JPH06111005A - Method and device for collating fingerprint by alignment window change system - Google Patents

Abstract

PURPOSE:To improve a collation rate by using plural alignment windows for the window alignment. CONSTITUTION:In the fingerprint collating device of a moving window system, this device is provided with a registration alignment window storage means 2, a window registration control means 4 for determining the sequence for using a prescribed number of registration collating windows as registration alignment windows, at the time of registering the registration collating window, and registering it in the registration alignment window storage means 2, and a window read-out control means 8 for outputting the registration collating window being in the next sequence from the registration alignment window storage means 2 as the registration alignment window for collating with the collating window corresponding to the registration collating window in an input fingerprint image, when a prescribed number of collations cannot be obtained by the collating means.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a positioning window changing type fingerprint collation method and apparatus for using the collation window as a positioning window when it is difficult to perform the positioning by the positioning window.

In recent years, with the introduction of computers into a wide range of social systems, it is necessary to maintain the security of those systems. For example, when entering a computer room or using a terminal device, an ID card or password that has been conventionally used as a means for confirming whether or not the person is the person still has a problem in ensuring security. Therefore, the use of a fingerprint collation device is being considered as a powerful alternative to it.
Fingerprint matching is used for fingerprints, "for everyone"
This is because it has two major characteristics, "lifelong immutability."

[0003]

2. Description of the Related Art Conventionally, fingerprint collation uses characteristic points of a fingerprint in order to effectively utilize the above-mentioned characteristics.
The characteristic point is a point (branch point) at which a ridge line formed by a series of convex portions of a fine unevenness formed on a fingertip is divided into two (branch point) or a dead end point (end point). These feature points form the two major features of the fingerprint.

In the fingerprint collation described above, it is possible to increase the speed and flexibly cope with the difference in how the force is applied when the fingerprint is input and the partial expansion and contraction of the fingerprint image caused by moving the glass with the finger attached. What can be done is a problem to be solved. In order to meet this demand, the applicant has already proposed a fingerprint registration / verification method called “moving window method”. The outline of this fingerprint registration / verification method is as follows.

In the fingerprint registration / verification method, one registration window and a plurality of verification windows that are obtained by surrounding the registration window are stored in the feature information storage unit in the registration. An example in which the one alignment window and a plurality of matching windows are shown on the input fingerprint image is shown in FIG.
Shown in. This will be described with reference to FIG. 5. The fingerprint signal from the fingerprint sensor 22 is binarized by the binarization circuit 22 and stored in the binarization memory 24. 2 of the fingerprint of the binarized memory 24
The feature image is thinned by the feature extraction unit 28 to extract feature points. The binary image in the vicinity of the extracted predetermined number of feature points is cut out as a “window” by the window cutout unit 30, and the window is registered in the feature information storage unit 32.

At the time of registration, as registration information,
The binary image in "window", the relative positional relationship between "window" and "window", the type of feature point in "window", and the like are registered in the feature information storage unit 32 and are also registered. Select one of the "windows" that is stable and relatively close to the center of the fingerprint to be the "registration registration window", and the rest as the "registration verification window".
Define as and register.

Next, in the "verification" process, the coordinates of the "registration registration window" registered in the characteristic information storage unit 32 are used by the main displacement amount changing unit 34 to read out the read address from the memory address designating unit 38. Is output, and the “registration registration window” is read from the characteristic information storage unit 32. Pixel value comparison means performs pattern matching between the "registered registration window" and the corresponding window in the input fingerprint image written in the binarization memory 24 as described above (hereinafter referred to as the input fingerprint image window). 40 (see S1 in FIG. 6),
The displacement amount indicating the difference between the input positions at the time of registration and at the time of verification is extracted. As shown in FIG. 6, the extraction of the displacement amount is performed until the candidate point is found in the collation between the “registered registration window” and the input fingerprint image (see Yes in S2 of FIG. 6). While moving the position of the "registration alignment window" by 1 or 2 dots (see S3 in FIG. 6), it is determined whether or not the movement is within a predetermined movement range (S4 in FIG. 6).
(See) Pattern matching is performed (see S1 in FIG. 6) to extract candidate points (see Yes in S2 in FIG. 6). In determining whether the movement is within a predetermined movement range,
If it is out of the range (see No in S4 of FIG. 6), the fingerprint collation for the input image is rejected and the fingerprint collation is rejected (see S5 of FIG. 6).

