KR20030086396A - Recognising human fingerprint method and apparatus independent of location translation , rotation and recoding medium recorded program for executing the method - Google Patents

Abstract

PURPOSE: A method and apparatus for recognizing the fingerprint irrespective of location movement or rotation are provided to determine reliable similarity irrespective of the location movement and rotation generated by time difference between a fingerprint registration and a fingerprint input by discriminating direction information of a feature point and a moved distance of an input fingerprint image. CONSTITUTION: A fingerprint image is acquired using a fingerprint input device(100). Direction information vertical with respect to the most direction of a Fourier spectrum in a sub-region is extracted from each pixel, using an x and y axis gradient by a Sobel operation with respect to the acquired fingerprint image(110). A portion except for a fingerprint is separated from the fingerprint image(120). An original image is divided into a ridge and a valley in only a region in which the fingerprint exists(130). A frame of the ridge is expressed for calculating a CN(Crossing count Number)(140). Feature points are extracted by a CN method(150). An incorrectly extracted feature point is removed among the extracted feature points(160). A moved distance of an input feature point with respect to a registration feature point is calculated using direction information of the extracted feature point(200). The similarity of two fingerprints is measured using the calculated distance for discriminating the truth of a user and indicating a verification result(300).

Description

Fingerprint recognition method independent of position movement and rotation, and recording medium recording fingerprint recognition device and program for executing the method {RECOGNISING HUMAN FINGERPRINT METHOD AND APPARATUS INDEPENDENT OF LOCATION TRANSLATION

The present invention relates to a fingerprint recognition method independent of position movement and rotation, and to a recording medium on which a fingerprint recognition device and a program for executing the method are recorded. Specifically, the position movement caused by the time difference between fingerprint registration and fingerprint input. The present invention relates to a fingerprint recognition method independent of rotation, a fingerprint recognition device, and a recording medium having recorded thereon a program for executing the method.

Fingerprints are formed by the uplift of sweat glands to form a constant flow.The unique characteristics of the fingerprints are different for each individual and do not change for life. It is used as an efficient personal authentication method. The advantages of a fingerprint recognition system are its acceptability, convenience and reliability. Existing research has shown that fingerprint as the means of authentication has the least intrusion among all intrusion prevention technologies, has less user's rejection, and fingerprint recognition equipment requires very little space and low cost equipment. There is this.

However, in the prior art, errors occurred due to aging of the hardware, errors in manufacturing, use under adverse conditions, and differences in brightness between partial regions, and many improved methods were studied in the so-called preprocessing step of extracting feature points to minimize such errors. . However, even though a reliable feature point was obtained through the preprocessing process, the recognition performance was greatly degraded due to the position shift or rotation due to the time difference between the fingerprint registration time point and the fingerprint input time point. It has been pointed out that the problem of making it difficult to process in real time for confirmation of the user.

In order to overcome the problems caused by the movement of the fingerprint image, the method of constructing and matching the tree structure using the feature points extracted through preprocessing as the vertices, that is, the method using the minimum spanning tree (MST) or the angle of the feature points The method using distance was used. However, the method of using the wood structure of the feature points is relatively logical and has the advantage of effectively setting the reference axis between enrollment / input fingerprints, but if the feature points are missing or pseudo feature points remain, or the fingerprint is locally twisted, the spanning tree ) Is severely deformed, greatly increasing the probability of error. In addition, the method using the angle and distance of the feature point has been proposed to reduce the errors caused by the rotation and movement of the fingerprint when comparing the coordinates and directions of the feature point, but this is also the case when the feature point is missing or the pseudo feature point remains or the fingerprint is locally distorted. In this case, the probability of false recognition is greatly increased, and since only the angles and distances are compared, there is a problem in that false recognition rates are increased because they are regarded as the same structure even though they are different structures.

The present invention has been made to solve the conventional problems as described above, an object of the present invention is to efficiently maintain a high recognition rate even when the fingerprint image is moved or rotated, missing feature points or pseudo feature points are not removed. Provided are a fingerprint recognition method independent of position movement and rotation, a fingerprint recognition device, and a recording medium having recorded thereon a program for executing the method.

1 is a flow chart showing the main steps of the fingerprint recognition method according to the present invention.

2 is a flowchart illustrating a process of measuring a position movement distance of an input fingerprint image with respect to a registered fingerprint image in the fingerprint recognition method according to the present invention.

