JP6366420B2 - Data creation program, data creation method, and data creation device - Google Patents

Description

  The present invention relates to a data creation program for creating data used for fingerprint authentication, a data creation method, and a data creation apparatus.

  In recent years, the performance of mobile devices such as smartphones and notebook computers has improved, and the information handled has become more sophisticated. For this reason, mobile devices having a fingerprint authentication function are widely used. When a fingerprint sensor is mounted on a mobile device, the fingerprint sensor is often small and has a unique shape due to design and space constraints. For example, a cellular phone in which a part that has been conventionally used as a button has a fingerprint sensor function is known (see Patent Document 1).

JP 2009-48418 A

  Accompanying miniaturization of the fingerprint sensor, the acquired fingerprint image also becomes smaller. In many cases, a user performs a fingerprint input operation by bringing the finger of the hand holding the mobile device into contact with a fingerprint sensor mounted on the mobile device. In this case, since the finger must be moved in an unnatural direction, the fingerprint input operation is likely to be unstable, and the fingerprint image is likely to be displaced or rotated. Therefore, there is a demand for a technique for performing highly accurate authentication of a fingerprint image (for example, a 5-10 mm square fingerprint image) acquired based on a smaller and unstable input operation.

  An object of the present invention is to provide a data creation program, a data creation method, and a data creation for creating data capable of authenticating a fingerprint image acquired based on a small and unstable input operation with sufficient accuracy. Is to provide a device.

  The data creation program according to the first aspect of the present invention has a first acquisition step of acquiring a fingerprint image and each of a plurality of arbitrary points of the fingerprint image acquired by the first acquisition step. A second acquisition step of acquiring the respective densities of the plurality of pixels corresponding to the circumference of the predetermined radius for each of the plurality of points; and the plurality of the plurality of points acquired for each of the plurality of points by the second acquisition step. An amplitude calculation step for calculating an amplitude spectrum for each of the plurality of points based on the order of density on the circumference, and each of the plurality of amplitude spectra calculated for each of the plurality of points by the amplitude calculation step. An amplitude storage step of storing a plurality of corresponding peak frequencies in the storage unit for each of the plurality of points is caused to be executed by a program of the data creation device.

  According to the first aspect, the data creation device that operates by executing the data creation program includes the density of pixels corresponding to a plurality of positions on a circumference centered on each of a plurality of arbitrary points of the fingerprint image. Based on the above, an amplitude spectrum is calculated for each of a plurality of points. The data creation device stores a plurality of peak frequencies corresponding to the calculated amplitude spectrum in a storage unit for each of a plurality of points as data that can be used for fingerprint authentication.

  In the above case, an authentication apparatus that performs fingerprint authentication using the peak frequency stored in the storage unit can collate the fingerprint by determining the similarity of each position of the plurality of peak frequencies. Since the peak frequency is specified for each of a plurality of points, a large number of peak frequencies can be specified even when the fingerprint image is small. Therefore, the authentication apparatus can perform fingerprint authentication with sufficient accuracy even when the fingerprint image is small and the fingerprint image is acquired based on an unstable input operation. Thus, the data creation device can create data that allows the authentication device to perform fingerprint authentication with sufficient accuracy.

  In the first aspect, the amplitude calculation step may calculate the amplitude spectrum by DFT processing based on the plurality of concentrations acquired for each of the plurality of points by the second acquisition step. In this case, the data creation device can calculate the amplitude spectrum of an arbitrary point based on the density of each of the plurality of pixels.

  In the first aspect, the amplitude storage step may store a predetermined number of peak frequencies in the storage unit in descending order of the intensity of the corresponding amplitude spectrum among the plurality of peak frequencies. In this case, the data creation device can suppress the capacity of data stored in the storage unit.

  In the first aspect, based on the order of arrangement of the plurality of concentrations acquired for each of the plurality of points by the second acquisition step on the circumference, phase information corresponding to the phase spectrum for each of the plurality of points is obtained. A phase calculating step for calculating each of the plurality of points, and a storage unit storing data related to the plurality of pieces of phase information calculated for each of the plurality of points by the phase calculating step in the storage unit. A data storage step may be further performed. In this case, the authentication device can perform fingerprint authentication based on the peak frequency and data related to the phase information stored in the storage unit. Therefore, the authentication device can further improve the accuracy of authentication compared to a case where fingerprint authentication is performed based only on the peak frequency.

  In the first aspect, when the phase calculation step represents a cumulative value of a plurality of real parts calculated by the DFT processing as ΣRe (n) and a cumulative value of a plurality of imaginary parts as ΣIm (n), arctan (ΣIm A value calculated by (n) / ΣRe (n)) may be calculated as the phase information. In this case, the data creation device can easily calculate the phase information.

  According to a second aspect of the present invention, there is provided a data generation method centered on a first acquisition step of acquiring a fingerprint image and a plurality of arbitrary points on the fingerprint image acquired by the first acquisition step. A second acquisition step of acquiring the respective densities of the plurality of pixels corresponding to the circumference of the predetermined radius for each of the plurality of points; and the plurality of the plurality of points acquired for each of the plurality of points by the second acquisition step. An amplitude calculation step for calculating an amplitude spectrum for each of the plurality of points based on the order of density on the circumference, and each of the plurality of amplitude spectra calculated for each of the plurality of points by the amplitude calculation step. An amplitude storage step of storing a plurality of corresponding peak frequencies in the storage unit for each of the plurality of points. According to the 2nd aspect, there can exist an effect similar to a 1st aspect.

  According to a third aspect of the present invention, there is provided a data creation device including a first acquisition unit configured to acquire a fingerprint image and a plurality of arbitrary points on the fingerprint image acquired by the first acquisition unit. Second acquisition means for acquiring the respective densities of a plurality of pixels corresponding to a circumference of a predetermined radius for each of the plurality of points; and the plurality of the plurality of points acquired for each of the plurality of points by the second acquisition means. Based on the order of density on the circumference, amplitude calculation means for calculating an amplitude spectrum for each of the plurality of points, and each of the plurality of amplitude spectra calculated for the plurality of points by the amplitude calculation means Amplitude storage means for storing a plurality of corresponding peak frequencies in the storage unit for each of the plurality of points. According to the 3rd aspect, there can exist an effect similar to a 1st aspect.

2 is a functional block diagram of the authentication system 1. FIG. It is a figure which shows the fingerprint image. FIG. 6 is a diagram for explaining a function for acquiring densities corresponding to a plurality of points on a circle. It is a graph which shows an example of the density | concentration in the position on the periphery of the circle | round | yen 7. It is a figure which shows the fingerprint image 61 after a filtering process. It is a graph which shows an example of the density | concentration smoothed by the moving average. It is a graph which shows an example of an amplitude spectrum. It is a figure for demonstrating the identification method of direction data. It is a figure for demonstrating the identification method of direction data. It is a figure which shows the relationship between the fingerprint image 6 and the image 65 corresponding to direction data. 2 is a block diagram showing an electrical configuration of a data creation device 3 and an authentication device 5. FIG. It is a flowchart of a 1st main process. It is a figure which shows the characteristic data. It is a figure which shows fingerprint data. It is a flowchart of the 2nd main process.

