JP2929802B2 - Fingerprint matching system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fingerprint collation system suitable for use in an entry control system for an important confidential room such as a computer room, a keyless entry system for a vehicle, and the like. The present invention relates to a fingerprint matching system that matches a fingerprint image with a registered fingerprint image.

[0002]

2. Description of the Related Art Conventionally, in a fingerprint collating system of this kind, as shown in Japanese Patent Application Laid-Open No. 63-228271, when a fingerprint image is input and photographed, the apex angle of a prism having a cross section of an isosceles triangular shape having a right-angle cross section. In a state where the finger is pressed against the surface opposite to, the light from the light source is made incident on the prism through one surface of both surfaces sandwiching the apex angle and reflected on the surface of the finger, and the top of the prism is The reflected light from the surface of the finger that exits through the other surface on both sides sandwiching the corner is CCD
In some cases, the image is received as a fingerprint image by a camera.

[0003]

However, in such a configuration, since the photographing by the CCD camera is performed by photographing the surface of the finger from an oblique direction, the surface of the finger pressed against the surface of the prism is taken. Of these, the surface part far from the CCD camera is photographed small, while the surface part close to the CCD camera is photographed large, resulting in distortion (commonly called trapezoidal distortion) in the input fingerprint image.
There is a problem that the collation accuracy is lowered. Therefore, in order to cope with such a problem, the present invention intends to perform fingerprint matching after correcting trapezoidal distortion of a registered fingerprint image and a matching fingerprint image in a fingerprint matching system.

[0004]

Means for Solving the Problems In solving the above problems, the structural features of the present invention are, as exemplified in FIG. 1, a light source, a first surface on which light is incident from the light source, and a finger pressed. A second surface that reflects the incident light from the first surface and a third surface that emits the reflected light from the second surface;
A prism having a surface, imaging means for receiving light emitted from a third surface of the prism on a light receiving surface and photographing the image as a fingerprint image representing a fingerprint of a finger, and a second surface of the prism and the imaging means A fingerprint photographing apparatus 1 for providing an optical arrangement relationship that causes trapezoidal distortion to the fingerprint image between the fingerprint image and a light receiving surface of the means; a registration means 2 for registering the fingerprint image as a registered fingerprint image; A fingerprint matching system comprising: forming means 3 for forming a fingerprint image of a finger taken by the imaging means after registration as a matching fingerprint image; and matching means 4 for matching the matching fingerprint image with the registered fingerprint image. At the vertical position of the trapezoidal distortion.
Flip and storage means 5 comprising storing a magnification representing the distortion degree of the width direction of the trapezoidal distortion that changes in a curve, and the previous SL registered image
The fingerprint image and the collation fingerprint image before registration.
Each trapezoidal distortion of the fingerprint image before formation by
And in the vertical direction based on the magnification.
Can and a correcting means 6 you correct each registration means 2
Corresponds to the fingerprint image before being registered as the registered image.
The corrected image by the correcting means 6 is registered as the registered image.
Recording means 2 for forming the collated fingerprint image
The corrected image of the previous fingerprint image by the correction means 6 is
The collation fingerprint image is formed.

[0005]

According to the present invention as described above , prior to the matching by the matching means 4, the correcting means 6
The fingerprint image and the reference before registration as a registration image
Each trapezoidal distortion of the fingerprint image before forming as a combined fingerprint image
In both the width and length directions.
Each correction is made based on the magnification. Where the magnification
Changes in a curve according to the vertical position of the trapezoidal distortion.
It indicates the degree of distortion in the width direction of the trapezoidal distortion
The fingerprint image before registration as the registration image and
The fingerprint image before forming as the collation fingerprint image,
In both cases, it is corrected as if trapezoidal distortion has been accurately removed.
You. Therefore, the registration means 2 registers the image as the registered image.
Correction image of the fingerprint image before correction by the correction means 6
Is registered as the registered image, and the forming means 2
Correction to the fingerprint image before forming as a fingerprint image
Forming the corrected image by the means 6 as the collation fingerprint image.
Then , the collation by the collation means 4 can always be performed accurately and appropriately.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 2 shows a schematic overall configuration of a fingerprint matching system according to the present invention. The fingerprint collation system includes a fingerprint photographing device S. The fingerprint photographing device S includes a prism 10 having an isosceles right triangle cross section, a scattering plate 20, and a light source 30.
And a CCD camera 40. In the fingerprint photographing apparatus S, the light from the light source 30 passes through the scattering plate 20 while the finger 50 is pressed on the surface 11 facing the apex angle (having a right angle) of the prism 10. The parallel light is incident on the prism 10 from one surface 12 of the two surfaces 12 and 13 sandwiching the apex angle and is reflected on the surface of the finger 50. The reflected light is converted from the prism 10 into parallel light through the other side surface 13. The parallel emitted light is passed through the convex lens 10a and received by the CCD camera 40 at its light receiving surface 41 to be photographed as a fingerprint image, and the photographed fingerprint image is generated from the CCD camera 40 as a pre-video signal. Has become. Note that FIG.
The symbol F indicates the focal point on the image space side of the convex lens 10a.
Further, the symbol L represents the prism 10, the convex lens 10a and C
The optical axis common to the light receiving surface 41 of the CD camera 40 is shown.

