amount of the feature point by inputting the invariant value change function, move and rotate the feature point by the change amount calculated based on the direction of the feature point to generate the changed fingerprint template, and perform personal authentication using the changed fingerprint template. It is to provide a variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image and a verification system using the same.

In order to achieve the object of the present invention as described above, the verification system using a variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the fingerprint for obtaining a fingerprint image from the user A sensor; A fingerprint image processing unit which processes the fingerprint image obtained by the fingerprint sensor; A fingerprint feature extraction unit for extracting a feature point of the fingerprint from the fingerprint image processed by the fingerprint image processing unit and extracting a constant value having little change in the movement and rotation of the input fingerprint image at each extracted feature point; A change function storage unit for storing a change function generated by using fingerprint information or by using a user ID; A change function generation unit for automatically generating a change function with a user ID or obtaining a change function previously generated and stored in the change function storage unit; A fingerprint template changer for obtaining a position shift and a direction change amount of a fingerprint feature point by using invariant values and a change function extracted from each feature point, and generating a changed fingerprint template by changing each feature point; A fingerprint information storage unit for storing the changed fingerprint template with the user's ID; And a matching unit that determines whether the user is present by matching fingerprint templates stored in the fingerprint information storage unit using the changed fingerprint feature points and the user ID.

In order to achieve the object of the present invention as described above, the variable fingerprint template generation method irrelevant to the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the image processing the image after processing the input fingerprint image Extracting feature points of the fingerprint from the; Extracting an invariant value with little change in movement and rotation of the fingerprint image at each feature point extracted through the above steps; A change function generation step of generating a change function using a user ID or fingerprint information of an individual; Obtaining the position and direction change of the feature point of the fingerprint by using the invariant value and the change function extracted from each feature point; And generating a variable fingerprint template by changing a feature point of the fingerprint based on the direction of the feature point using the change amount obtained in the step.

In the variable fingerprint template generation method irrelevant to the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the invariant value having little change in the movement and rotation of the input fingerprint image is the fingerprint feature point extracted in the fingerprint information extraction step. It is a value obtained from a fingerprint image in a certain region or a fingerprint image processed based on a reference value, and includes direction information of a fingerprint pattern around a feature point, image tilt information around a feature point, frequency information of a fingerprint pattern around a feature point, and The invariant value of each feature point is obtained by using statistical information of the fingerprint pattern around the feature point.

In the variable fingerprint template generation method irrespective of movement and rotation of the input fingerprint image according to an embodiment of the present invention, the change function is a method of automatically generating or changing function storage according to a user ID to obtain a new change if necessary. Create a function or obtain a new change function from the change function storage.

In the variable fingerprint template generation method irrelevant to the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the change of the position and the orientation of the feature point may include a constant value having little change in the movement and rotation of the input fingerprint image. As the input of the change function (definite range), the position shift (distance) and the direction change amount of the feature point are obtained from the output of the corresponding change function (range), and the position and It is characterized by changing the direction.

In the variable fingerprint template generation method irrelevant to the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the constant value having small change in the input and rotation may be obtained by using an input fingerprint image or an image processed fingerprint image. In addition, the method extracts and uses moments irrespective of rotation or movement change in a predetermined region around each feature point.

In the variable fingerprint template generation method irrelevant to the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the invariant value having a small change in the input and rotation is based on the difference between the direction value of each feature point and the peripheral direction value. The feature vector is generated, and a normalized dot product of the same random vector as the feature vector generated by the user ID is used.

In the variable fingerprint template generation method irrespective of the movement and rotation of the input fingerprint image according to an embodiment of the present invention, the change function is a function in which the slope of the change function is adjustable by the developer, and thus the same probability of occurrence within a predetermined interval is provided. Characterized by generating a sum of random variables having.

Hereinafter, with reference to the accompanying drawings an embodiment of the present invention will be described in more detail.

1 is a block diagram of a verification system using a variable fingerprint template generation method independent of movement and rotation of an input fingerprint image according to the present invention.

Referring to FIG. 1, the fingerprint sensor 11 obtains a fingerprint image from a user.

The fingerprint image processing unit 12 processes the fingerprint image acquired by the fingerprint sensor 11 into a ridge-enhanced fingerprint image or a thinned fingerprint image to facilitate feature point extraction.