  An embodiment of the present invention will be described with reference to the drawings. In addition, the concrete several numerical value illustrated in the following embodiment is an example, and this invention is not limited to these several numerical values.

<Outline of authentication system 1>
An overview of the authentication system 1 will be described with reference to FIG. The authentication system 1 is a system for authenticating a user with a fingerprint. The authentication system 1 includes a data creation device 3 and an authentication device 5. The authentication device 5 is an electronic device having a fingerprint authentication function. In the present embodiment, the authentication device 5 is a known smartphone. The data creation device 3 generates fingerprint data (see FIG. 14) necessary for the authentication device 5 to perform authentication by fingerprint, and stores the fingerprint database (DB) stored in the flash memory 504 (see FIG. 11) of the authentication device 5. ) Register at 60. In the present embodiment, the data creation device 3 is a known PC.

  The authentication system 1 is not limited to the above configuration. For example, the authentication system 1 may have only a single device having the functions of the data creation device 3 and the authentication device 5. Specifically, the authentication system 1 may be an authentication device 5 having a fingerprint data generation function and a fingerprint authentication function. The authentication device 5 is not limited to a smartphone, and may be any of a notebook PC, a tablet PC, a mobile phone, an ATM, and an entry / exit management device. The data creation device 3 is not limited to a PC, and may be a server or a dedicated device for generating fingerprint data.

  The fingerprint DB 60 may be stored in the HDD 304 (see FIG. 11) of the data creation device 3. The authentication device 5 may execute fingerprint authentication by referring to the fingerprint DB 60 stored in the HDD 304 of the data creation device 3. The fingerprint DB 60 may be stored in a server (not shown) connected to the network. In this case, the data creation device 3 may register the fingerprint data generated in the fingerprint DB 60 stored in the server. The authentication device 5 may perform fingerprint authentication by referring to the fingerprint DB 60 stored in the server.

<Function of Data Creation Device 3>
The data creation device 3 includes a fingerprint image acquisition unit 32, a polar coordinate acquisition unit 33, a DFT calculation unit 34, an amplitude spectrum calculation unit 35, a peak frequency extraction unit 36, a phase information calculation unit 37, a direction data specification unit 38, and a registration unit. The processing corresponding to each of the functional blocks 39 is executed by the CPU 301 (see FIG. 11). A fingerprint sensor 31 is connected to the data creation device 3.

  The fingerprint sensor 31 is a capacitance type surface sensor that determines the unevenness of the fingerprint from the charge amount of the matrix-like surface electrode. The size of the fingerprint sensor 31 is preferably the same size as the fingerprint sensor 51 described later or larger than the fingerprint sensor 51. The size of the fingerprint sensor 31 is not particularly limited, but for example, a 5 × 5 mm fingerprint sensor 31 can be applied. The resolution of the fingerprint sensor 31 is preferably the same as or higher than the fingerprint sensor 51 described later. Although the resolution of the fingerprint sensor 31 is not particularly limited, for example, the fingerprint sensor 31 of 508 dpi or 363 dpi can be applied. Hereinafter, the case where the resolution of the fingerprint sensor 31 is 508 dpi will be described as an example.

  Note that the detection method of the fingerprint sensor 31 is not limited to the capacitance method, and other methods (for example, an electric field method, a pressure method, and an optical method) may be used. The fingerprint sensor 31 is not limited to a surface type, and may be a linear type. The size and resolution of the fingerprint sensor 31 are not limited to the above specific example.

  The fingerprint image acquisition unit 32 acquires an image in which the unevenness of the fingerprint of the finger touching the fingerprint sensor 31 is shown in gray scale based on a signal output from the fingerprint sensor 31 while the finger is in contact with the fingerprint sensor 31. Is done. Hereinafter, an image acquired based on a signal output from the fingerprint sensor 31 is referred to as a “fingerprint image”. One dot of the fingerprint sensor 31 corresponds to one pixel of the fingerprint image. FIG. 2 shows a fingerprint image 6 which is an example of a fingerprint image acquired by the fingerprint image acquisition unit 32. The number of pixels of the fingerprint image 6 is 200 pixels (horizontal direction) × 400 pixels (vertical direction). The fingerprint image 6 includes an end point 6A of a ridge that is a series of convex portions of the fingerprint, a branch point 6B divided into two, and the like. Hereinafter, end points, branch points, and the like of fingerprints are collectively referred to as “a plurality of feature points”.

  The horizontal direction in FIG. 2 is called the X direction of the fingerprint image 6, and the vertical direction in FIG. 2 is called the Y direction of the fingerprint image 6. The right direction in FIG. 2 is referred to as a positive direction in the X direction, and the left direction in FIG. 2 is referred to as a negative direction in the X direction. The downward direction in FIG. 2 is referred to as a positive direction in the Y direction, and the upward direction in FIG. 2 is referred to as a negative direction in the Y direction. The position of the upper left pixel of the fingerprint image 6 is defined as the origin (0, 0) of the orthogonal coordinates. The position of a pixel that is separated by x pixels in the X direction from the origin of the orthogonal coordinates and separated by y pixels in the Y direction from the origin is denoted as orthogonal coordinates (x, y). Hereinafter, the case where the fingerprint image 6 is acquired by the fingerprint image acquisition unit 32 will be described as an example. Note that the number of pixels of the fingerprint image 6 is not limited to the above specific example.

  As shown in FIG. 1, the polar coordinate acquisition unit 33 specifies a plurality of positions from the fingerprint image 6 by the following method, and acquires the density corresponding to each of the specified plurality of positions. First, as shown in FIG. 3A, any one of a plurality of pixels included in a frame 6D (see FIG. 2) of the fingerprint image 6 acquired by the fingerprint image acquisition unit 32 is a circle 7 ( This is determined as the center point 7A in FIG. The value of the radius r is not particularly limited. For example, the value is 15 pixels when the resolution of the fingerprint image 6 is 508 dpi, and 11 pixels when the resolution of the fingerprint image 6 is 363 dpi. In the present embodiment, since the resolution of the fingerprint sensor 31 is 508 dpi, a specific description will be given below with an example in which the radius r is 15 pixels. Note that the value of the radius r is not limited to the above specific example.

  Next, as shown in FIG. 3B, a predetermined number of positions on the circumference of the circle 7 having the radius r centered on the center point 7A are determined in order. Hereinafter, the predetermined number is expressed as “K1”. Specifically, when the center point 7A is the origin of the polar coordinates, the position of the polar coordinates (r, θ) (position on the circumference of the circle 7) indicated by the radius r and the declination angle θ is the declination angle θ. K1 pieces are determined in order while changing clockwise from 0 (direction extending in the horizontal direction to the right from the center point 7A) to 2π in increments of (2π / K1) degrees. Hereinafter, each of the determined K1 positions is referred to as a “sample”. The value of K1 is not particularly limited. Hereinafter, the case where K1 is “128” will be described as an example.

  Next, as shown in FIG. 3C, the density corresponding to each of the determined 128 samples is acquired. Specifically, based on the well-known bilinear method, the density corresponding to each sample is calculated by weighting each density of pixels adjacent to the top, bottom, left, and right of the position of 128 samples on the circumference of the circle 7. Calculated. Note that the weighting coefficient for performing the weighting calculation may be stored in the HDD 304 in advance.