The video signal processing circuit 60 includes a CCD camera 4
The pre-video signal from 0 is converted to an NTSC video signal. The microcomputer 70 has a CPU 71. The CPU 71 includes a ROM 74, a RAM 75, and a multivalued image input / output circuit 7 via a control bus 72 and an image bus 73.
6. In cooperation with the image memory 77, the binarization circuit 78 and the memory backup circuit 79, the computer program is executed in accordance with the flowcharts shown in FIGS.
During this execution, various arithmetic processes necessary for outputting the collation determination signal are performed. However, the multi-value image input / output circuit 76
Receives the captured fingerprint image as a multivalued image in response to the NTSC video signal from the video signal processing circuit 60. The image memory 77 is a memory for storing the multi-valued image from the multi-valued image input / output circuit 76. The binarizing circuit 78 is a memory 7
7 is converted into a binary image. The ROM 74 stores the above-described computer program and initialization data in advance. The RAM 75 temporarily stores system data and fingerprint image data. The memory backup circuit 79 backs up data stored in the memory 77 after the microcomputer 70 stops operating.

In the present embodiment having the above-described configuration, when the system of the present invention is in operation, the CPU 71 of the microcomputer 70 executes the computer program in step 100 (see FIG. 3) according to the flowcharts of FIGS. Start. Thus, immediately after the fingerprint photographing apparatus S is mounted on the system of the present invention, if the correction for the mounting error due to the mounting has not been made yet, the CPU 71 determines “NO” in step 200 and executes the computer program. The process proceeds to the next trapezoidal distortion correction data calculation routine 300 (see FIGS. 3, 5, and 6). On the other hand, the determination in step 200 is “YES”
If so, the CPU 71 advances the computer program directly to step 400. Step 200
Is determined to be "NO" on the premise that the appropriate operation switch attached to the system of the present invention has not been operated immediately after the fingerprint photographing apparatus S is assembled to the system of the present invention.

When the computer program proceeds to the operation routine 300 as described above, the CPU 71
Starts the execution of the arithmetic routine 300 in step 300a. In this state, as shown in FIG. 8, if the lattice standard figure 80 is attached to the surface 11 of the prism 10 and photographed by the CCD camera 40, the photographed standard image by the CCD camera 40 becomes as shown in FIG. And
Obtained as trapezoidal Katachiibitsu standard image 90. Flop Thus trapezoidal distortion standard image 90 taken by from CCD camera 40
When output as Rebide O signals, the Purebide OH signal is output to the multi-value image input-output circuit 76 as a NTSC video signal from the video signal processing circuit 60.

Then, the CPU 71 via the control bus 72
, The multi-level image input / output circuit 76 multi-values the NTSC video signal from the video signal processing circuit 60 into trapezoidal distortion standard image multi-valued data, and the binarization circuit 78 The standard image multi-valued data is binarized to trapezoidal distortion standard image binarized data in the image memory 77 in step 31
0 is stored. Thereafter, based on the storage of the trapezoidal distortion standard image binarized data in step 310, the CPU
If 71 determines “YES” in step 320,
The CPU 71 sets the coordinates Jon to J in step 320a.
On = 0 is set, and in the next step 320b, the magnification M (Jon) is set according to the coordinates Jon = 0, the width Wi of the standard image 80, and the width Wo ( Jon) of the trapezoidal distortion standard image 90 based on Equation 1. ) Is calculated.