The displacement amount is used for matching the registration matching window with the input fingerprint image window. That is, the registration collation window moved by the displacement amount is compared with the input fingerprint image window corresponding to the registration collation window. In matching each registration matching window with the corresponding input fingerprint image window, for each window,
A displacement corresponding to the displacement amount of the registration matching window is applied to the registration matching window, and a spiral shape is formed in the pattern matching between the "registration registration window" and the corresponding registration window in the input fingerprint image. The movement is changed to the auxiliary displacement amount changing means 36.
The registration collation window is read from the feature information storage unit 32 by the memory address designating means 38 in the state of being given to the registration collation window by and the collation of the registration collation window with the corresponding collation window in the input image is performed by the pixel value. The comparison means 40 performs (S in FIG. 6).
6). This is because human fingers are very soft
This is to absorb the partial expansion and contraction of the input fingerprint image due to the difference in how the force is applied during pressing.

In the matching (pattern matching), the matching is obtained (passed) and the pixel value comparing means 4 is obtained.
When a pass signal is output from 0 (S7 in FIG. 6)
If the number of passes is larger than a predetermined threshold value (see Yes in S8 of FIG. 6), a signal that the person's confirmation is obtained is output from the pass window evaluation means 42 (see FIG. 6). See S11).

It is not the above-mentioned pass (see No in S7 of FIG. 6) or the number of passes is less than the threshold value (in S8 of FIG. 6).
If the above process is not completed for all windows (see No in S9 of FIG. 6), the characteristic information storage unit 32
Using the coordinate information about the next registered matching window registered in (see S10 of FIG. 6), the above-described processing for the matching window, that is, steps S6 to S9 is performed.

Although the above-described processing steps 6 to S8 have been carried out for all windows, if the number of passes still does not reach the threshold value (see Yes in S9 of FIG. 6), the verification processing for the person to be verified is performed. Refuse (see S12 in FIG. 6).

[0012]

[Problems to be Solved by the Invention]
The “window method” performs fingerprint collation processing as described above, but since there is only one registration registration window, if the collation fails, collation will be performed on the subsequent collation windows as well. Will not be possible. This is still a problem from the viewpoint of improving the matching rate.

The present invention was made in view of the above technical problems, and a window-changing fingerprint matching method for position alignment for improving the matching rate by using a plurality of position alignment windows for window position alignment, and The purpose is to provide the device.

[0014]

The invention according to claim 1, that is, the invention of the registration window changing type fingerprint collation method, includes a registration registration window and a registration registration window corresponding to the registration registration window in the input fingerprint image. By comparing with the registration window, the displacement amount between the registered registration window and the registration window corresponding to the registered registration window in the input fingerprint image is determined,
Relative correction according to the displacement amount is performed between a predetermined number of registration verification windows used for verification and the verification windows in the input fingerprint image corresponding to each of the registration verification windows. When a predetermined number of registration matching windows in the relevant order and each matching window in the input fingerprint image corresponding to each of the registered matching windows cannot be obtained, as a registration position matching window. By comparing the registration matching window in the next order to be used (hereinafter referred to as the temporary registration positioning window) with the registration window in the input fingerprint image corresponding to the temporary registration positioning window. , Determining a displacement amount between the temporary registration registration window and a registration window corresponding to the temporary registration registration window in the input fingerprint image, and making a relative correction according to the displacement amount. Other than the registration verification window used as the temporary registration registration window The matching process performed between the matching window in the input image corresponding to a predetermined number of registered matching window and the registration verification window and repeating in the ranking.

According to a second aspect of the present invention, that is, the invention of the registration window changing type fingerprint collation device, as shown in FIG. 1, the registration registration window corresponds to the registration registration window in the input fingerprint image. Of the registration matching window and the registration matching window in the input fingerprint image. In a fingerprint collation device that collates with the collation windows corresponding to each of the collation windows, a registration position registration window storage unit 2 and a predetermined number of registered collation windows are registered when registering the registered collation windows. Window registration control means 4 for determining the order used as the registration window and registering it in the registration registration window storage means 2, the predetermined number of registration verification windows, and the registration verification window in the input fingerprint image. The collation means 6 performs collation with the collation window corresponding to each of If not, the registration matching window in the next order in the determined order from the registration position matching window storage means 2 is used for matching with the matching window corresponding to the registered matching window in the input fingerprint image. A window read control means 8 for outputting as a registration position adjustment window is provided, and the registration position adjustment window output from the window read control means 8 is used for position adjustment corresponding to the registration position adjustment window in the input fingerprint image. It is characterized in that it is used to determine the amount of displacement from the window.

[0016]

In the invention according to claim 1, first, by comparing the registration position alignment window with the position alignment window in the input fingerprint image corresponding to the registration position alignment window, the displacement amount between them is determined. It is determined.

After the relative correction according to the displacement amount is performed between a predetermined number of registration matching windows and the matching windows in the input fingerprint image corresponding to each of the registration matching windows. , Match between the windows.