  FIG. 4 shows an example of a graph in which the density corresponding to each of the 128 samples when the deflection angle θ is changed from 0 to 2π is plotted and connected by a straight line. As described above, the result when the concentrations are arranged in the order of sample acquisition, that is, in the order of the deflection angle θ, is a polygonal line having a plurality of peaks arranged periodically.

  3 (A) to 3 (C), the plurality of 200 × 400 pixels included in the fingerprint image 6 are arranged inward by a radius r (15 pixels) from each of the upper end, the lower end, the left end, and the right end. The process is executed for each of the plurality of arranged pixels in the frame 6C in FIG. A plurality of pixels (170 (= (200−2 × 15)) × 370 (= (400−2 × 15)) are arranged in the frame 6C.

  Note that the polar coordinate acquisition unit 33 may perform a filtering process on the fingerprint image 6 before acquiring the density based on the fingerprint image 6. Specifically, it is as follows. For example, as shown in FIG. 5, a filtering process for removing high-frequency components as noise is performed on the fingerprint image 6 (FIG. 5A) to generate a fingerprint image 61 (FIG. 5B). May be. Then, the processing of FIGS. 3A to 3C may be performed on the fingerprint image 61 from which the high-frequency component has been removed, and the density may be acquired. Note that, by executing the filtering process, the shading change of the edge portion of the original fingerprint image 6 becomes gentle. Any of a known low-pass filter, Gaussian filter, moving average filter, median filter, and averaging filter may be used as a filter used in the filtering process. Further, for example, a filtering process for extracting only a specific frequency band component may be executed on the fingerprint image 6. As a specific frequency band, a band including the period of the concave and convex portions of the fingerprint may be selected. A known band pass filter may be used as a filter used for the filtering process.

  In addition, after the density is acquired based on the fingerprint image 6 in the polar coordinate acquisition unit 33, the polygonal line waveform shown in FIG. 4 may be smoothed by a moving average. FIG. 6 shows the result of smoothing the graph shown in FIG. 4 by moving average. As shown in FIG. 6, by performing smoothing, the change in the broken line waveform of the density is smoothed.

  The reason for performing the above filtering process or smoothing is as follows. In fingerprint images acquired from fingerprint sensors 31 of different detection methods and types, there may be differences in characteristics such as resolution and gradation. Further, depending on the detection method and type of the fingerprint sensor 31, there are cases where the acquired fingerprint image includes a lot of noise, or the fingerprint image is acquired with black and white reversed. In such a case, the difference may be absorbed by filtering processing or smoothing to be in a state suitable for subsequent processing.

  As shown in FIG. 1, the DFT calculation unit 34 uses a known discrete Fourier transform (Discrete Fourier Transform) based on 128 densities corresponding to each of a plurality of 170 × 370 pixels acquired by the polar coordinate acquisition unit 33. DFT) operation is performed. A real part and an imaginary part are calculated for every 128 frequency components by the DFT calculation. Hereinafter, the calculated real part is expressed as “Re” and the imaginary part is expressed as “Im”. Re and Im corresponding to each of the 128 frequency components calculated by the DFT operation are denoted as “Re (n)” and “Im (n)”, respectively. n = 0, 1, 2, 3,... 127. Re (n) and Im (n) are calculated for each of a plurality of pixels of 170 × 370.

In the amplitude spectrum calculation unit 35, based on Re (n) and Im (n) calculated for each of the plurality of pixels of 170 × 370 by the DFT calculation unit 34, the amplitude spectrum is expressed by the following equation (1). Is calculated. Hereinafter, the calculated amplitude spectrum is expressed as “A (n)”. A (n) is calculated for each of a plurality of pixels of 170 × 370.
A (n) = √ (Re (n) ² + Im (n) ² ) (n = 0, 1, 2,... 127) (1)

  FIG. 7 is an example of an amplitude spectrum corresponding to one pixel among a plurality of pixels of 170 × 370, and plots amplitude spectra (A (0) to A (63)) corresponding to n = 0 to 63. And a graph connected with a straight line. According to the sampling theorem, the amplitude spectrum (A (64) to A (127)) corresponding to n = 64 to 127 has a shape obtained by horizontally inverting A (0) to A (63). Therefore, the amplitude spectrum (A (64) to A (127)) is omitted in FIG.

  As shown in FIG. 1, in the peak frequency extraction unit 36, n corresponding to a predetermined number of peaks in the descending order of amplitude among the amplitude spectrum calculated by the amplitude spectrum calculation unit 35 is extracted as a peak frequency by a predetermined number. Is done. Hereinafter, the predetermined number is expressed as “K2”. In this case, the number of extracted peak frequencies can be adjusted by adjusting K2. Further, by adjusting K2, there is a case where the data amount of the peak frequency corresponding to one pixel can be suppressed as compared with the case where n corresponding to all the peaks is extracted as the peak frequency. Hereinafter, a case where K2 is “4” will be described as an example. In addition, about the value of K2, it is not limited to said specific example, What is necessary is just two or more arbitrary values.

  When the amplitude spectrum is calculated by the amplitude spectrum calculation unit 35 based on the fingerprint image 6 (for example, the fingerprint image 61 (see FIG. 5)) on which the filtering process has been performed on the fingerprint image 6, a high frequency component that is a noise component The peak corresponding to is lower. In this case, an appropriate peak frequency is easily extracted.

  In the registration unit 39, data indicating the peak frequency extracted by the peak frequency extraction unit 36 is registered in the fingerprint DB 60 stored in the flash memory 504 (see FIG. 11) of the authentication device 5 described later. In the registration unit 39, direction data specified by a direction data specifying unit 38 described later is also registered in the fingerprint DB 60. Details will be described later.

  The phase information calculation unit 37 calculates phase information by the following method based on Re (n) and Im (n) calculated for each of a plurality of pixels of 170 × 370 by the DFT calculation unit 34. First, a value obtained by accumulating each Re (n) from a predetermined first number to a predetermined second number is calculated. Hereinafter, a value obtained by accumulating Re (n) is represented as “ΣRe (n)”. Next, a value obtained by accumulating each Im (n) from the first number to the second number is calculated. Hereinafter, a value obtained by accumulating Im (n) is referred to as “ΣIm (n)”. The predetermined first number is expressed as “K3”, and the predetermined second number is expressed as “K4”. Hereinafter, the case where K3 is “1” and K4 is “15” will be specifically described as an example. In addition, about the value of K3 and K4, it is not limited to said specific example. K3 may be an arbitrary number of 1 or more. K4 may be any value between “K3 + 1” and 127 or less. For example, K3 may be “2” and K4 may be “17”.

  In the above, the reason for setting the value of n from 1 to 15 or from 2 to 17 is to suppress an increase in the processing amount of the cumulative calculation. Note that n greater than about 30 corresponds to a frequency greater than the frequency of the fingerprint irregularities. For this reason, the values of Re (n) and Im (n) corresponding to n greater than about 30 are likely to be very small. Therefore, the difference in the calculated phase information is often small when n is 0 to 127 and when n is limited to 1 to 15 or 2 to 17.