3A is an exemplary diagram showing an example of feature points extracted from a registered fingerprint image in the fingerprint recognition method according to the present invention.

3B is an explanatory diagram showing an example of a specific point extracted from an input fingerprint image in the fingerprint recognition method according to the present invention and showing the position moving distance of the input fingerprint with respect to the registered fingerprint.

4 is a flowchart illustrating a process of determining a specific point similarity between a registered fingerprint and an input fingerprint according to the present invention.

5 is an experimental result table showing a boundary value for the conventional matching method.

6 is an experimental result table showing the boundary value in the fingerprint recognition method of the present invention.

7 is a graph showing the results of experiments showing the distribution of data matching rate in the conventional fingerprint recognition method.

8 is a graph showing the results of experiments showing the distribution of data matching rate in the fingerprint recognition method of the present invention.

9 is a graph of experimental results showing a distribution of data agreement rates of others in the conventional fingerprint recognition method.

FIG. 10 is a graph of experimental results showing a distribution of data agreement rates of others in the fingerprint recognition method of the present invention. FIG.

11 is a graph showing experimental results showing FRR and FAR for the conventional fingerprint recognition method.

12 is a graph of experimental results showing FRR and FAR for the matching method for the fingerprint recognition method of the present invention.

13 is a block diagram showing the configuration of a fingerprint recognition device according to the present invention.

In order to achieve the above object, the present invention provides a fingerprint recognition method irrelevant to position movement and rotation, and a fingerprint recognition device and a recording medium recording a program for executing the method, firstly fingerprint recognition irrespective of position movement and rotation. The method will be described.

The fingerprint recognition method independent of position movement and rotation includes (a) obtaining direction information and position information of a fingerprint ridge from a fingerprint image, extracting feature points of the fingerprint ridge, and then direction information and position information of the extracted feature of the fingerprint ridge. Registering and storing in the database; (b) obtaining direction information and position information of the fingerprint ridge from the input fingerprint image and extracting feature points of the fingerprint ridge; (c) averaging the moving distance from the position of the registered feature point stored in the database to the position of the input feature point using the direction information and the position information of the feature of the registration fingerprint ridge in step (a) and the input fingerprint ridge in step (b) (D) removing feature points outside the area overlapping the registered fingerprint with respect to the input fingerprint by using the calculated moving distance; (e) location information and direction information of the registered fingerprint feature point and the input fingerprint feature point; Determining the similarity between the registration fingerprint and the input fingerprint using the.

Hereinafter, a preferred embodiment according to the fingerprint recognition method of the present invention will be described with reference to the accompanying drawings.

Figure 1 is a flow chart of the main steps of the fingerprint recognition method according to a preferred embodiment of the present invention, as shown in the present invention is a fingerprint image acquisition step 100, direction information extraction step 110, the background region separation step ( 120), binarization step 130, thinning step 140, feature point extraction step 150, pseudo feature point removal step 160, repositioned distance calculation 200, feature point matching step 300.

Fingerprint image acquisition step 100 is a first step of the fingerprint recognition method according to the present invention to obtain a fingerprint image using a fingerprint input device.

In the direction information extracting step 110, using the x and y axis gradients obtained by Sobel calculation on the fingerprint image acquired in the fingerprint image obtaining step 100, each pixel is expressed as Equation 1 below. Direction information is extracted perpendicular to the largest direction of the Fourier spectrum in the subregion.

Here, G _x (x, y), G _y (x, y) are Sobel calculated values and θ _d is direction information of each subregion. The direction information is used to extract ridge flow information in a subregion (ω × ω) of a predetermined size and to be used in subsequent binarization and matching steps.

After that, a smoothing process is performed to remove false direction information due to noise. After doubling θ _d obtained from Equation 1 to separate the x and y components, noise is removed by using a low pass filter (LPF).

Where ω is the size of the filter and the preferred size of the filter is 5 × 5.

Direction information is calculated as shown in Equation 4 using the smoothed x and y components calculated in Equations 2 and 3.

The background region separation step 120 refers to a step of separating a portion other than the fingerprint from the fingerprint image and shortens the calculation time. Since the direction of the ridge is not constant in the fingerprint image, the size of the gradient in the subregion is small. Therefore, as shown in Equation 5, when the magnitude of the gradient is smaller than the threshold Th, it can be separated into a background area.