[0011]

## EQU1 ## where M (Jon) = Wo (Jon) / Wi where Equation 1 is derived as follows. As shown in FIG. 9A, the standard image 80 is positioned on the graphic-side coordinate plane defined by the two coordinate axes Ii and Ji orthogonal to each other, and the image-side coordinates defined by the two coordinate axes Io and Jo orthogonal to each other. As shown in FIG. 9B, a trapezoidal distortion standard image 90 is positioned on the surface. Also, the rectangular coordinates on the figure-side coordinate plane are represented by (Iim, Jin).
Assuming that the orthogonal coordinates on the image-side coordinate plane are represented by (Iom, Jon), the width of the standard image 80 is constant as Wi as shown in FIG. Has a coordinate J as shown in FIG. 9B.
It is given as Wo ( Jon) with on as a variable. Therefore, Equation 1 is stored in the ROM 74 in advance as a relational expression for specifying the magnification M (Jon). Also, a trapezoidal distortion standard image 90
In FIG. 9B, each width Wo ( Jon) of Jon =
.., 512 are stored in advance in the ROM 74 as data determined in relation to 0, 1,. Note that the magnification M
If (Jon) is expressed in relation to the coordinates Jon, FIG.
As shown by, it is specified by the curve P. In FIG. 9B, (Ci, Cj) represents coordinates on the trapezoidal distortion standard image 90 passing through the optical axis L (see FIG. 8) (hereinafter referred to as optical axis position coordinates).

Thereafter, the CPU 71 proceeds to step 320c.
At step 320a, "1" is added to Jon = 0 in step 320a, and this addition result is updated to Jon. In step 330, based on Jon = 1 <512 at the present stage, "NO" And returns the calculation routine 300 to step 320b. Thereafter, each step 320b,
In the process of repeating the calculation circulating through 320c and 330, the CPU 71 calculates each magnification M (Jon) in step 320c according to the coordinate Jon increasing by “1”. In this process, if the latest coordinate Jon = 512 is satisfied in step 320c, the CPU 71
Is determined to be “YES” in step 330, and in the next step 330a, the reference position coordinates (Io) on the image-side coordinate surface are determined.
r, Jos) is set to (256, 256) (FIG. 10).
reference). This means that the reference position coordinates (Ior, Jos)
== (256, 256) means starting distortion correction in the direction of the coordinate axis Io on the image-side coordinate plane.

Thus, the CPU 71 determines in step 330
In step b, coordinate Jon = 0 is set, and step 330c is set.
, The width correction value Whn is calculated based on the following equation (2).
(Jon) = Wo (0) and magnification M (Jon) = M
Calculate according to (0).

Whn = Wo (Jon) / M (Jon) where, Expression 2 is stored in the ROM 74 in advance. Next, the CPU 71 determines in step 330d that the coordinates Jon
= 0 is added to "1" and the result of the addition is updated to Jon. At the next step 340, "NO" is determined based on Jon = 1 <512, and the data calculation routine 300 returns to step 330c.

Thereafter, in the process of repeating the operation circulating through steps 330c, 330d and 340, the CPU
71 calculates each width correction value Whn based on Equation 2 according to the coordinates Jon and the magnification M (Jon) that increase by “1” in Step 330d. In such a process, when the latest coordinate Jon = 512 in step 330d is established, the CPU 71 determines in step 340 that “YE
S ". This means that the reference position coordinates (Ior,
Jos) = (256, 256) means that the distortion correction in the direction of the coordinate axis Io on the image-side coordinate plane has been completed. Thereby, the corrected image H as shown in FIG.
1 is obtained.

After that, the CPU 71 determines in step 340a
Sets the coordinate Jon = 0. This means that distortion correction in the direction of the coordinate axis Jo on the image-side coordinate plane is started around the reference position coordinates (Ior, Jos) = (256, 256). Thus, the CPU 71 calculates the longitudinal distortion correction value Lhn based on the following equation 3 in step 340b as follows.

Lhn = Lo (Jon) / M (Jon) where, in Equation 3, the sign Lo (Jon) is the coordinate Jo.
n represents a deviation from the reference coordinate Jos along the direction of the coordinate axis Jo (hereinafter, referred to as longitudinal distortion). This longitudinal distortion Lo (Jon)
Calculates the distortion correction of the trapezoidal distortion standard image 90 as the reference position coordinates (Io
(r, Jos). Then, each longitudinal distortion Lo (Jo
n) is determined on the trapezoidal distortion standard image 90 on the image-side coordinate plane in relation to each coordinate Jon, and is stored in advance in the ROM 74 together with the equation (3).