Next, calculation is performed based on the following equation (2). The calculated value is referred to as “phase information” and expressed as “B (i)”. i indicates the order in which each of the plurality of pixels of 170 × 370 is acquired by the polar coordinate acquisition unit 33, and i = 1, 2,... 62900 (= 170 × 370)).
B (i) = arctan (ΣIm (n) / ΣRe (n)) (i = 1, 2,... 62900) (2)
Note that B (i) takes any value between −π / 2 and π / 2.

  Note that the phase information calculated by the above method is a pixel represented by a phase spectrum calculated for each frequency component based on Re (n) and Im (n) where n is 1 to 15, respectively. There is a correlation with the angle of the synthesized vector synthesized every time. That is, the above calculation method suppresses the amount of calculation processing by reducing the angle of the combined vector based on the phase spectrum calculated based on Re (n) and Im (n) where n is 1 to 15. It can be said that this is a simple calculation method.

The direction data specifying unit 38 calculates a direction for each of the plurality of pixels of 170 × 370 based on B (i) calculated for each of the plurality of pixels of 170 × 370 by the phase information calculation unit 37 by the following method. , Generate direction data. First, B (i) calculated by the phase information calculation unit 37 is converted into a density corresponding to each phase information by the following equation (3). Hereinafter, the calculated density is expressed as “D (i)”.
D (i) = ((B (i) + π / 2) × 255) / π (i = 1, 2,... 62900) (3)

  This will be specifically described with reference to FIG. FIG. 8A shows the fingerprint image 6 acquired by the fingerprint image acquisition unit 32. FIG. 8B shows the fingerprint image 62 after the smoothing process (see FIG. 6) is performed on the fingerprint image 6. FIG. 8C shows a phase information map 63 in which the grayscale image of D (i) calculated by the above method is arranged at the corresponding pixel position based on the fingerprint image 62. FIG. 8D is an enlarged view of a part of the phase information map 63 shown in FIG. Next, as shown in FIG. 8E, a phase information map 64 including a plurality of arbitrary 5 × 5 pixels is extracted from the phase information map 63. In the phase information map 64 of FIG. 8E, the number in each of the 5 × 5 matrix indicates the value of D (i) of the pixel at the corresponding position.

  Hereinafter, each position of a plurality of 5 × 5 pixels is denoted as (Xi, Yi), and the corresponding density is denoted as “D (Xi / Yi)”. Xi indicates the position of the pixel in the horizontal direction by the number of pixels in the right direction (1 to 5) with the leftmost pixel as a reference. Yi indicates the position of the pixel in the vertical direction by the number of pixels in the downward direction (1 to 5) when the pixel at the upper end is used as a reference.

Next, the direction corresponding to the position (3, 3) among the plurality of 5 × 5 pixels is specified as follows. Each of the following calculation results is expressed as “R (j)”. In particular,
“D (3/3) −D (2/1)” is expressed as “R (2/1)”. “D (3/3) −D (3/1)” is expressed as “R (3/1)”. “D (3/3) −D (4/1)” is expressed as “R (4/1)”. “D (3/3) −D (1/2)” is expressed as “R (1/2)”. “D (3/3) −D (2/2)” is expressed as “R (2/2)”. “D (3/3) −D (4/2)” is expressed as “R (4/2)”. “D (3/3) −D (5/2)” is expressed as “R (5/2)”. “D (3/3) −D (1/3)” is expressed as “R (1/3)”. “D (3/3) −D (5/3)” is expressed as “R (5/3)”. “D (3/3) −D (1/4)” is expressed as “R (1/4)”. “D (3/3) −D (2/4)” is expressed as “R (2/4)”. “D (3/3) −D (4/4)” is expressed as “R (4/4)”. “D (3/3) −D (5/4)” is expressed as “R (5/4)”. “D (3/3) −D (2/5)” is expressed as “R (2/5)”. “D (3/3) −D (3/5)” is expressed as “R (3/5)”. “D (3/3) −D (4/5)” is expressed as “R (4/5)”. First, R (j) is compared with “−128”. When R (j) is smaller than “−128”, 256 is added to R (j). When R (j) is larger than “127”, 256 is subtracted from R (j).

Next, R (j) calculated as described above is applied to the following equations (4) to (11), and P1 to P8 are calculated.
P1 = R (5/3) ² + R (1/3) ² (4)
P2 = R (5/4) ² + R (1/2) ² (5)
P3 = R (4/4) ² + R (2/2) ² (6)
P4 = R (4/5) ² + R (2/1) ² (7)
P5 = R (3/5) ² + R (3/1) ² (8)
P6 = R (2/5) ² + R (4/1) ² (9)
P7 = R (2/4) ² + R (4/2) ² (10)
P8 = R (1/4) ² + R (5/2) ² (11)
The calculated P1 to P8 indicate the magnitude of the difference between the density of each of the two pixels for calculating each and the density of the pixel at the position (3, 3).

  The matrix 8 in FIG. 9A includes a plurality of 5 × 5 pixels of two pixels corresponding to two D (Xi / Yi) used when each of P1 to P8 is calculated. Indicates the position. P1 is calculated based on two D (Xi / Yi) corresponding to the pixels arranged in the right direction and the left direction as viewed from (3, 3) among the plurality of 5 × 5 pixels. P2, P3, and P4 are two D (Xi) corresponding to the pixels arranged in the diagonally lower right direction and the diagonally upper left direction as viewed from (3, 3) among the plurality of 5 × 5 pixels, respectively. / Yi). The angle with respect to the horizontal direction of the straight line connecting the two pixels for calculating each of P2, P3, and P4 increases in the order of P2, P3, and P4. P5 is calculated based on two D (Xi / Yi) corresponding to the pixels arranged in the upward direction and the downward direction as viewed from (3, 3) among the plurality of 5 × 5 pixels. P6, P7, and P8 are two D (Xi) corresponding to the pixels arranged in the diagonally upper right direction and the diagonally lower left direction as viewed from (3, 3) among the plurality of 5 × 5 pixels, respectively. / Yi). The angle with respect to the horizontal direction of the straight line connecting the two pixels for calculating each of P6, P7, and P8 decreases in the order of P6, P7, and P8.

  Next, the minimum value is extracted from the calculated P1 to P8. As is clear from the equations (4) to (11), P1 to P8 are the differences between the respective densities of the two pixels for calculating each and the densities of the pixels at the positions (3, 3). Indicates the size. Therefore, by extracting the minimum value among P1 to P8, the difference between the density of each of the two corresponding pixels and the density of the pixel at the position (3, 3) is the smallest, in other words, the change in density. Can be extracted.

  For example, when D (Xi / Yi) of each of the 5 × 5 pixels of the phase information map 64 shown in FIGS. 8E and 9B is applied, (4) to (11) By applying the equation, P1: 200, P2: 25, P3: 164, P4: 1381, P5: 1810, P6: 1517, P7: 1096, and P8: 1889 are calculated. P2 is extracted as the minimum value (25).

  Next, the direction of the straight line connecting the centers of the two pixels corresponding to any of the extracted P1 to P8 is the direction corresponding to the position of (3, 3) among the plurality of 5 × 5 pixels. Identified. For example, when P1 to P8 are calculated by applying D (Xi / Yi) of each of a plurality of 5 × 5 pixels of the phase information map 64 (FIG. 9B), and the minimum P2 is extracted, As shown in FIG. 9C, the direction 64A rotated clockwise by 0.15π with respect to the horizontal direction is specified as the direction corresponding to the position (3, 3).