The fingerprint feature extractor 13 extracts a feature point of the fingerprint from the fingerprint image processed by the fingerprint image processor 12 and extracts a constant value having little change in movement and rotation of the input fingerprint image at each feature point extracted.

The change function storage unit 14 stores the change function generated using the fingerprint information or the user ID.

The change function generation unit 15 automatically generates a change function with the user's ID or obtains the change function previously generated and stored in the change function storage unit 14.

The fingerprint template changing unit 16 obtains the position shift and direction change amount of the fingerprint feature point by using the invariant value and the change function extracted from each feature point, and changes the feature points to generate a changed fingerprint template.

The fingerprint information storage unit 17 stores the changed fingerprint template together with the ID of the user.

The matching unit 18 determines the identity by matching the fingerprint templates stored in the fingerprint information storage unit 17 using the changed fingerprint feature point and the user ID.

The variable fingerprint template generation method applied to the verification system using the variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image according to the embodiment of the present invention configured as described above is performed as follows.

Referring to FIG. 2, when a fingerprint image of the user is obtained through the fingerprint sensor 11, the fingerprint image processing unit 12 processes the fingerprint image to facilitate extraction of a feature point, and then the fingerprint feature extraction unit 13. In step S21 and S22, feature points of the fingerprint are extracted from the input fingerprint image.

For reference, FIG. 3 illustrates an input fingerprint image, a fingerprint feature point extracted from the image, and a feature point of a modified variable fingerprint template according to the present invention. FIG. 3 is a fingerprint feature point extracted from three input fingerprint images a. The conversion result of (b), that is, the feature point (c) of the modified variable fingerprint template is shown.

When the feature point of the fingerprint is extracted as described above, the fingerprint feature extracting unit 13 extracts an invariant value having little change in the position and rotational movement of the input fingerprint at the position of each feature point (S23).

In this case, an invariant value having little change in the movement and rotation of the input fingerprint is obtained from image information of a predetermined region including the feature point at the location of the extracted fingerprint feature point, and uses direction information formed by a fingerprint pattern around the feature point, The invariant value may be extracted using image slope information around the feature point, frequency information of the fingerprint pattern around the feature point, or statistical information.

For example, when the invariant value m _i having little change in movement and rotation is extracted by the fingerprint feature extraction unit 13, the change function generation unit 15 uses a user ID to change the function (L _ID (m)). , Θ _ID (m)) is automatically generated.

As described above, when the invariant value (m _i ) having a small change in the movement and rotation of the input fingerprint and the change function are generated, the fingerprint template changing unit 16 changes the invariant value calculated at each feature point into the change function generator. The change amount of the position movement (distance) and the direction of the feature point is calculated using the corresponding output (range) value as the input (definite range) of the change function generated in (15), and the change amount calculated based on the direction of each feature point. Variable fingerprint templates are generated by changing the feature points as much (S24, S25).

4 is a graph showing a feature change function according to the present invention, and if one feature point M _i = (x _i , y _i , θ _i ) before the change, the feature point M _t = (x _t , y _t , θ _t after the change ) Is obtained as in Equation 1 below.

The change amounts Δx _i , Δy _i , and Δθ _i of the feature points in Equation 1 may be obtained as in Equation 2 below.

In Equation 2, Δx _i and Δy _i represent the amount of change in the position of the feature point, and Δθ _i represents the amount of change in the direction of the feature point. And m _i represents an invariant value that does not change in position and rotation for the i th feature point, and L _ID (m _i ) and Θ _ID (m _i ) correspond to the invariant m _i as the input of the change function (inverse). The position movement (distance) and the direction change amount of the i-th feature point obtained by the output (range).

Here, the position change of the feature point is the distance change amount L _ID (m _i ) in the direction of the direction change amount Θ _ID (m _i ) based on the direction θ _i having the feature point before the change, as shown in FIG. 5. Will be moved by).

By creating a variable fingerprint template with the fingerprint feature point information changed as described above, and using this, the variable fingerprint template with the newly changed fingerprint feature point information is replaced by replacing a change function corresponding to the user ID when the feature point information is stolen or stolen in the system for authenticating an individual. It can be created.

In the variable fingerprint template generation method irrespective of the movement and rotation of the input fingerprint image according to the present invention, the method for obtaining the invariant value (m _i ) obtains a moment in the region around the feature point and calculates the moment of the input fingerprint image. There are a method of using a constant value with little change in rotation and movement, and a method for extracting an invariant value with little change in rotation and movement of an input fingerprint image using direction information formed by a feature point and a fingerprint pattern around the feature point.