Then, the CPU 71 determines in step 340
At b, the longitudinal distortion correction value Lhn is calculated based on Equation 3 by using the longitudinal distortion Lo (J
on) = Lo (0) and the magnification M (Jon) = M (0). Then, the CPU 71 determines in step 34
At 0c, “1” is added to the coordinates Jon = 0, and the addition result is updated to Jon. At the next step 350, Jo is added.
It is determined as “NO” based on n = 1 <512, and the data calculation routine 300 returns to step 340b. Thereafter, in the process of repeating the calculation circulating through the steps 340b, 340c and 350, the CPU 71
c, the coordinate Jon increasing by “1” at each time, and each longitudinal distortion Lo (J
on) and the magnification M (Jon), and calculates the longitudinal distortion correction value Lhn based on Equation 3. In such a process, when the latest coordinate Jon = 512 in step 340c is established, the CPU 71 determines in step 350 that “YE
S ". This means that the reference position coordinates (Ior,
(Jos) = (256, 256) means that the longitudinal distortion correction in the direction of the coordinate axis Jo on the image-side coordinate plane has been completed. Thereby, the corrected image H as shown in FIG.
2 is obtained.

In this manner, the reference position coordinates (Ior, J) for the trapezoidal distortion standard image 90 in the respective coordinate axes Io, Jo directions.
When the enlargement correction centering on os) is completed, the CPU 71
Is the width X of each grid of the corrected image H2 in step 360.
1, X2, X3 and heights Y1, Y2, Y3 (see FIG. 10C) are calculated, and (X1 / X2), (X2 / X3), (Y1 /
Y2) and (Y2 / Y3) are calculated, and in step 370, the coordinate axes Io are calculated according to (X1 / X2) and (X2 / X3).
A deviation ΔIo between the reference position coordinates Ior of the direction and the optical axis position coordinates Ci is calculated, and (Y1 / Y2) and (Y2 /
In accordance with Y3), a deviation ΔJo between the reference position coordinate Jos in the coordinate axis Jo direction and the optical axis position coordinate Ci is calculated. Thereafter, the CPU 71 determines in step 380 that the two shifts ΔI
o, optical coordinates (Ci, Cj) are determined based on Jo,
The optical coordinates (Ci, Cj) are stored in the RAM 75 together with the respective magnifications M (Jon) in step 320b, and the computer program proceeds to step 400.

[0018] Proceeding from the computer program step 200 or data computation routine 300 as described above in step 400 (see FIG. 3), CPU 71 has, at the same step 400, it determine matching or registration of the fingerprint.
When determining that the fingerprint is to be registered, the CPU 71 advances the computer program to step 410 . If the finger 50 is pressed against the surface 11 of the prism 10 and photographed by the CCD camera 40 as shown in FIG.
The fingerprint image captured by the D camera 40 is
0 is output as Purebide O signals from. Then this
Purebide O signal is output to the multi-value image input-output circuit 76 as a NTSC video signal from the video signal processing circuit 60.

Next, the CPU 71 via the control bus 72
, The multi-valued image input / output circuit 76 multi-values the NTSC video signal from the video signal processing circuit 60 into fingerprint image multi-valued data, and the binarization circuit 78 The data is binarized and stored in the image memory 77 as fingerprint image binarized data in step 410.
Thereafter, when the computer program proceeds to the trapezoidal distortion correction routine 420 (see FIGS. 3 and 7), the CPU 71
However, in step 420a, the trapezoidal distortion correction routine 42
0, and in step 421, the image plane side coordinates J
By setting on = 0, the trapezoidal distortion correction routine 420 proceeds to step 422. Then, this step 422
At step 380, the CPU 71 sets the width Wo (Jon) = Wo under the condition of Jon = 0 based on the optical axis position coordinates (Ci, Cj) in step 380 and based on the following equation 4.
The width correction value Whnf of the fingerprint image of the fingerprint image binary data is calculated according to (0) and the magnification M (Jon) = M (0).

Whnf = Wo (Jon) / M (Jon) where Whn is replaced by Whnf in Expression 2 and stored in the ROM 74 in advance.