When P1 is extracted as the minimum value, the horizontal direction is specified as the direction corresponding to the position (3, 3). When each of P3 to P8 is extracted as the minimum value, the directions rotated 0.25π, 0.35π, 0.5π, 0.65π, 0.75π, 0.85π clockwise relative to the horizontal direction, respectively. Is specified as the direction corresponding to the position (3, 3).

  The above processing is executed for all of the 5 × 5 plurality of pixels in which each of the 170 × 370 plurality of pixels is arranged at the position (3, 3). In this case, among a plurality of pixels of 170 × 370, 166 (= 170− (2 × 2)) × 366 (= 370− (2 × 2)) arranged inside two pixels from the top, bottom, left, and right ends. The direction is specified for each of the plurality of pixels. Further, data indicating each specified direction by a clockwise rotation angle (0 to π) with respect to the horizontal direction is specified as “direction data”. FIG. 9D is a diagram in which line segments indicating each of a plurality of specified directions are arranged at corresponding pixel positions. In order to make each line segment easy to see, line segments are arranged at intervals of 8 pixels in the X direction and the Y direction.

  As shown in FIG. 1, in the registration unit 39, the direction data specified by the direction data specifying unit 38 is registered in a fingerprint DB 60 stored in a flash memory 504 (see FIG. 11) of the authentication device 5 described later. That is, in the registration unit 39, data indicating the peak frequency and direction data are registered in the fingerprint DB 60.

<Function of Authentication Device 5>
The authentication device 5 includes a fingerprint image acquisition unit 52, a polar coordinate acquisition unit 53, a DFT calculation unit 54, an amplitude spectrum calculation unit 55, a peak frequency extraction unit 56, a phase information calculation unit 57, a direction data specification unit 58, and a verification unit 59. The processing corresponding to each functional block is executed by the CPU 501 (see FIG. 11). The authentication device 5 is provided with a fingerprint sensor 51.

  The fingerprint sensor 51 is a sensor that operates on the same principle as the fingerprint sensor 31. The size of the fingerprint sensor 51 is not particularly limited. For example, a fingerprint sensor 51 of 4 × 10 mm, 8 × 8 mm, or 10 × 10 mm can be applied. The resolution of the fingerprint sensor 51 is not particularly limited. For example, the fingerprint sensor 51 having a resolution of 508 dpi or 363 dpi can be applied. Note that the detection method of the fingerprint sensor 51 is not limited to the capacitance method, and other methods (for example, an electric field method, a pressure method, and an optical method) may be used. The fingerprint sensor 51 is not limited to a surface type, and may be a linear type.

  The fingerprint image acquisition unit 52, polar coordinate acquisition unit 53, DFT calculation unit 54, amplitude spectrum calculation unit 55, peak frequency extraction unit 56, phase information calculation unit 57, and direction data specification unit 58 are respectively included in the data creation device 3. The same functions as the fingerprint image acquisition unit 32, polar coordinate acquisition unit 33, DFT calculation unit 34, amplitude spectrum calculation unit 35, peak frequency extraction unit 36, phase information calculation unit 37, and direction data identification unit 38 can be executed. These detailed explanations are omitted.

  The collation unit 59 refers to at least one of the peak frequency and direction data registered in the fingerprint DB 60 stored in the flash memory 504 (see FIG. 11), and the fingerprint of the finger touching the fingerprint sensor 51 and the data creation device 3 is compared with the fingerprint of the finger touching the third fingerprint sensor 31. Specifically, when the peak frequency is used, the peak frequency extracted by the peak frequency extraction unit 56 and the peak frequency registered in the fingerprint DB 60 are collated. When the direction data is used, the direction data specified by the direction data specifying unit 58 and the direction data registered in the fingerprint DB 60 are collated. If the collation is successful, it is determined that the user who has a fingerprint registered in advance via the data creation device 3 and the user who has a fingerprint touching the fingerprint sensor 51 of the authentication device 5 are the same person. If the collation fails, it is determined that the user having a fingerprint registered in advance via the data creation device 3 and the user having a fingerprint touching the fingerprint sensor 51 of the authentication device 5 are different persons.

  The outline of verification using the peak frequency is as follows. Hereinafter, the four peak frequencies extracted for each of the plurality of pixels by the peak frequency extraction unit 56 are referred to as “four acquisition frequencies”. The peak frequencies registered for each of a plurality of pixels in the fingerprint DB 60 are referred to as “four registered frequencies”.

  A combination of pixels in which the four acquisition frequencies and the four registration frequencies all match is extracted. If there is no combination of pixels that match all four, a combination of pixels that match two or three of the four acquisition frequencies and the four registration frequencies is extracted. When a plurality of pixel combinations are extracted, the positional relationships of the extracted pixel combinations are compared, and the correlation of the positional relationships is obtained. The similarity is calculated based on the number of coincidence between the four acquisition frequencies and the four registered frequencies and the correlation of the positional relationship. If the calculated similarity is greater than a predetermined value, it is determined that the matching has been successful. If the similarity is less than the predetermined value, it is determined that the matching has failed. That is, the above collation method uses a peak frequency instead of each type (end point, branch point, core, delta, etc.) of a plurality of feature points in the conventional minutiae method.

  In the above case, the collation can be performed even if the fingerprint image 6 does not include feature points such as end points, branch points, cores, and deltas, as in the collation method based on the well-known minutiae method. In addition, a large number of acquisition frequencies and registration frequencies to be compared at the time of collation can be extracted from the small fingerprint image 6. Accordingly, even when the size of the fingerprint image 6 is small, more peak frequencies can be compared, so that the accuracy of matching can be improved.

  The outline of collation when using the direction data is as follows. Hereinafter, the direction indicated by the direction data specified for each of the plurality of pixels by the direction data specifying unit 58 is referred to as “acquisition direction”. The direction indicated by the direction data registered for each of the plurality of pixels in the fingerprint DB 60 is referred to as “registered direction”.

  A combination of pixels whose acquisition direction and registration direction match is extracted. When a plurality of pixel combinations are extracted, the positional relationships of the extracted pixel combinations are compared, and the correlation of the positional relationships is obtained as the similarity. If the similarity is greater than a predetermined value, it is determined that the matching has been successful, and if the similarity is less than the predetermined value, it is determined that the matching has failed. That is, the above collation method uses direction data instead of the directions of the plurality of feature points in the conventional minutiae method.

  As shown in FIG. 10, a case where an image 65 (FIG. 10B) in which the acquisition direction or the registration direction is indicated by a line segment is superimposed on the fingerprint image 6 (FIG. 10A) is taken as an example. . In this case, as shown in FIG. 10C, the line segment indicating the acquisition direction or the registration direction is in good agreement with the direction of the ridges of the fingerprint indicated by the fingerprint image 6. In addition, by comparing the registration direction and the acquisition direction as described above, feature points such as end points, branch points, and deltas are included in the fingerprint image 6 as in the matching method based on the well-known minutiae method. Even if not, verification can be performed. In addition, a large number of acquisition directions and registration directions to be compared at the time of collation can be extracted from the small fingerprint image 6. Therefore, even when the size of the fingerprint image 6 is small, more directions can be compared, so that the accuracy of matching can be improved.

