JPH02224076A - Method and device for collating fingerprint - Google Patents

Abstract

PURPOSE:To improve a collation rate by executing a collation of a collation point having a feature point and an non-feature point of an inputted fingerprint image and a collation point recorded already in a collation point record list. CONSTITUTION:Information of a non-feature point being a point in which a feature point does not exist is also utilized. For instance, a point shown by a black square is set as a non-feature point 32, and in a collating operation, by considering a position of a given non-feature point and its shift, its vicinity is searched, and it is confirmed that a feature point does not exist in its feature point search area 33. Also, as a result of search of a place corresponding to the feature point search area 33, sometimes a feature point 35 is detected in an adjacent area 36 of the non-feature point shown by a void square. In such a way, the collation rate can be improved.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to a fingerprint verification method and device.

[Conventional technology]

The principle of conventional technology regarding fingerprint verification is as follows. That is, the fingerprint pattern is composed of lines called ridges. The ridge has parts where it branches (hereinafter referred to as bifurcation points) and parts where it breaks (hereinafter referred to as end points). These branch points and endpoints are called feature points. In principle, the positions and relationships of these points in the two fingerprints match exactly.1. It is determined that they are the same when

The details of the fingerprint matching device based on this principle are published in 1986-3.
Reference can be made to Publication No. 4508. In this device, minutiae and their relationships are extracted from the fingerprint to be registered at the time of registration and are recorded. During verification, minutiae and their relationships are extracted from the input fingerprint image, and the extracted minutiae and the relationship between the minutiae points and the minutiae recorded during registration and the relationship between the minutiae points are compared. . This comparison shows that the fingerprints do not match.
The results are compared.

However, this device requires a large amount of calculation to extract feature points and relationships between the feature points from the fingerprint image.

Therefore, it is powerful for crime scanning, which uses large general-purpose computers and supercomputers to identify suspects from latent fingerprints.

However, as in entry management, an individual's fingerprint is identified at the time of verification, and the process of matching the identified fingerprint with the fingerprint given for entry is a quick process using inexpensive processing equipment with low computing power. Need to be processed on time.

Technologies that meet this demand include, for example, the paper "Fingerprint Verification Terminal for Personal Identification" by Takakazu Morita and Hiroshi Asai (IEICE Study Group Materials, AL85-86, pp. 97-104).
(1985).

In the terminal described in this document, when high-speed processing is not required during registration, a process is executed to extract minutiae and the relationship between the minutiae points, and a list of minutiae points of the fingerprint to be registered is created and recorded. . When high-speed verification is required, the presence or absence of the minutiae on the input fingerprint image is detected according to the minutiae information listed in the recorded minutiae list, and if the minutiae is detected. The relationship between the feature points and 1. The number of ridges is counted. Next, the fingerprints are compared based on the presence or absence of the minutiae and the number of ridges, and it is determined whether the fingerprints match or do not match.

In this way, in the matching process of a processing device with low computational power, it is necessary to check whether the minutiae listed in the registered minutiae list exists or not, and whether the number of ridges between two minutiae points is as described. A method is used to check whether this is the case.

[Problem to be solved by the invention]

Therefore, if a minutiae listed in the minutiae list exists and there is a number of ridges between the two minutiae points, even if there is a minutiae that should not exist, it will be treated as a match of 1-. There is a first drawback. That is, since no matching target information can be found in an area where there are no feature points, there is a drawback that there is a high possibility of erroneous recognition.

Additionally, there are other security drawbacks, such as:

In other words, if the mechanism that executes the matching process and requires the results does not have the ability to detect the feature points and count the number of ridges, and there is a device outside the mechanism that has these capabilities, the feature point Assume a situation where it is required to prevent the contents of the list from being leaked outside the organization.

As a first example, assume that a person wants to use his/her fingerprint to confirm the identity of the user in order to access an IC card.

Generally, integrated circuits built into digital cards do not have image processing capability due to various limitations. Moreover, since the external device is a small-scale terminal device, it does not have the ability to quickly create a minutiae list from the input fingerprint image.

In this situation, if the feature point list registered in the IC card is given to an external device, there is a drawback that it will no longer function as "identity verification". As a second example, assume a system in which a center that centrally manages registered feature point lists and a terminal device installed at a location where entry management is performed are connected via a communication line for entry management using fingerprints.

The terminal C in this system does not have the ability to quickly create a feature point list from an input fingerprint image. Another drawback is that it may not be possible to send the input fingerprint image to the center due to line capacity.