Then, the CPU 71 adds "1" to the coordinates Jon = 0 at step 423 and updates the result of the addition to Jon. At the next step 424, Jon = 1.
Based on <512, “NO” is determined, and the keystone distortion correction routine 420 is returned to step 422. Hereafter, each step 4
In the process of repeating the calculations circulating through 22, 423, and 424, the CPU 71 determines the width in accordance with the coordinate Jon, each width Wo (Jon), and each magnification M (Jon) that increase by "1" in step 423, based on the mathematical expression 4. Correction value Whnf
Is calculated. In such a process, step 340
When the latest coordinate Jon = 512 in c is established, C
The PU 71 determines “YES” in step 424. This means that the lateral distortion correction in the coordinate axis Jo direction of the fingerprint image has been completed with the optical axis position coordinates (Ci, Cj) in step 380 as the center.

Thereafter, in step 425, the CPU 71 sets the coordinates Jon = 0. This means that the coordinate axis J of the fingerprint image is centered on the optical axis position coordinates (Ci , Cj ).
This means that distortion correction in the o direction is started. Then
In step 426, the CPU 71 sets the vertical distortion correction value Lhnf of the fingerprint image to the vertical distortion Lo (Jon) =
Calculation is performed according to Lo (0) and magnification M (Jon) = M (0).

Lhnf = Lo (Jon) / M (Jon) where Lhn is Lhnf in Expression 3 and is stored in the ROM 74 in advance.

Next, at step 427, the CPU 71 adds "1" to the coordinates Jon = 0, updates this addition result to Jon, and at the next step 428, Jon = 1.
Based on <512, “NO” is determined, and the trapezoidal distortion correction routine 420 is returned to step 426. Hereafter, each step 4
In the process of repeating the calculations circulating through 26, 427, and 428, the CPU 71 determines in step 427 the coordinate Jon that increases by “1”, each longitudinal distortion Lo (Jon), and each magnification M (Jon) based on Equation 3. Vertical distortion correction value Lhnf
Is calculated. In such a process, step 427 is performed.
When the latest coordinate Jon = 512 is established, CP
U71 determines "YES" in step 428.
This means that the vertical distortion correction in the direction of the coordinate axis Jo of the fingerprint image around the optical axis position coordinates (Ci, Cj) has been completed. In this manner, the trapezoidal distortion correction routine 420
Is completed, in step 430, the CPU 71 extracts a registered image area from the final corrected fingerprint image in the trapezoidal distortion correction routine 420 , and in step 440, extracts the extracted registered image area. The image is stored in the image memory 77 as a registered fingerprint image.

After the registration of the registered fingerprint image, if it is determined in step 400 that the fingerprint verification is to be performed, the CPU 71 advances the computer program to step 450. As described above, as shown in FIG. 2, when the finger 50 is pressed against the surface 11 of the prism 10 and photographed by the CCD camera 40, the fingerprint image photographed by the CCD camera 40 is output from the CCD camera 40.
It is output as Purebide Oh signal. Then, this Prebi
De O signal is output to the multi-value image input-output circuit 76 as a NTSC video signal from the video signal processing circuit 60.

Next, the CPU 71 via the control bus 72
, The multi-valued image input / output circuit 76 multi-values the NTSC video signal from the video signal processing circuit 60 into fingerprint image multi-valued data, and the binarization circuit 78 The data is binarized and stored in the image memory 77 at step 450 as fingerprint image binarized data.
Thereafter, when the computer program proceeds to the trapezoidal distortion correction routine 460, the CPU 71 executes the trapezoidal distortion correction routine 42 in the trapezoidal distortion correction routine 460.
Keystone distortion correction of the fingerprint image similar to 0 is performed to obtain a collation fingerprint image.

Then, the CPU 71 determines in step 470
At step 480 (see FIG. 4), the registered fingerprint image is read from the image memory 77, and the matching fingerprint image and the registered fingerprint image are aligned at step 480.
Perform pattern matching in the same alignment state,
In step 500, the number of unmatched pixels after pattern matching is measured. When the measurement of the number of mismatched pixels is completed in this way, the CPU 71 determines in step 510 that the matching fingerprint image and the fingerprint registration image are matched or mismatched in relation to the measured mismatched pixel number. I do. If it is determined that they match, the CPU 71 proceeds to step 5
At 20, a coincidence output signal is output. On the other hand, if it is determined that they do not match, the CPU 71 outputs a mismatch output signal in step 530.