  Furthermore, the success or failure of the collation may be determined by comparing an added value of the similarity calculated based on the peak frequency and the similarity calculated based on the direction with a predetermined value. Thereby, since collation can be performed based on the peak frequency and direction, the accuracy of collation can be further improved.

  The above collation method is an example. Therefore, the collation unit 59 may collate the acquired frequency and the registered frequency by different methods. Moreover, the collation part 59 may collate an acquisition direction and a registration direction by a different method.

<Electrical configuration>
The electrical configuration of the data creation device 3 will be described with reference to FIG. The data creation device 3 includes a CPU 301, a ROM 302, a RAM 303, an HDD 304, a display unit 305, an input unit 306, a communication interface (I / F) 307, and a drive device 308. The CPU 301 controls the data creation device 3. The ROM 302, RAM 303, HDD 304, display unit 305, input unit 306, communication interface (I / F) 307, and drive device 308 are electrically connected to the CPU 301. In addition, the CPU 301 is electrically connected to the fingerprint sensor 31 via a connector (not shown). The ROM 302 stores a BIOS, a boot program, and initial setting values. The RAM 303 stores various temporary data. The HDD 304 stores a program executed by the CPU 301 for controlling the data creation device 3 and an OS. The display unit 305 is an LCD. The input unit 306 is a keyboard and a mouse. The communication I / F 307 is a controller for executing communication with an external device (for example, the authentication device 5) via the cable 407. The drive device 308 can read information stored in the storage medium 308A. The CPU 301 can read out the program stored in the storage medium 308 </ b> A by the drive device 308 and store it in the HDD 304.

  The data creation device 3 may have a communication I / F (not shown) that can be connected to the Internet. The CPU 301 may receive a program stored in a server connected to the Internet via a communication I / F not shown. The CPU 301 may store the program received via the communication I / F in the HDD 304.

  The electrical configuration of the authentication device 5 will be described with reference to FIG. The authentication device 5 includes a CPU 501, a ROM 502, a RAM 503, a flash memory 504, a display unit 505, a touch panel 506, a communication I / F 507, and a fingerprint sensor 51. The CPU 501 controls the authentication device 5. The CPU 501 is electrically connected to the ROM 502, RAM 503, flash memory 504, display unit 505, touch panel 506, communication I / F 507, and fingerprint sensor 51. The ROM 502 stores a BIOS, a boot program, and initial setting values. The RAM 503 stores various temporary data. The flash memory 504 stores a program executed by the CPU 501 for controlling the authentication device 5 and an OS. The flash memory 504 stores a fingerprint DB 60 (see FIG. 1). The display unit 505 is an LCD. The touch panel 506 is provided on the surface of the display unit 505. The communication I / F 507 is a controller for executing communication with an external device (for example, the data creation device 3) via the cable 407.

<Details of processing executed by data creation device 3>
The first main process executed by the CPU 301 of the data creation device 3 will be described with reference to FIG. The first main process is started when the CPU 301 executes a program stored in the HDD 304 when activation of an application for fingerprint registration is prompted. The fingerprint sensor 31 is in a standby state in which finger contact can be detected.

  When the fingerprint sensor 31 detects a finger contact, the fingerprint sensor 31 outputs a signal that can identify a fingerprint image to the CPU 301. The CPU 301 receives a signal output from the fingerprint sensor 31. The CPU 301 acquires the fingerprint image 6 (see FIG. 2) based on the received signal (S11). Note that the processing of S11 corresponds to the function of the fingerprint image acquisition unit 32 in FIG. The fingerprint image 6 includes an image of the entire fingerprint (for example, a plurality of pixels of 200 × 400). The CPU 301 selects one of a plurality of pixels of 170 × 370 in the frame 6C (see FIG. 2) among the plurality of pixels included in the acquired fingerprint image 6 in the following order (S13).

  CPU301 selects the pixel of the position of (15,15), when performing the process of S13 first. When the CPU 301 next executes the process of S13, the Y coordinate is fixed at “15”, and the X coordinate is incremented by “1” from “15” to select the position of the pixel to be selected (Xc, 15) (Xc = 15, 16,...) And a pixel is selected. When the process is repeated and Xc becomes larger than 185, the CPU 301 adds “1” to the Y coordinate “15” and returns the X coordinate to “15”. The CPU 301 determines the position of the pixel to be selected as (Xc, 16) (Xc = 15, 16,...), And selects the pixel. The above processing is repeated until all of the plurality of 170 × 370 pixels are selected by the processing of S13.

  Note that the selection method when sequentially selecting any one of the plurality of pixels in the frame 6C is not limited to the above method, and may be another method. For example, the CPU 301 may select one pixel at a time from a plurality of pixels of 170 × 370. The coordinate value indicating the position of the selected pixel may be determined by a known random number generation algorithm. Moreover, you may select preferentially the pixel arrange | positioned in the vicinity of several feature points among several pixels.

  The CPU 301 sequentially determines the positions of 128 samples on the circumference of a circle 7 (see FIG. 3) having a radius r (= 15) centered on the pixel selected in the process of S13. The position of each of the 128 samples is a position indicated by polar coordinates (r, θ) with the selected pixel as the origin, and the declination angle θ was changed from 0 to 2π by (2π / 128) degrees. It is a position that is sometimes specified in order.

  Note that the order in which the positions of 128 samples on the circumference of the circle 7 are determined can be changed. For example, the CPU 301 may sequentially determine the positions indicated by the polar coordinates (r, θ) when the declination angle θ is rotated by 2π from any angle other than 0 as the positions of 128 samples.

  Even when the order of determining the positions of the 128 samples changes, the amplitude spectrum calculated by the process of S19 described later does not change. The reason is that the phase component is eliminated in the process of calculating the amplitude spectrum (A (n)) from Re (n) and Im (n) calculated by the DFT operation. Similarly, even when the order of determining the positions of 128 samples changes, the direction data calculated by the process of S25 described later does not change.

  The CPU 301 acquires the density corresponding to each of the determined 128 samples based on the bilinear method (S15, see FIG. 4). Note that the processing of S15 corresponds to the function of the polar coordinate acquisition unit 33 in FIG. The method for calculating the density is not limited to the bilinear method, and other methods may be used. For example, the concentration may be calculated based on the known nearest neighbor method or the bicubic method. Further, the CPU 301 may acquire the densities of the pixels arranged at the determined 128 sample positions as they are. Also, brightness may be calculated instead of density.

  The CPU 301 performs a DFT operation based on the 128 densities on the circumference of the circle 7 acquired by the processing of S15, and calculates a real part (Re (n)) and an imaginary part (Im (n)) ( S17). Note that the processing of S17 corresponds to the function of the DFT operation unit 34 in FIG. The CPU 301 calculates the amplitude spectrum by applying the above equation (1) based on the calculated Re (n) and Im (n) (S19, see FIG. 7). The CPU 301 extracts n corresponding to four peaks in descending order of amplitude from the calculated amplitude spectrum as a peak frequency (S19). Note that the processing of S19 corresponds to the functions of the amplitude spectrum calculation unit 35 and the peak frequency extraction unit 36 in FIG.

  The CPU 301 calculates phase information by applying the above equation (2) based on Re (n) and Im (n) calculated by the processing of S17 (S21). Note that the processing of S21 corresponds to the function of the phase information calculation unit 37 in FIG.