In the method using the moment, a region of radius R is set based on each feature point in an input fingerprint image or an image processed fingerprint image (a ridge-enhanced fingerprint image or a thinned fingerprint image), and the input fingerprint image is set in the set region. Find the moment with little change even if rotation or movement occurs.

The moment is to use the statistical information of the image, it is possible to use the moment value of the first, second, third, fourth, and fifth or a combination of the moment value of each order.

For reference, FIG. 6 illustrates an example of a process of obtaining an invariant value having a small change in movement and rotation of an input fingerprint image according to an embodiment of the present invention, wherein a change in rotation or movement of the input image within a radius R at each feature point is shown. The method of obtaining a small moment is shown.

The variable fingerprint template irrelevant to the movement and rotation of the fingerprint image according to the fingerprint feature point changing method described above is obtained by using the moment value of each feature point obtained as described above as an invariant value (m _i ) with little change in rotation and movement of the input fingerprint image. Create

In the method of using the feature point and the direction information around the feature point, as shown in FIG. 7, a local direction θ _Z1 formed by an image in a certain area is obtained at a position Z1 separated by R1 in the direction of each feature point. Local directions (θ _Z2 , θ _Z3 , ...) formed by the image within a certain area at the positions (Z2, Z3, ...) obtained while rotating by an angle α in a clockwise or counterclockwise direction with reference to )

In this way, the local directions θ _Zi ,..., Θ _Zn of the image within a certain area are obtained at distances separated by another distance R2, R3, RN. The feature vector Vi for each feature point is obtained by using the angle difference between the direction of each position thus obtained and the direction θ _Z0 of the feature point.

Here, when the direction of the i th feature point is θ _Z0i and the direction around the feature points of the positions obtained by the above method is θ _Z1i , θ _{Z2i ,} ..., θ _Zni , the feature vector Vi is obtained by the following equation (3). Become.

In Equation 3, min (a, b) selects the minimum value of a and b, and the invariant value (m _i ) having small change in movement and rotation is the same generated through the obtained feature vector Vi and user ID. Using the random vector (K _ID ) of the dimension is generated by the following equation (4).

In Equation 4, | Vi | and | K _ID | refer to a norm of Vi and K _ID vectors.

The variable fingerprint template is independent of the movement and rotation of the fingerprint image according to the fingerprint feature point changing method described above using the constant value obtained as the constant value m _i of each feature point.

In the present invention, in the variable fingerprint template generation method independent of the movement and rotation of the fingerprint image, the change function should be capable of adjusting the amount of change of the function and easily generating a plurality of other change functions. There is a function such as Equation 5.

In Equation 5, x _n is a random variable and has the same occurrence probability in the intervals [-b, -a], [a, b] as shown in FIG. 8, and y _n is a random variable. ) Have the same probability of occurrence in the intervals [-d, -c], [c, d] as shown in FIG. 8 shows graphically the probability density functions of the random variables x _n and y _n . Here, by adjusting a, b, c, and d, the developer can generate a variable fingerprint template that is independent of the movement and rotation of the fingerprint image suitable for the application.

The variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image according to the present invention described above and the verification system using the same are not limited to the above-described embodiments, and the gist of the present invention as claimed in the following claims. Without departing from the technical spirit of the present invention to the extent that anyone who has ordinary knowledge in the field to which the present invention can be variously changed.

The present invention as described above does not authenticate the individual by using the feature information of the original fingerprint, but changes the feature information of the original fingerprint to authenticate the individual with the changed feature information. It is impossible to restore to the fingerprint template, and it is possible to cancel the stolen fingerprint information and create a new changed feature by replacing the change function when the fingerprint information is stolen, thus protecting the privacy infringement and personal fingerprint information that may occur when the fingerprint information is stolen. Can solve the problem.

In particular, in the present invention, since the method of changing the feature point is determined from the image information around the feature point and the change function generated by the user's ID, the alignment between the input fingerprint images is unnecessary and the variable fingerprint template has the same format as the existing fingerprint template. Therefore, even if it is applied to the fingerprint system using the existing matching method without changing the matching method, personal authentication is possible with the variable fingerprint template.