The main objectives of the present invention are to eliminate the above-mentioned drawbacks: 1. An object of the present invention is to provide a method and apparatus for verifying l-fold fingerprints so as to improve the verification rate.

The purpose of the pond of the present invention is to eliminate the above-mentioned drawbacks.

An object of the present invention is to provide a fingerprint verification method and device that take measures against security.

[Means to solve the problem]

A fingerprint matching method according to a first aspect of the present invention includes an input step of inputting a fingerprint image, a matching point having minutiae and non-minutiae points of the fingerprint image input in this input step, and a matching point 1/1/1 of the fingerprint image input in this input step. and a matching step of performing matching with matching points already registered in the code list.

According to a second aspect of the present invention, a fingerprint matching device includes a matching point record holding means for holding matching point information having minutiae points and non-minutiae points, and matching point information from the matching point record holding means. An IC card having a sending means for sending data to an image processing unit, and searching for a matching point in response to matching point information from the sending means of this IC card, and transmitting the search results to the IC card.
An image processing unit having a matching point search means for sending data to a card.

朲三恭ふ罷, characterized by including knits.

The fingerprint verification device according to the third aspect of the present invention is characterized in that the verification point record holding means in the IC card of the fingerprint verification device according to the second aspect is accessed by random numbers. Zuro.

According to a fourth aspect of the present invention, the lid fingerprint verification device includes the second
A UC card having a fingerprint verification device according to the aspect of the invention further includes a method in which the permissible number of transmissions of the transmission means in response to the search results from the verification point search means of the image processing unit is greater than a predetermined value. The feature is that it can be set to any value.

〔Example〕

First, we will explain how to solve the problem No. 1 shown in the background of the invention. From the viewpoint of a decrease in recognition rate, in the matching process of the processing device with poor computational ability, There is a problem in confirming only the number of ridges in between. In order to solve this problem, one of the features of the present invention is to use the information that a feature point does not exist or the relationship between any two points as information for verification.

Referring to FIG. 3A, in the registered fingerprint image, black circles indicate registered minutiae 21 used for verification.

For example, assume that a fingerprint image as shown in FIG. 3B is input. In this input fingerprint image, all the minutiae 21 in the fingerprint image shown in FIG. 3A, that is, the registered minutiae 22 shown in FIG. 3B and the registered minutiae points 22 shown in FIG. 3A, are indicated by white circles. It has feature points 23 below the registered feature points. Therefore, the third A
The fingerprint image shown in the figure and the fingerprint image shown in Figure 3B are not identical. (1)-167 In the conventional matching method, only the registered feature point 21 shown in FIG. 3A is retrieved.

Therefore, since the registered minutiae point 2 in FIG. 3A is the same as the registered minutiae point 22 in FIG. 3B, both fingerprint images are treated as the same.

The present invention is characterized in that it also uses information on non-feature points, which are points where there are no feature points.
In the fingerprint image that is equal to the fingerprint image shown in the figure, the point indicated by the black square shown in Figure 4A is the non-feature point 32.
shall be. In the matching operation, the vicinity of the given non-feature point is searched taking into account the position and displacement of the non-feature point, and it is confirmed that the feature point does not exist in the feature point search area 33.

Therefore, when a fingerprint image equal to the fingerprint image shown in FIG. 3B is given, the minutiae search area 33 shown in FIG. 4A is
The location corresponding to is searched. As a result, a region 36 near the non-feature point indicated by the white square shown in FIG. 4B
, a feature point 35 that does not exist in the fingerprint image of FIG. 4A is found. This is the fingerprint image shown in Figure 4A and the fingerprint image shown in Figure 4B.
The fingerprint images shown in the figure are not identified as the same.

Next, one embodiment of the present invention will be described in detail with reference to the drawings.

First, an example of the personal fingerprint data generation phase will be described in order to perform verification, which is a feature of the present invention.

Referring to FIG.
．． Direction detection 3. Pre-processing 17 including the steps of directional smoothing/smoothing and ridge enhancement 5; and positioning information generation 22. Feature point detection 23° Non-feature point detection 24. The registration process 20 includes the steps of relation detection 25 and matching point record creation/recording 26. Verification points also include eloquence feature points of feature points consisting of end points and branch points.

An example of this phase is the registered fingerprint image input step 21.
It starts from. In Step 21, in addition to the method of obtaining an image by applying ink to the fingertip, the SIN ratio is The ``total internal reflection method'' can be used to obtain high quality fingerprint images.