The binarization step 130 separates the background region 120 and then divides the original image into ridges and valleys only for the region with the fingerprint. In this case, when the threshold value of the same value is set for the entire fingerprint image, a bad result is obtained due to the partial brightness difference in the poor quality image. Adaptive binarization (block binarization) is used to improve the performance of binarization. After dividing the original fingerprint image into blocks of a certain size, the average value of each block is determined as a threshold, and binarization is performed on the block. Preferably divided into 8 × 8 blocks.

Thinning step 140 is a process of expressing the skeleton of the ridge to calculate the CN in the process of extracting candidate feature points. In the present invention, Zhang Suen's thinning algorithm, which is one of the parallel processing methods, is preferably applied.

Extracting the feature point 150 uses a Crossing Count Number (CN) value of Equation 6.

In Equation 6, C ₁ to C ₇ refer to neighboring pixels centered on pixels that are ridges in the thinned image. If the value of CN is 2 using Equation 6, it is preferable to extract the feature point as an ending point and 6 as a bifurcation.

Removal of the pseudo feature point 160 is a process of removing a feature point that is incorrectly extracted from the feature points, and is divided into five structures of short ridges, broken ridges, twig structures, circular structures, and cross structures.

Next, using the direction information of the feature point extracted through the above preprocessing, the position shifted distance of the input feature point relative to the registered feature point is calculated (200), and the similarity of the two fingerprints is measured using the calculated position shifted distance ( 300), the authenticity of the user is determined based on the result, and the verification result is displayed. Hereinafter, a method 200 for calculating a position shifted distance of an input fingerprint image with respect to a registered fingerprint image using direction information and a matching method for measuring similarity using the calculated position shift distance The 300 will be described in detail.

The step 200 of calculating the moved distance of the input fingerprint image with respect to the registered fingerprint image by using the direction information will be described with reference to FIG. 2. 2 is a flowchart illustrating a process of calculating a position moving distance of an input fingerprint image on the basis of a registered fingerprint image according to the present invention. The direction information input step 210, the N feature point storage step 220, and the position movement feature amount comparison Steps 230 and 240 may include a distance calculated step 250.

In the direction information input step 210, the direction information extracted in the direction information extraction step 110 of FIG. 1 is used.

In the N feature point storing step 220, the N feature points in which the direction of the ridge changes rapidly are detected and stored in the present invention. In order to detect the above point, in the present invention, the sum (θ _tot ) of the direction difference from the target pixel is preferably arranged in four directions, that is, a predetermined distance (n) from up, down, left, and right. . In this case, the ridge information θ (x, y) is represented, and x and y mean x and y coordinate values of the target pixel. The following equation (7) is used to detect the point where the direction of the ridge changes abruptly. In Equation 7, φ _x1 , φ _x2 , φ _y1 , and φ _y2 are direction differences between the target pixel and the point separated in four directions, and as the sum of these increases, the direction of the ridge changes rapidly.

In the case of detecting the point where the direction change of the ridge is large as the direction difference as described above, the direction difference with the point away from n pixels is obtained, which is a result of experimenting with various values. Because of the degree of compression, the noise is susceptible to noise, and an error from the distance difference occurs when the direction difference with a pixel that is too far away is obtained. It is preferable to set to 3 pixels experimentally in this invention. Further, in the present invention, although N pieces are extracted in the order that the sum of the direction differences φ _tot is large as described above, ten pieces are experimentally preferable. Also, for each of N, x, y coordinates, φ (xn, y), φ (x + n, y), φ (x, yn), φ (x, y + n) and φ (x, y) Stored as feature quantities, the present invention defines these feature quantities as displacement feature quantities.

The comparing step 230 of the position movement feature amount searches for whether the position information can be regarded as the same position by comparing the direction information with respect to the position movement feature amounts of the N registered fingerprints and the input fingerprints as shown in Equation 8 below.

Here, φ represents direction information of a pixel spaced apart from the target pixel in four directions among the extracted N position changing feature amounts, and subscript r denotes a registration fingerprint and i denotes an input fingerprint. Th _φ is a previously set value, so the rotation may occur when the fingerprint is input.

In the position movement distance calculation step 250, when the above Equation 8 is satisfied among the N position change feature quantities, the input fingerprint and the registration fingerprint are regarded as the same position, and each of their x and y axes is expressed as in Equation 9 below. Compute the position shifted distance (X _d , Y _d ) as the mean for.