  In the above equation (2), the values of n of ΣRe (n) and ΣIm (n) for calculating the phase information may be 0 to 127. That is, based on ΣRe (n) obtained by accumulating Re (n) where n is 0 to 127 and ΣIm (n) obtained by accumulating Im (n) where n is 0 to 127, Information (B (i)) may be calculated.

  The CPU 301 determines whether or not all of the plurality of 170 × 370 pixels within the frame 6C (see FIG. 2) of the fingerprint image 6 have been selected by the process of S13 (S23). CPU301 returns a process to S13, when there exists an unselected pixel among several pixels of 170x370 (S23: NO). The CPU 301 next selects any one of the plurality of pixels that are not selected, and repeats the processes of S15 to S21. If the CPU 301 determines that all of the plurality of pixels of 170 × 370 have been selected (S23: YES), the process proceeds to S25.

  The CPU 301 calculates the density (D (i)) by applying the above equation (3) to each of the phase information calculated for each of the plurality of pixels of 170 × 370 by repeating the processing of S21. To do. The CPU 301 calculates P1 to P8 (see FIG. 8) by applying the above equations (4) to (11) to each of the calculated 170 × 370 D (i). The CPU 301 specifies the direction corresponding to each of the plurality of 166 × 366 pixels based on the calculated P1 to P8, and specifies the direction data (S25). Note that the processing of S25 corresponds to the function of the direction data specifying unit 38 in FIG.

  The method for specifying the direction data can be changed. For example, the CPU 301 may calculate a phase spectrum for each frequency component based on Re (n) and Im (n) for each of a plurality of pixels of 170 × 370 obtained as a result of the DFT calculation (S17). Next, the CPU 301 may combine the vectors corresponding to specific frequency components (for example, n = 1 to 15) among the vectors indicated by the calculated phase spectrum to obtain phase information.

  The CPU 301 associates the data indicating the peak frequency extracted by the process of S19 and the direction data specified by the process of S25 with the coordinates indicating the position of the corresponding pixel, and generates feature data (S27). FIG. 13 shows an example of the generated feature data. As shown in FIG. 13, in the feature data, the data indicating the peak frequency is associated with each of a plurality of coordinates of 170 × 370, and the direction data is associated with each of a plurality of coordinates of 166 × 366. Yes. In FIG. 13, F (X, Y) indicates four peak frequencies corresponding to the pixel at the position (X, Y). G (X, Y) indicates direction data corresponding to the pixel at the position (X, Y). Note that the direction data corresponding to the coordinates of any of 15, 16, 169, 170 and the coordinates of any of 15, 16, 369, 370 of the Y coordinate are not specified by the processing of S25. In this case, flag data “−1” indicating that the direction data has not been specified is stored as G (X, Y).

  As shown in FIG. 12, the CPU 301 adds an error detection code to the feature data generated by the process of S27 and encrypts it. The CPU 301 associates the encrypted feature data with the sensor type, the size of the fingerprint image 6, and the falsification / error detection method, and generates fingerprint data (S29). The sensor type indicates the model, size, resolution, and the like of the fingerprint sensor 31. The size of the fingerprint image 6 indicates the number of pixels of the fingerprint image 6 acquired based on the signal output from the fingerprint sensor 31. The falsification / error detection method indicates a feature data encryption method and an error detection method. The CPU 301 registers the generated fingerprint data in the fingerprint DB 60 (see FIG. 1) stored in the flash memory 504 of the authentication device 5 (S31). Note that the processing of S31 corresponds to the function of the registration unit 39 in FIG. The CPU 301 ends the first main process.

<Details of processing executed by authentication device 5>
The second main process executed by the CPU 501 of the authentication device 5 will be described with reference to FIG. The second main process is started when the CPU 501 executes a program stored in the flash memory 504 when activation of an application for fingerprint authentication is prompted. In addition, since the process of S11-S25 is the same as the process of S11-S25 of a 1st main process, description is abbreviate | omitted.

  The CPU 501 refers to the fingerprint DB 60 (see FIG. 1) stored in the flash memory 504 (S41). The CPU 501 decodes the feature data of the fingerprint data registered in the fingerprint DB 60 based on the falsification / error detection method and checks whether there is any error.

  When the CPU 501 performs collation using only the peak frequency, the CPU 501 is based on the peak frequency (acquired frequency) extracted by the process of S19 and the peak frequency (registered frequency) among the feature data of the fingerprint data registered in the fingerprint DB 60. Then, the success or failure of the collation is determined (S43). When the CPU 501 performs collation using only the direction data, the direction (acquisition direction) indicated by the direction data specified by the processing of S25 and the direction indicated by the direction data among the feature data of the fingerprint data registered in the fingerprint DB 60. Based on (registration direction), the success or failure of the collation is determined (S43). When performing the verification based on the peak frequency and direction data, the CPU 501 determines whether the verification is successful based on the verification result of the acquisition frequency and the registration frequency and the verification result of the acquisition direction and the registration direction (S43). The CPU 501 causes the display unit 505 to display an image indicating whether the verification is successful (S45). The CPU 501 ends the second main process.

  In the above description, the CPU 301 of the data creation device 3 may register the generated fingerprint data in the fingerprint DB 60 when the request data transmitted from the authentication device 5 is received via the cable 407 (see FIG. 11). . The CPU 501 of the authentication device 5 may transmit the request data to the data creation device 3 via the cable 407 after the end of the processing of S25 of the second main processing. The CPU 501 may execute the processes of S41, S43, and S45 after the fingerprint data is registered in the fingerprint DB according to the request data.

<Effect>
The CPU 301 of the data creation device 3 performs a DFT calculation based on the density of pixels corresponding to a plurality of positions on the circumference of the circle 7 centering on each of a plurality of center points 7A in the fingerprint image 6. (S17). The CPU 301 calculates an amplitude spectrum based on Re (n) and Im (n) (S19). The CPU 301 extracts four peak frequencies corresponding to each of the calculated amplitude spectra (S19). The CPU 301 registers fingerprint data including data indicating the extracted four peak frequencies in the fingerprint DB 60 (S31). As a result, the peak frequency data registered in the fingerprint DB 60 can be used as data for the CPU 501 of the authentication device 5 to perform fingerprint matching.

  In the above case, the CPU 501 can perform fingerprint collation by comparing the peak frequency registered in the fingerprint DB 60 with the peak frequency corresponding to the fingerprint image 6 acquired via the fingerprint sensor 51. In this case, the feature points of the fingerprint may not be included in the fingerprint image 6 as in the conventional minutiae method. In addition, since four peak frequencies are extracted for each of a plurality of pixels, a large number of peak frequencies can be compared. Therefore, the CPU 501 can perform fingerprint authentication with sufficient accuracy even when the fingerprint image 6 is small. Further, even if the fingerprint image 6 is displaced or rotated based on an unstable input operation in which the finger of the hand holding the authentication device 5 is brought into contact with the fingerprint sensor 51, the fingerprint authentication is performed with sufficient accuracy. It can be performed. In this way, the CPU 301 of the data creation device 3 can generate fingerprint data for the authentication device 5 to perform fingerprint authentication with high accuracy based on the peak frequency.