Next, in an image acquisition step 22, the fingerprint image is optically read, converted into an electrical signal, and stored in a memory as image data. The image data collected in this way is
In the direction extraction step 3, the direction of the ridge is extracted as follows. After that, (8, 8
) The sum of the absolute values of the differences between the gradation values of every three pixels and the gradation value of the center pixel on pixels arranged on a ray line showing eight directions with the pixel as the center is a two-dimensional plane as shown in Figure 6. plotted above. Next, the plotted point and the origin are connected, the connected line segment is considered a vector, and 8 vectors are synthesized, and the direction perpendicular to the direction of the resulting synthesized vector is 256
J16, which is assigned to the (0 to 255) direction, is the direction of the streamline in the local area.

Furthermore, the magnitude of this composite vector is taken as the directional stability of the local region.

Next, in direction smoothing step 4, the directions of the ridges are adjusted so that they are continuous in adjacent blocks. That is, the direction and the direction stability are obtained from the eight neighboring local regions of the direction-extracted local region using the same method as in the direction extraction step 3. Thereafter, nine composite vectors obtained from eight neighboring local regions including the central local region are resynthesized, and smoothing is performed in consideration of the direction of the neighboring regions.

Next, in a ridge emphasis step 5, the ridges are filtered based on the direction of the extracted ridges, and the ridges are emphasized. That is, for all pixels in the local area that has been subjected to direction extraction and direction smoothing, a filter according to the ridge direction as shown in FIG. 7 is applied with weighting as shown in FIG.
Depending on the directionality, 1- or more shades of light or light are emphasized.

Then, in a positioning information generation step 22, a 5×5 block with a drastic change in ridge direction is set. That is, for reference alignment between the registered fingerprint and the input fingerprint in the matching process 19 shown in FIG. 1, a 5×5 directional pattern is selected from among the 16×16 directional patterns on the registered fingerprint image.

Next, in a feature point detection step 23, the position and direction of the feature point are detected. That is, the positions of end points/branch points are detected on the binary image by scanning the binary image, which is a binary image of grayscale images of 256×256 pixels stored in the image memory, block by block. . The direction is the same as the direction found in the direction smoothing step 4 of the block to which the detected feature point belongs8.
Note that a matching point number is assigned to the detected feature point.

Next, in the non-feature point detection step 24, non-feature points are detected as follows. A non-feature point is defined as the center of a region of a predetermined size where there is no feature point. For non-feature points, the center point of an area without feature points is detected by the same operation performed in the feature point detection step 23. The direction is the same as the direction determined in the direction smoothing step 4 to which the detected non-feature point belongs. Note that matching point numbers are also assigned to detected non-feature points.

In the relation detection step 25, four quadrants 1 to 4 are determined by setting one of the matching points detected at a certain matching point i as the origin. The closest feature point in each quadrant is selected from among the detected feature points. If this feature point does not exist within the quadrant, the closest non-feature point to the feature point at the origin is determined. The feature points or non-feature points in each quadrant determined in this way are called neighboring matching points, and each matching point number i becomes the neighboring matching point number6. Ba,
The number of ridges connecting two points is the value of the relation. In addition,
In other cases, the value is "0".

After such processing, a matching point record list 2 shown in FIG.
7 is formed. This list 27 has an average of 40 1-no codes for each fingerprint.

Next, an example of the matching phase of an embodiment of the present invention will be described in detail.

In the verification phase of one embodiment of the present invention, for the purpose of personal identification, the personal fingerprint data consisting of the verification point record list 27 and alignment data 28 registered in the above-mentioned personal fingerprint data generation phase and the registrant's A one-to-one comparison with the input fingerprint input at the time of comparison is performed. Therefore, one embodiment of the present invention employs a ``top-down verification method.'' In this method, the registrant enters his or her ID number or information in its place at the time of verification.Personal index of registered fingerprint Data 27 and 28 are accessed and the registered fingerprint characteristics within the data are read.

This results in direct verification on the input fingerprint image.

An example of the verification steps according to an embodiment of the present invention is: fingerprint image input step 1; image collection 2. Pre-processing 1 with steps of direction detection 3, direction smoothing 4 and ridge enhancement 5
7; Position correction processing 18; and search matching point selection 7. Local area settings 8. Search matching point search9. Nearby matching point selection 10. Neighborhood matching check ill, ll-tion search 12. Scoring 13, number judgment 14. Scoring 15. It includes a matching process 19 having steps of threshold determination 16 and matching result output 17. The operations in fingerprint image input step 1 and preprocessing 7 are the same as those in the code registration phase described above.