Where x _ik and y _ik are the x, y coordinate values, and x _rk and y _rk are the respective x and y coordinate values in the registration feature's _relocation feature list in the input fingerprint. Is the number of feature quantities satisfying the condition of Equation 8 among the N lists stored as the position change feature quantities.

X _d and Y _d calculated by Equation 9 above have the following meanings. 1) If X _d has a positive value, it moves to the left with respect to the enrolled fingerprint and the input fingerprint enters. 2) If X _d has a negative value, it refers to the registered fingerprint. in the case on the position moved to the right contains the fingerprint, 3) Y _d is negative (- a case, if having a value of) the input fingerprint entered by the moving position upwards relative to the registration fingerprint, 4) Y _d is If it has a positive value, it means that the input fingerprint enters by moving downward based on the registered fingerprint.

Figure 3 (a) is an exemplary view showing an example of a feature point extracted from the registered fingerprint image in the fingerprint recognition method according to the invention, Figure 3b is a view of the specific point extracted from the input fingerprint image in the fingerprint recognition method according to the present invention It is an explanatory drawing which shows the example and shows the position moving distance of the input fingerprint with respect to a registration fingerprint.

Determining the feature point similarity 300 will be described with reference to FIG. 4. FIG. 4 is a flowchart illustrating a process of determining a specific point similarity between a registration fingerprint and an input fingerprint according to the present invention. An input step 310, a feature point deletion step 320, a similarity comparison step 330, 340, 350, and a similarity detection step 360 are included.

In the distance input step 310 in which the feature point and the position are moved, a position shift distance of the feature point extracted in the above process, that is, X _d , Y _d , a feature point of the registered fingerprint, and a feature point list of the input fingerprint are input.

In the feature point deletion step 320, the feature points outside the region included in the two fingerprints are removed from the feature point list using the X _d and Y _d values. In more detail, when X _d has a positive value, since the input fingerprint is moved to the left compared to the enrolled fingerprint, the x coordinate value is 0 to X _d in the feature point list of the enrolled fingerprint. If it is located in between, it is removed from the feature point list.In the case of an input fingerprint, if the x coordinate value is located between (X _image -X _d ) and X _image in the feature point list, it is also removed from the feature point list. X _image means the x-axis size of the fingerprint image. On the other hand, X _d is negative (-) at a value a is because the x coordinate value (X _image + X _d) in the feature point list of the registered fingerprint when the position moves to the input fingerprint right when compared to the fingerprint when having the X _image If it is located between, it is removed from the feature point list. Also, in the case of an input fingerprint, if the x coordinate value is located between 0 and -X _d in the feature point list, it is removed from the feature point list. For the y-axis, the feature points are removed in the same manner as the above procedure.

The similarity determination step (330, 340, 350) is performed on the feature point list after the above process. The feature points of the feature list registration are the feature points located in the common area of the registration fingerprint and the input fingerprint. Determine similarity for In order to measure the similarity in the feature point direction, the absolute value of the difference between the direction θ _r of the registered fingerprint feature point and the direction θ _i of the input fingerprint feature point is compared with a threshold for rotation previously set (340). In order to determine the similarity of the position of the feature point after determining the degree of similarity with respect to the direction of the feature point, the input fingerprint for the position (x _r , y _r ) of the registered fingerprint feature point in consideration of the X _d , Y _d information as shown in Equation 10 Calculate the distance of the feature point position (x _i , y _i ).

Here, Th _d is a threshold for the movement, and the coordinates of the two feature points are equalized by summing the distance (X _d , Y _d ) that has been moved to the position of the input fingerprint feature point.

If the above Equation 10 is satisfied (350), it belongs to a common area of the registration / input fingerprint and is regarded as the same feature point.

In the similarity detection step 360, the same number of PMs is common to both fingerprints for the sum of the number of fingerprint fingerprint points EM _r and the number of input fingerprint features EM _i , as shown in Equation 11 below. As a percentage value of, the matching ratio of two fingerprint feature points (MR) is obtained.

Compared with the verification reference value for verifying the matching rate and similarity, if the matching rate is equal to or greater than the verification reference value, the identification is determined by the same fingerprint, and other fingerprints are used to determine the authenticity of the user.