Claims (8)

A fingerprint sensor for obtaining a fingerprint image from a user; A fingerprint image processing unit which processes the fingerprint image obtained by the fingerprint sensor; A fingerprint feature extraction unit for extracting a feature point of the fingerprint from the fingerprint image processed by the fingerprint image processing unit and extracting a constant value having little change in the movement and rotation of the input fingerprint image at each extracted feature point; Equation 5 below using a constant variable (m _i ) extracted by the fingerprint feature extraction unit and a random variable (x _n , y _n ) generated by using an ID of a user as an initial value of a normal random number generator.

(L (m _i ) and Θ (m _i ) are the positional shift (distance) and the direction change of the i th characteristic point determined by the output (range) of the invariant value m _i as the input (definition) of the change function. ) A change function storage unit for storing a change function generated as described above; Automatically generate the change function or change the previously generated and stored change function by using the invariant value extracted by the fingerprint feature extractor and a random variable generated by using the user ID as an initial value of a normal random number generator. A change function generation unit obtained by the storage unit; A fingerprint template changer for obtaining a position shift and a direction change amount of a fingerprint feature point by using invariant values and a change function extracted from each feature point, and generating a changed fingerprint template by changing each feature point; A fingerprint information storage unit for storing the changed fingerprint template with the user's ID; And A matching unit which determines whether the user is present by matching fingerprint templates stored in the fingerprint information storage unit using the changed fingerprint feature points and the user ID; Verification system using a variable fingerprint template method independent of the movement and rotation of the input fingerprint image, characterized in that consisting of. Extracting the feature points of the fingerprint from the processed image after image processing the input fingerprint image; Extracting an invariant value with little change in movement and rotation of the fingerprint image at each feature point extracted through the above steps; Equation 5 below using a random variable (x _n , y _n ) generated by using the extracted invariant (m _i ) and the user ID as an initial value of a normal random number generator.

(L (m _i ) and Θ (m _i ) are the positional shift (distance) and the direction change of the i th characteristic point determined by the output (range) of the invariant value m _i as the input (definition) of the change function. ) A change function generation step of generating a change function such as; Obtaining the position and direction change of the feature point of the fingerprint by using the invariant value and the change function extracted from each feature point; And Generating a variable fingerprint template by changing a feature point of the fingerprint based on the direction of the feature point using the change amount obtained in the step; Variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image, characterized in that consisting of. 3. The method of claim 2, wherein the invariant value with little change in the movement and rotation of the input fingerprint image is a value obtained from a fingerprint image of a predetermined region or a fingerprint image subjected to image processing based on the fingerprint feature point extracted in the fingerprint information extraction step. Invariant values of each feature point using direction information of the fingerprint pattern around the feature point, image tilt information around the feature point, frequency information of the fingerprint pattern around the feature point, and statistical information of the fingerprint pattern around the feature point Method for generating a variable fingerprint template independent of the movement and rotation of the input fingerprint image, characterized in that obtaining. The method of claim 2, wherein the change function is a method of automatically generating or changing a function of acquiring the change function according to a user ID, and generating a new change function or acquiring a new change function from the change function storage if necessary. A variable fingerprint template generation method independent of movement and rotation of an input fingerprint image. 3. The method of claim 2, wherein the change of the position and the direction of the feature point is based on an invariant value having a small change in the movement and rotation of the input fingerprint image as an input (definite range) of the change function. The position movement (distance) and direction change of the feature point are calculated, and the position and the direction of the feature point are changed by the amount of change obtained based on the direction of each feature point of the input fingerprint. How to create a fingerprint template. 4. The method of claim 3, wherein the constant value of the change that is small in the input and rotation is obtained by using an input fingerprint image or an image processed fingerprint image and extracting a moment irrelevant to the rotation or movement change in a predetermined area around each feature point. Variable fingerprint template generation method independent of the movement and rotation of the input fingerprint image, characterized in that. The random vector of claim 3, wherein the constant value having small change in the input and rotation generates a feature vector using a difference between the direction value of each feature point and the surrounding direction value, and is the same random vector as the feature vector generated by the user ID. A variable fingerprint template generation method independent of movement and rotation of an input fingerprint image, characterized by using a normalized dot product of. 3. The movement and rotation of an input fingerprint image according to claim 2, wherein the change function is a function in which the slope of the change function is adjustable by the developer and is generated as a sum of random variables having the same probability of occurrence within a predetermined interval. How to create a variable fingerprint template irrespective of.

Images