In the center alignment step 6 of the position correction process 18, the 5×5 direction data pattern obtained as the alignment data 28 generated in the alignment information generation step 22 of the registration phase is moved on the direction data of the input fingerprint, Alignment is performed by searching for the best matching position.

The fingerprint image that has been aligned in this way is input to the verification process 19.

In the search and matching point selection step 7, matching point records are sequentially selected by referring to the contents in the matching point record list 27.

Assume now that matching point record 1 has been selected.

Next, in local area setting step 8, the matching position (
X, Y) is set on the input fingerprint image.

Matching point record 1 is added to the input fingerprint image set in this way.
It is detected whether or not there is a matching point corresponding to the matching point.

In the neighboring matching point selection step 10, a matching point record j having the matching point number indicated by the neighboring matching point number in the matching point record 1 selected in the selection step 7 is searched.

Next, in a search step 11, it is detected whether or not a matching point exists at the position of the input fingerprint image indicated by the matching point in this record, i. Then, in the relation search step 11, the relation field in record 1 is referenced L/, the number of ridges on the line segment connecting the two points corresponding to record 1 and record j in the input fingerprint image, and the ridge in the referenced record description. It is detected whether the number of lines is the same.

In the scoring step 13, the degree of matching is weighted based on the difference between the detected relation and the relation in the matching point record list 27. For example, when the difference is large, the score is set low, and when the difference is small, the score is set high.

In step 14 for determining the predetermined number of times, four quadrants are detected, so four times is the predetermined number of times. However, in order to expect an improvement in accuracy, a power of 4 can be set to a predetermined number of times.

In the scoring step 15, four approximate matching points are detected for one matching point serving as the origin, and the matching points serving as the origin are registered according to the predetermined summation value for weighting by the relation. A determination is made for one matching point record as to whether it matches the existing matching point.

In threshold judgment step 16, scoring step 1
The number of matching point records determined in step 4 reaches a predetermined number t7
It is determined whether or not it is. This number may be, for example, 20'' for matches or 5'' for mismatches.

In this way, in 1-, personal fingerprint data of matching points including non-minutiae points are generated and verified.

By the way, an example of a method for determining non-feature points is as follows.

The positions of the non-feature points are predetermined so as to be areas where the absence of feature points can be stably detected, and are stored together with the feature point list. At this time, the direction of the nearby ridge to be searched may also be recorded to facilitate the search.

From the viewpoint of image processing speed, the search and match processing speed for non-feature points can be executed as fast as the search and match processing speed for feature points. A second embodiment in which an example using this random number table is realized by separating the image processing example and the IC card will be described in detail below with reference to the drawings.

Referring to FIG. 9, a second embodiment of the present invention performs image processing. At step 31, a collation request is first generated (
Step 32). In response to this verification request, alignment data 28, which is part of the personal fingerprint data in IC card 30, is read out. Along with this, the image processing unit 3
1, in response to a verification request, fingerprint image 1° preprocessing 17 and position correction processing 18 are performed as in the first embodiment.

In this position correction process J8, the positioning data 28 given from the IC card 30 is used.

In this second embodiment, the IC card 3, which has the feature
In addition to the use of 0, the second feature, matching point search using random numbers, will be described in detail below.

First, the random number table 34 is accessed by the lower several bits of the timer 33 built into the IC card 30. Next, the matching point 1 record list 27 is accessed using the random number read from the random number table 34, and the matching point record is read out. In the search matching point selection step 7, the location of the matching point is determined based on the contents of the read matching point record! (X, Y)
and the direction are determined and provided to the image processing unit 31.

In the image processing unit 31, μ-
Cal area setting step 8 and search matching point searching step 9 are executed, and the type of matching point is sent to the IC card 30 as a search result. In the IC card 30, the same record as that referred to in the selection step 7 is referred to based on the search result, the position and direction of the nearby matching point are determined, and the determined position and direction are provided to the image processing unit 31. In the image processing tool I/31, based on the position and direction of the neighboring matching point, the neighboring front 8 point search step 11 and the relation searching step 12 are executed in the same way as in the first embodiment, and the search results are compiled and the matching point is The type and relay version are sent to the IC card 30.