Experimental Example

Experimental results of evaluating the performance of fingerprint recognition by the above method will be described with reference to FIGS. 5 to 12. In this experiment, 455 thumb fingerprint images were inputted from 35 persons, 13 per person, using a semiconductor sensor (Authen Tec, FingerLoc AF-S2, 500dpi, 256 × 256). The software for preprocessing, feature extraction and matching of the entire fingerprint recognition system uses Microsoft Visual C ++ 6.0.

The experiment uses 5 fingerprint images as registration data for each individual, and the verification compares 8 fingerprint images and 8 fingerprint images of other persons except for registration data of the same person, and proves the improvement effect of matching part. Experiment with the matching method using the angle and distance of existing feature points.

In this experiment, the results are expressed as False Reject Rate (FRR) and False Accept Rate (FAR). FRR is the rate of rejection if you reject yourself because you consider yourself to be someone else, and FAR is the rate of misrecognition if you mistakenly accept someone as yourself. After FRR and FAR are found, the trade-off is used to find the point in the graph that minimizes both.

5 to 12 show the results of this experiment. 5 is a test result table showing a boundary value for a conventional matching method, FIG. 6 is a test result table showing a boundary value in the fingerprint recognition method of the present invention, FIG. Figure 8 is a graph showing the results of the experiment, Figure 8 is a graph of the experimental results showing the distribution of the data match rate of his own in the fingerprint recognition method of the present invention, Figure 9 is an experiment showing the distribution of data matching rate of others in the conventional fingerprint recognition method Figure 10 is a graph of the experimental results showing the distribution of data agreement rate of others in the fingerprint recognition method of the present invention, Figure 11 is a graph of the experimental results showing the FRR and FAR for the conventional fingerprint recognition method, 12 is a graph of experimental results showing FRR and FAR for the matching method for the fingerprint recognition method of the present invention.

In the graph of FIG. 11, when the point that minimizes both FRR and FAR is found, the boundary value at that time is 53.5%, FRR is 7.000%, FAR is 6.664%, and both FRR and FAR are shown in the graph of FIG. When the minimum point is found, the boundary value at that time is 44.5%, the FRR is 2.000%, and the FAR is 2.023, indicating that the present invention contributes to the performance improvement in the fingerprint matching method.

A fingerprint recognition device independent of position movement and rotation according to another embodiment of the present invention will be described with reference to FIG. 13. FIG. 13 is a block diagram showing a configuration of a fingerprint recognition device according to the present invention, which includes a direction information extracting unit, a feature point extracting unit, a database storage unit, a moving distance calculator, a common area extra feature point removing unit, and a similarity determining unit.

The direction information extracting unit obtains the direction information and the position information of the fingerprint ridge from the input or registration fingerprint image. The feature point extraction unit performs the CN method through the steps of fingerprint region and background region separation 120 using the magnitude of gradient, adaptive binarization 130 and thinning 140 to distinguish ridges and valleys as described in the fingerprint recognition method. The feature point extractor extracts the feature points. The database storage registers and stores the direction information and the position information of the feature points of the fingerprint ridges extracted by the feature point extractor in the database. The moving distance calculator calculates a moving distance of the input feature point from the position of the registered feature point stored in the database by using the direction information and the location information of the feature points of the input fingerprint ridge and the registered fingerprint ridge. The non-common area feature point removal unit removes the feature points outside the area overlapping the registered fingerprint with respect to the input fingerprint by using the calculated movement distance. The similarity determining unit determines the similarity between the registered fingerprint and the input fingerprint using the location information and the direction information of the registered fingerprint feature point and the input fingerprint feature point. See the detailed description of the fingerprint recognition method for detailed functions and actions performed by each component of the fingerprint reader.

A computer readable recording medium having recorded thereon a program for executing a fingerprint recognition method according to another embodiment of the present invention will be described. Recording of a program for executing the fingerprint recognition method according to the present invention with a computer The medium includes the steps of: (a) acquiring direction information and location information of the fingerprint ridge from the fingerprint image, extracting feature points of the fingerprint ridge, and registering and storing the extracted direction information and location information of the feature of the fingerprint ridge in a database; (b) obtaining direction information and location information of the fingerprint ridge from the input fingerprint image and extracting feature points of the fingerprint ridge; (c) averaging the moving distance from the position of the registered feature point stored in the database to the position of the input feature point using the direction information and the position information of the feature of the registration fingerprint ridge in step (a) and the input fingerprint ridge in step (b) (D) removing the feature points outside the area overlapping the registration fingerprint with respect to the input fingerprint by using the calculated travel distance; and (e) location information and direction information of the registration fingerprint feature point and the input fingerprint feature point. A program for performing a procedure for determining the similarity between the enrolled fingerprint and the input fingerprint is recorded. The functions and actions performed by each procedure of the program are as described above in the detailed description of the fingerprint recognition method, which will be omitted here.