  The CPU 301 of the data creation device 3 calculates the amplitude spectrum and extracts the peak frequency based on Re (n) and Im (n) calculated by performing the DFT operation (S17) (S19). Therefore, the CPU 301 can appropriately extract the peak frequencies corresponding to all of the plurality of pixels in the fingerprint image 6 based on the density acquired by the process of S15.

  The CPU 301 of the data creation device 3 extracts n corresponding to four peaks in descending order of amplitude from the calculated amplitude spectrum as a peak frequency (S19). In this case, the CPU 301 can suppress the volume of fingerprint data registered in the fingerprint DB 60 as compared with a case where n corresponding to all peaks of the amplitude spectrum is extracted as a peak frequency.

  The CPU 301 of the data creation device 3 calculates phase information based on Re (n) and Im (n) calculated by performing the DFT operation (S17) (S21). The CPU 301 calculates the density (D (i)) based on each of the phase information calculated for each of the plurality of pixels of 170 × 370, and further calculates P1 to P8 (see FIG. 8). Based on the calculated P1 to P8, the CPU 301 identifies direction data corresponding to each of the plurality of 166 × 366 pixels (S25). The CPU 301 registers fingerprint data including the specified direction data in the fingerprint DB 60. The CPU 501 of the authentication device 5 can improve the accuracy of verification by performing fingerprint verification based on the peak frequency and direction data. In this manner, the CPU 301 of the data creation device 3 can generate fingerprint data for the authentication device 5 to perform fingerprint authentication with higher accuracy based on the direction data.

  The CPU 301 calculates phase information by applying equation (2). In this case, the CPU 301 can calculate the phase information more easily than the case where the combined vector is calculated based on the calculated Re (n) and Im (n) and the direction is calculated as the phase information.

<Others>
The present invention is not limited to the above embodiment, and various modifications can be made. In the above embodiment, the CPU 301 may extract only the peak frequency and not specify the direction data. The CPU 301 may create fingerprint data including only the peak frequency and register it in the fingerprint DB 60.

  In the above embodiment, the CPU 301 calculates the real part (Re) and the imaginary part (Im) by performing the DFT operation. The CPU 301 may calculate a real part (Re) and an imaginary part (Im) by performing Fast Fourier Transform (FFT) calculation instead of DFT. In this case, the number of samples K1 can be any one of “128”, “256”, “512”, “1024”, “2048”,.

  In the above-described embodiment, the CPU 301 selects a plurality of pixels included in the fingerprint image 6 one by one and sets it as the center point 7 A of the circle 7. On the other hand, the CPU 301 may select any of a plurality of positions of the fingerprint image 6 as the center point 7 A of the circle 7.

  The process of S11 is an example of the “first acquisition step” in the present invention. The process of S15 is an example of the “second acquisition step” in the present invention. The process of S19 is an example of the “amplitude calculation step” in the present invention. The process of S31 is an example of the “amplitude storage step” and the “related data storage step” in the present invention. The processing of S21 and S25 is an example of the “phase calculation step” in the present invention. The CPU 301 that performs the process of S11 is an example of the “first acquisition unit” in the present invention. The CPU 301 that performs the process of S15 is an example of the “second acquisition unit” in the present invention. The CPU 301 that performs the process of S19 is an example of the “amplitude calculating means” in the present invention. The CPU 301 that performs the process of S31 is an example of the “amplitude storage means” in the present invention.

1: Authentication system 3: Data creation device 5: Authentication device 6: Fingerprint image 7: Circle 7A: Center point 31: Fingerprint sensor 51: Fingerprint sensor 60: Fingerprint image 301: CPU
501: CPU
504: Flash memory

Claims (8)

A first acquisition step of acquiring a fingerprint image;
The respective densities of a plurality of pixels corresponding to the circumference of a predetermined radius centered on each of a plurality of arbitrary points of the fingerprint image acquired by the first acquisition step are acquired for each of the plurality of points. A second acquisition step,
An amplitude calculating step of calculating an amplitude spectrum for each of the plurality of points based on the order of arrangement of the plurality of the concentrations acquired for each of the plurality of points by the second acquiring step;
An amplitude storage step of storing a plurality of peak frequencies corresponding to each of the plurality of amplitude spectra calculated for each of the plurality of points by the amplitude calculation step in a storage unit for each of the plurality of points; Data creation program to be executed by other programs.   2. The data creation according to claim 1, wherein the amplitude calculation step calculates the amplitude spectrum by DFT processing based on the plurality of concentrations acquired for each of the plurality of points by the second acquisition step. program.   The amplitude storage step stores the predetermined number of peak frequencies in the storage unit in descending order of the intensity of the corresponding amplitude spectrum among the plurality of peak frequencies. Data creation program. Based on the order of arrangement of the plurality of concentrations acquired for each of the plurality of points in the second acquisition step on the circumference, phase information corresponding to the phase spectrum of each of the plurality of points is obtained for each of the plurality of points. A phase calculation step to calculate
A related data storing step of storing data related to the plurality of phase information calculated for each of the plurality of points in the phase calculating step in a storage unit for each of the plurality of points; The data creation program according to claim 1 or 3. Based on the order of arrangement of the plurality of concentrations acquired for each of the plurality of points in the second acquisition step on the circumference, phase information corresponding to the phase spectrum of each of the plurality of points is obtained for each of the plurality of points. A phase calculation step to calculate
A related data storing step of storing data related to the plurality of phase information calculated for each of the plurality of points in the phase calculating step in a storage unit for each of the plurality of points; The data creation program according to claim 2.   In the phase calculation step, when the cumulative value of the plurality of real parts calculated by the DFT processing is expressed as ΣRe (n) and the cumulative value of the plurality of imaginary parts is expressed as ΣIm (n), arctan (ΣIm (n) / ΣRe 6. The data creation program according to claim 5, wherein the value calculated by (n)) is calculated as the phase information. A first acquisition step of acquiring a fingerprint image;
The respective densities of a plurality of pixels corresponding to the circumference of a predetermined radius centered on each of a plurality of arbitrary points of the fingerprint image acquired by the first acquisition step are acquired for each of the plurality of points. A second acquisition step,
An amplitude calculating step of calculating an amplitude spectrum for each of the plurality of points based on the order of arrangement of the plurality of the concentrations acquired for each of the plurality of points by the second acquiring step;
An amplitude storage step of storing a plurality of peak frequencies corresponding to each of the plurality of amplitude spectra calculated for each of the plurality of points by the amplitude calculation step in a storage unit for each of the plurality of points. Characteristic data creation method. First acquisition means for acquiring a fingerprint image;
The respective densities of a plurality of pixels corresponding to the circumference of a predetermined radius centered on each of a plurality of arbitrary points of the fingerprint image acquired by the first acquisition unit are acquired for each of the plurality of points. Second obtaining means for performing,
Amplitude calculating means for calculating an amplitude spectrum for each of the plurality of points based on the order of arrangement on the circumference of the plurality of concentrations acquired for the plurality of points by the second acquiring means;
Amplitude storage means for storing a plurality of peak frequencies corresponding to each of the plurality of amplitude spectra calculated for each of the plurality of points by the amplitude calculation means in a storage unit for each of the plurality of points. Characteristic data creation device.