With this method, matching can be performed without having to previously inform the image processing device of the types of matching points and relational symbols.

Note that the scoring step 132 in the IC card 30
Number of times judgment 7/Up】4. The scoring step 15 is performed in the same manner as in the first embodiment. Since the image processing unit I.31 in this second embodiment cannot know whether there is a registered feature point at the queried position, the image processing unit 31 does not know whether or not there is a registered feature point at the queried position. It is difficult to infer the content from the inquiry. However, even if it is difficult to infer from the query, if the response is combined with the result of matching, information can easily be leaked depending on the matching method. For example, by inquiring only about the - point and referring to the verification result of only its validity, it can be seen whether the response from the image processing example 31 was as described in the verification point record list.

If the verification is considered successful only when the results of the inquiry are all correct (1-), a third party eavesdropping will know the correct answers to all the inquiries. Also,
On the other hand, in terms of reliability, there is a possibility that even a legitimate person may give an incorrect answer due to noise in the input image. Therefore, in step 16 of one embodiment of the present invention, 1. A threshold is set and when the number of correct answers reaches that value, it is considered that the match has been made, and on the other hand, when the number of incorrect answers reaches the threshold17, it is concluded that the person is not the real person8.In this case, the number of correct answers or the number of incorrect answers is the threshold value. Reached 1~
If the determination is reported to the image processing example 31 at this point, it can be seen that the threshold value has been reached based on the response to the last question. That is, if it is verified, it means that the last response was the correct answer, and if it is not verified, it is known that the last response was incorrect. This leaves a problem regarding the confidentiality of the content of the feature point list regarding the last question.

Stella card 3 for a predetermined number of times in the second embodiment of the present invention
0 queries the image processing side 31 about the positions and directions of the matching points and neighboring matching points, obtains the search results, and then informs the image processing side 31 of the matching results and ends the process. However, there is no concern about wiretapping by a third party, and I4. ．． If there is no need to hide the list of matching points, it is a waste of time and communications to continue asking a certain number of questions even though the number of correct answers has reached a threshold7. When the number of incorrect answers reaches the threshold value, the matching result is sent to the image processing example 31, while if the number of incorrect answers exceeds the threshold value, a certain number of questions are continued.

〔Effect of the invention〕

The present invention has the effect of improving the matching rate because matching is performed using matching points that include not only minutiae points but also non-minutiae points.Furthermore, the present invention provides an image processing unit to have a matching point record list on the IC card side. This has the effect that it is difficult to distinguish whether a searched matching point is a feature point or a non-feature point, and the true feature point, neighboring feature points, and the state of the relation sequence cannot be recognized. Furthermore, the present invention uses random numbers to access the verification point record list on the IC card, uses a different verification code each time, and reduces the number of times the verification point information is sent to the image processing unit. By setting it to an arbitrary number greater than or equal to the value, the contents of the matching point record list can be more effectively concealed.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a first embodiment of the present invention; FIG. 2 is a diagram showing a first embodiment of the present invention;
Figures showing personal fingerprint data created in the embodiment: Figures 3A, 3B, 4A, J, and 4B are diagrams for explaining the operation of fingerprint verification; FIG. 8 is a diagram for explaining the preprocessing 17 according to the embodiment of the present invention; and FIG. 9 is a diagram showing a second embodiment of the present invention. In the figure, 1, 2]... Input step, 17
...Pre-processing, 18...Position correction processing,
19... Verification processing, 20... Registration processing 9 End Z diagram agent Patent attorney Uchi Hara Jingu Topo Mz diagram

Claims (1)

[Claims] 1. An input step of inputting a fingerprint image, and matching points having minutiae and non-minutiae points of the fingerprint image input in this input step and matching points already registered in the matching point record list. A fingerprint matching method, comprising: a matching step of performing a matching step. 2. An IC card having a matching point record holding means for holding matching point information including minutiae points and non-feature points, and a sending means for sending out the matching point information from the matching point record holding means to an image processing unit; and an image processing unit having a matching point search means for searching for a matching point in response to matching point information from an IC card sending means and sending the search result to the IC card. 3. The fingerprint verification device according to claim 2, wherein the verification point record holding means of the IC card is accessed by random numbers. 4. The IC card is characterized in that the permissible number of times of sending of the sending means in response to the search result from the matching point searching means of the image processing unit is set to an arbitrary value larger than a predetermined value. The fingerprint verification device according to item 2 of the scope.