In the collation, when the collation of the predetermined number of registration collation windows and each collation window in the input fingerprint image corresponding to each of the registration collation windows cannot be obtained, the following processing is performed. A registration matching window in the next order used as a registration positioning window (hereinafter referred to as a temporary registration positioning window), and registration in the input fingerprint image corresponding to the temporary registration positioning window. After determining the displacement amount between the temporary registration registration window and the registration window corresponding to the temporary registration registration window in the input fingerprint image by collating with the registration window, the displacement amount is determined. A predetermined number of registration matching windows other than the registration matching window used as the temporary registration positioning window and matching in the input image corresponding to each of the registration matching windows. The process of performing matching with the windows is repeated in the order, and in the order, a predetermined number of registered matching windows and matching windows in the input fingerprint image corresponding to each of the registered matching windows are matched. When you get it, verify your identity with the fingerprint. When not obtained, the same processing is repeated in the further next place.

As described above, since the registration registration window used for verification is originally used as the registration registration window as the temporary registration registration window, even if the registration registration window fails to match. Even in that case, the collation process for the fingerprint is not rejected by itself, so that the fingerprint collation rate is improved.

According to the second aspect of the present invention, when registering the registration matching windows, the order of using a predetermined number of registration matching windows as the registration positioning windows is determined and stored in the registration positioning window storage means 2. register.

When a predetermined number of verifications cannot be obtained in the verification of the predetermined number of registration verification windows in the verification means 6 and the verification windows in the input fingerprint image corresponding to each of the registration verification windows. The window read control means 8 sets the registered collation window in the next order registered in the registered position alignment window storage means 2 as the collation window corresponding to the registered collation window in the input fingerprint image. It is read out as a registered registration window for position matching.

The registration positioning window is used for position matching with the matching window corresponding to the registered positioning window (which is originally used as the matching window) in the input fingerprint image, and both of them are used. The amount of displacement between windows is determined, and the amount of displacement is used for fingerprint matching by the moving window method.

The invention according to claim 2 is also different from claim 1 in the verification of a predetermined number of registration verification windows and the verification windows in the input fingerprint image corresponding to each of the registration verification windows. Only by having, it is possible to obtain substantially the same effect as the invention according to claim 1.

[0024]

FIG. 2 shows an embodiment of the invention according to claims 1 and 2. This embodiment is also a fingerprint collation device by the Mubinck window method. In this figure,
The same reference numbers as indicate the same components,
The details are omitted. In the embodiment shown in FIG. 2, the characteristic part of the invention according to claim 2 resides in the alignment window order determination part 31 connected to the window cutout part 30 and the characteristic information storage part 33, and also the evaluation of the number of passed windows. The means 43 also constitutes a characteristic part. The pass window number evaluation means 43 outputs a registration position adjustment window change signal in addition to the principal confirmation signal.

The personal characteristic information includes, for example, as in the conventional case,
It includes a name, employee number, binary image information, window coordinate information, and the like. In addition, the personal characteristic information includes position alignment window use order information. The positioning window use order information indicates the order in which the matching window is used as the positioning window in the positioning process, and the order indicates that the matching window is close to the positioning window in terms of position. The ranking is determined, or the ranking is determined from weak to strong correlation with the background image.

In FIG. 2, the feature information storage unit 33 corresponds to the registered position alignment window storage unit 2 of FIG. 1, and the feature extraction unit 28, the window image cutout unit 30, and the alignment window use order determination unit 31 are included. 1 corresponds to the window registration control means 4 in FIG.
The main displacement amount changing means 34, the sub displacement amount changing means 36, the memory address specifying means 38, and the pixel value comparing means 41 correspond to the collating means 6 in FIG. The pass window number evaluation means 43 is shown in FIG.
This corresponds to the window read control means 8 of FIG.

The operation of the apparatus having such a configuration will be described below with reference to FIGS. For convenience of illustration, step S2 of FIG. 3 represents step S8 of FIG. In the registration of the window information, the fingerprint signal read by the fingerprint sensor 20 is binarized by the binarization circuit 22 and stored in the binarization memory 24, as described in the [Prior Art] section. . The input image is thinned by the feature extraction unit 26 to extract feature points. Window extraction is performed by the window cutout unit 30 using the feature points. In the extraction of the window, that is, the registration registration window and the registration verification window, personal characteristic information is added to the window in the same manner as in the conventional case. Then, the position alignment window use order determination unit 31 assigns a position alignment window use order to each matching window.

The personal characteristic information for each of the registration position adjusting window and the registration collating window is registered in the characteristic information storage unit 33. In the fingerprint collation processing, the input image read from the binarization memory 24 according to the conventional method and the first order designated by the main displacement amount changing unit 34 and the memory address designating unit 38 from the feature information storage unit 33. The pixel value comparison means 40 performs the pattern matching according to FIG.

In the pattern matching (comparison), if alignment is possible, the registration matching window and the matching matching window in the input fingerprint image are matched as in the conventional case. Then, in the matching of the matching window, if the number of passes is larger than a predetermined threshold value (Yes in step S8 of FIG. 6).
s)), and confirm that the person to be verified is the person.