As described above, in the trapezoidal distortion correction data calculation routine 300, the respective magnifications M (Jon) and the optical axis position coordinates (Ci, Cj) required for correction of the trapezoidal distortion are calculated and obtained as correction data. Based on the correction data, the trapezoidal distortion correction routines 420 and 460 perform correction so as to remove the trapezoidal distortion of each fingerprint image necessary for registration and collation. Without knowing the mounting angle of the prism 40, trapezoidal distortion, which includes the mounting error of the prism 10 and the CCD camera 40 and is necessarily mixed in the registered fingerprint image and the collation fingerprint image, can be corrected with high accuracy. As a result, fingerprint collation is performed after reliably eliminating an error inherent in the fingerprint photographing apparatus S, and the fingerprint collation accuracy can be improved. In practicing the present invention, if each error of the mounting angle α of the prism 10 and the mounting angle β of the CCD camera 40 can be neglected, the prism 1
The magnification or the optical axis position can be calculated by assembling the camera such that the focus is 0 or the CCD camera 40 is focused. Therefore, the trapezoidal distortion correction data calculation routine 30
0 may be omitted.

[Brief description of the drawings]

FIG. 1 is a diagram corresponding to the description in the claims.

FIG. 2 is a block diagram showing one embodiment of the present invention.

FIG. 3 is a part of a flowchart showing an operation of a CPU in FIG. 2;

FIG. 4 is a remaining part of the flowchart.

FIG. 5 is a first part of a detailed flowchart of a trapezoidal distortion correction data calculation routine of FIG. 3;

FIG. 6 is a latter part of the detailed flowchart.

FIG. 7 is a detailed flowchart of a trapezoidal distortion correction data calculation routine of FIG. 3;

FIG. 8 is a diagram schematically showing a standard figure and its trapezoidal distortion standard image in relation to a fingerprint photographing apparatus.

FIG. 9 is a diagram illustrating a standard graphic figure-side coordinate plane, a diagram illustrating a trapezoidal distortion standard image on an image-side coordinate plane, and a graph illustrating a magnification with respect to a coordinate axis Jo direction.

FIG. 10 is a diagram showing a trapezoidal distortion standard image on an image-side coordinate plane. FIG.
FIG. 6 is an explanatory diagram showing a case where correction is performed in the direction, and a diagram showing a case where the trapezoidal distortion standard image is corrected in the direction of the coordinate axis Jo with reference to the reference position coordinates.

[Explanation of symbols]

S: fingerprint imaging device, 10: prism, 30: light source, 40
... CCD camera, 70 ... microcomputer.

────────────────────────────────────────────────── ─── Continuation of front page (56) References JP-A-2-176984 (JP, A) JP-A-3-91877 (JP, A) (58) Fields investigated (Int.Cl. ⁶ , DB name) G06T 7/00 G06T 1/00

Claims (1)

(57) [Claims] 1. A light source, a first surface to which light is incident from the light source, a second surface to which a finger is pressed to reflect incident light from the first surface, and a reflected light from the second surface. A prism having a third surface from which light is emitted, and imaging means for receiving light emitted from the third surface of the prism on a light receiving surface and photographing the image as a fingerprint image representing a fingerprint of a finger; A fingerprint photographing apparatus that provides an optical arrangement relationship that causes trapezoidal distortion in the fingerprint image between a surface and a light receiving surface of the imaging unit; a registration unit that registers the fingerprint image as a registered fingerprint image; Forming a fingerprint image of a finger photographed by the imaging unit as a collation fingerprint image after registration by a collation fingerprint image, and a collation unit performing collation between the collation fingerprint image and the registered fingerprint image. You The trapezoidal distortion changes in a curved shape according to the vertical position of the distortion.
A storage means for storing a magnification representing the distortion degree of the width direction of the trapezoid distortion, the fingerprint image and the before registering as before Symbol registered image
Each trapezoid of the fingerprint image before forming as a collation fingerprint image
Eliminate distortion in both width and length directions
Based on the magnification and a correction means that to correct each
The registration unit may be configured to register the image before registering the image.
The registered image of the fingerprint image by the correction means is registered.
Registered as a recorded image, and the forming unit is configured to form the image before forming the collated fingerprint image.
The corrected image of the fingerprint image by the correcting unit is
Fingerprint matching system comprising a forming child as serial collation fingerprint image.