As described above, the fingerprint recognition matching method according to the present invention is a matching method for determining similarity using the direction information of the feature point and the position shifted distance of the input fingerprint image with respect to the registered fingerprint image. Reliable similarity can be determined regardless of the position shift and rotation caused by the time difference and the image is damaged in the fingerprint itself or the distribution of contrast value is irregular, especially when the missing feature point or the pseudo feature point is not removed. Because it has relatively accurate information, it has the effect of reducing False Acceptance Rate (FAR) and False Rejection Rate (FRR).

In addition, the feature points other than the overlapping areas of the registration and input fingerprints are removed and compared only with respect to the remaining feature points according to the position moving distance.

What has been described above is only one embodiment for implementing the fingerprint recognition matching method using the direction information and the position movement distance of the fingerprint ridge, the present invention is not limited to the above-described embodiment, it is claimed in the claims As will be apparent to those skilled in the art to which the present invention pertains without departing from the gist of the present invention, the technical spirit of the present invention will be described to the extent that various modifications can be made.

Claims (8)

(a) acquiring direction information and location information of the fingerprint ridge from a fingerprint image, extracting feature points of the fingerprint ridge, and registering and storing the extracted direction information and location information of the feature of the fingerprint ridge in a database; (b) obtaining direction information and position information of the fingerprint ridge from the input fingerprint image and extracting feature points of the fingerprint ridge; (c) averaging the moving distance from the position of the registered feature point stored in the database to the position of the input feature point using the direction information and the position information of the feature of the registration fingerprint ridge in step (a) and the input fingerprint ridge in step (b) Calculating by; (d) removing feature points outside the area overlapping the registered fingerprint with respect to the input fingerprint using the calculated movement distance; (e) determining the similarity between the registration fingerprint and the input fingerprint by using the position information and the direction information of the registered fingerprint feature point and the input fingerprint feature point. The method of claim 1, The extraction of the feature points of step (b) is performed by first comparing the sum of the pixel values of the ridge direction peripheral pixels of the center pixel to be processed and the sum of the peripheral pixel values above and below the ridge direction located at a vertically adjacent distance. Fingerprint recognition method comprising an adaptive binarization step. The method of claim 1, Calculating the movement distance of step (c) calculates the distance moved by averaging the distance between each registered feature point and the input feature point using only the predetermined feature points extracted in the order of rapidly changing ridge flow. Fingerprint recognition method independent of rotation. The method according to claim 1 or 3, Extraction of the feature points used in the movement distance calculation in step (c) is performed by calculating the direction difference with neighboring blocks in four directions from each block of the direction image, and extracting the sums in the order of increasing sum. Fingerprint recognition method irrespective of. The method of claim 1 or 3, The calculation of the movement distance of step (c) further includes the step of measuring the position movement distance only at the points where the direction difference of the registered feature point with respect to the input feature point satisfies the threshold for rotation. Unrelated fingerprint recognition method. The method of claim 3, wherein The calculation of the movement distance of step (c) further includes the step of measuring the position movement distance only at the points where the direction difference of the feature points satisfies the threshold for rotation. Fingerprint recognition method independent of position movement and rotation . A direction information extraction unit for obtaining direction information and location information of the fingerprint ridge from an input or registration fingerprint image; A feature point extraction unit for extracting feature points of the fingerprint ridge; A database storage unit for registering and storing the extracted direction information and location information of the feature points of the fingerprint ridge in a database; A moving distance calculator for calculating a moving distance of the input feature point from the position of the registered feature point stored in the database by using the direction information and the position information of the feature of the input fingerprint ridge and the registration fingerprint ridge; A non-common area feature point removal unit for removing a feature point other than an area overlapping a registered fingerprint with respect to an input fingerprint using the calculated moving distance; A fingerprint recognition device irrelevant to position movement and rotation, comprising: a similarity determining unit for determining similarity between a registered fingerprint and an input fingerprint using location information and direction information of a registered fingerprint feature point and an input fingerprint feature point. A computer-readable recording medium having recorded thereon a program for executing the method of any one of claims 1 to 6.