Unable to confirm the person (in step S2 of FIG. 3)
(See No)), if all of the matching windows ranked as the alignment windows are not used for alignment (see No in step S3 of FIG. 3), the pixel value comparison means 41.
Is transferred to the pass window number evaluation means 43 from.
The pass window number evaluation unit 43 transfers the alignment window change signal to the feature information storage unit 33. As a result, the process shown in FIG. 6 is repeated with the second matching window as the registered positioning window or the next matching positioning window as the positioning window being the next positioning window. If the person can be confirmed in the matching process in any of the positioning windows (see Yes in step S2 of FIG. 3), the person currently being matched is confirmed as the person. On the contrary, if all the matching windows ranked as the positioning windows are used for the positioning (see Yes in step S3 of FIG. 3), the matching process is rejected because the person cannot be confirmed. To do. In the above embodiment, the case where all the matching windows are used has been described, but when the matching windows used as the alignment windows are used in a predetermined number, it is considered that the person cannot be confirmed. The matching process may be rejected. The registered registration window and / or the registered collation window may be moved relative to the input fingerprint image window in response to the displacement amount.

[0031]

As described above, according to the present invention, since the collation window is also used as the alignment window, even if the picked-up fingerprint has a scratch or the like which is different from that at the time of registration, the conventional method is used. Compared to using one alignment window,
The matching rate can be improved.

[Brief description of drawings]

1 is a principle block diagram of an invention according to claim 2;

FIG. 2 is a diagram showing an embodiment of the invention according to claim 1 and claim 2;

FIG. 3 is a diagram showing a processing flow of an embodiment of the invention according to claim 1 and claim 2;

FIG. 4 is an explanatory diagram of a known moving window method.

FIG. 5 is a configuration diagram of a conventional fingerprint matching device.

FIG. 6 is a diagram showing a processing flow of the apparatus shown in FIG. 5;

[Explanation of symbols]

 2 Registration Positioning Window Storage Means 4 Window Registration Controlling Means 6 Matching Means 8 Window Reading Controlling Means 28 Feature Extraction Unit 30 Window Image Cutout Unit 31 Positioning Window Utilization Order Judgment Unit 33 Feature Information Storage Unit 34 Main Displacement Amount Changing Means 36 Sub-displacement amount changing means 38 Memory address specifying means 41 Pixel value comparing means 43 Pass window number evaluating means

Claims (2)

[Claims] 1. The registration registration window and the registration position in the input fingerprint image are obtained by matching the registration registration window with a registration window corresponding to the registration registration window in the input fingerprint image. The amount of displacement between the alignment window and the alignment window corresponding to the alignment window is determined, and a predetermined number of registration collation windows that are used for collation and make a relative correction according to the displacement amount are associated with each of the registration collation windows. The comparison is performed between the input fingerprint image and the matching window in the input fingerprint image, and a predetermined number of registered matching windows in the order and the input fingerprint image corresponding to each of the registered matching windows are registered. When the collation with each collation window cannot be obtained, a registration collation window in the next order used as a registration position alignment window (hereinafter referred to as a temporary registration position alignment window) and the temporary registration position. Alignment window in the input fingerprint image corresponding to the alignment window By comparison,
A displacement amount between the temporary registration position adjustment window and the position adjustment window corresponding to the temporary registration position adjustment window in the input fingerprint image is determined, and a relative correction is performed according to the displacement amount. A process for performing matching between a predetermined number of registration matching windows other than the registration matching window used as the temporary registration positioning window and the matching window in the input image corresponding to the registration matching window. A window-changing type fingerprint matching method for positioning, characterized by repeating in the order. 2. A displacement amount between a registration registration window and a registration window corresponding to the registration registration window in the input fingerprint image is determined, and a predetermined number of registration verifications are performed according to the displacement amount. In a fingerprint collation device for performing a collation between the predetermined number of registration collation windows and the collation windows corresponding to the respective registration collation windows in the input fingerprint image, a registration position registration window When registering the storage means (2) and the registration matching window, the order of using a predetermined number of registration matching windows as registration positioning windows is determined and registered in the registration positioning window storage means (2). The window registration control means (4), the predetermined number of registration matching windows, and the matching windows corresponding to each of the registered matching windows in the input fingerprint image are matched by the matching means (6). If there is not, it is determined from the registered registration window storage means (2). Window reading control means (8) for outputting the registration matching window in the next order in the order as a registration position matching window for matching with the matching window corresponding to the registered matching window in the input fingerprint image. And using the registered registration window output from the window read control means (8) for determining the displacement amount between the registered registration window and the registration window corresponding to the registered registration window. A unique window matching fingerprint matching device.