JPH0821083B2 - Fingerprint matching device and fingerprint matching method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a fingerprint collation device and a fingerprint collation method for personal identification.

(Prior Art) With the spread of various types of cards, IC cards are entering the practical period. When using the IC card, the personal identification number is used to confirm the individual. However, it has a problem that it is easy to forget the personal identification number and that it may be stolen by others.

Therefore, in recent years, fingerprint information representing the physical characteristics of the card user is registered in the IC card instead of the personal identification number,
A method has been proposed in which a fingerprint collation device as a terminal is used to collate a fingerprint each time it is used to confirm an individual.

In this way, the fingerprint information is used instead of the personal identification number to enhance the security of the IC card, and the fingerprint information is further improved to protect the privacy of the fingerprint and further improve the card security including the alteration of the fingerprint data. An IC card system has been proposed that has been encrypted in real time and realized in near real time at the microcomputer level.

Actually, the position and direction data of the end points (points at which the ridges terminate) and branch points (points at which the ridges branch) of the fingerprint ridges (contact portions) are used as feature point data and random numbers are used. Use it to encrypt and register in the IC card memory. When using an IC card, as shown in FIG.
Is printed, taken by the TV camera 3, and taken into the computer 5 via the A / D converter 4, and the computer 5
The processing is carried out in step 1 to extract the characteristic points. The data of this characteristic point and the IC read from the IC card reader / writer 10.
Matches the data in the card with template matching independent for each end point and branch point. In this case, thinning and matching are performed by parallel processing using the two micro CPUs 6 and 7. 8 is a bus selection circuit,
Reference numeral 9 is a shared memory (RAM).

FIG. 6 is an overall schematic processing flowchart of the fingerprint collation.

First, input a rich flame image to the main CPU (step), then perform thinning by parallel processing of the main CPU and sub CPU (step), then perform feature point extraction by the main CPU (step), The registered fingerprint data in the IC card is collated (step), and it is determined whether they match (step).

Next, FIG. 7 shows an end point processing flowchart of the fingerprint collation. That is, (1) First, in the input fingerprint, the shared feature point and the direction difference are ±
One is selected and it is set as a corresponding point candidate (step).

(2) Next, the shared feature point is overlaid on each corresponding point candidate, and other feature points are moved by the amount of movement (step).

(3) Next, the matching feature points are searched for and a score is given (step).

(4) Next, the true corresponding point of the corresponding point candidate having the highest score is set (step).

Such processing is performed, and the same processing is further performed for the branch points.

In that case, as shown in Fig. 5, the end point processing is the main CP.
U6 and branch point processing are separately processed in the sub CPU 7.

(Problems to be Solved by the Invention) However, in the above-described conventional fingerprint collation device, when the registered fingerprint and the input fingerprint are collated, the end points and the branch points thereof are processed separately. Although they have common processing data, they are handled individually, the CPU throughput is reduced, and there is a problem that it is difficult to avoid mutual confirmation due to lack of synchronization.

(Means for Solving the Problems) In order to achieve the above-mentioned object, the present invention (1) extracts a feature point from an input fingerprint, stores position data and direction data of the feature point, and stores it as a registered fingerprint. In a fingerprint collation device that performs collation and performs feature point extraction processing using two central processing units, (a) extracting end points and branch points as feature points from an input fingerprint, and performing the end point processing and the branch point processing. A first central processing unit that performs position data movement processing in processing one of the first characteristic points; and (b) executes processing of second characteristic points that are not processed by the first central processing unit. In doing so, a second central processing unit to which the movement amount of the position data of the first characteristic point is transmitted from the first central processing unit is provided, and (c) the first central processing unit is Before according to the movement amount of the position data of the first feature point The second central processing unit performs matching with the second central processing unit, and the second central processing unit corresponds to the second amount corresponding to the movement amount of the position data of the first feature point transmitted from the first central processing unit. Performing the movement processing of the feature points, and after the movement processing of the position data of the first and second feature points by the first and second central processing units, the first central processing unit Rotation processing of the direction data of the feature points, and the second central processing unit performs the rotation processing of the direction data of the second feature points, respectively, to find the number of matching feature points corresponding to the registered fingerprint. It is configured as follows.

(2) Fingerprint matching in which a feature point is extracted from an input fingerprint, position data and direction data of this feature point are stored, a match with a registered fingerprint is performed, and feature point extraction processing is performed using two central processing units. In the method, (a) an end point and a branch point as feature points are extracted from the input fingerprint, and a position in the processing of the first feature point of either one of the end point processing and the branch point processing is performed by the first central processing unit. Move data,
(B) When executing the processing of the second feature points that are not processed by the first central processing unit, the first central processing unit transfers the position data of the first characteristic points to the second central processing unit. Of the position data of the first characteristic point of the first central processing unit,
The second central processing unit performs a movement process of the position data of the second characteristic points corresponding to the movement amount of the position data of the first characteristic points transmitted from the first central processing unit, The first central processing unit executes the rotation processing of the direction data of the first feature points, the second central processing unit performs the rotation processing of the direction data of the second feature points, and the registered fingerprint. The above steps for obtaining the number of matching feature points are sequentially performed.

(Operation) According to the present invention, since it is configured as described above, in the fingerprint collation device using the feature point extraction method, it is common to perform end point processing and branch point processing when collating a registered fingerprint and an input fingerprint. The processed data can be effectively used to improve the CPU throughput, and the end point processing and the branch point processing can be synchronized to avoid mutual erroneous confirmation.

(Example) Hereinafter, the Example of this invention is described in detail, referring drawings.

FIG. 1 is a schematic configuration diagram of a fingerprint collation device to which the present invention is applied, FIG. 2 is a flow chart showing a fingerprint collation method of the present invention, FIG. 3 is an explanatory diagram of extraction of a fingerprint image and its characteristic points, and FIG. The figure shows a matching method for fingerprint matching.

The fingerprint collation apparatus shown in FIG. 1 is configured to be able to send the command signal Sc for transmitting the data of the movement amount and the rotation amount between the main CPU 6 and the sub CPU 7 which will be described later. Further, the main CPU 6 is provided with a CRT 12 with an input / output device via an interface 11. Other points are the above-mentioned fifth.
It is similar to that in the figure.

Originally, the fingerprint collation device relies on position matching by template matching between the input fingerprint and the feature point data of the registered fingerprint. Hereinafter, this point will be described in detail.

First, as shown in FIG. 3, end points (points at which ridges terminate) a and branch points (points at which ridges branch) b are extracted from the input fingerprint, and their position data x, y is read into memory.

Further, as shown in FIGS. 3B and 3C, the direction data of the end point a and the branch point b are read into the memory as quantized data of eight directions.

Therefore, this data is made to correspond to the registered fingerprint, and four feature points and four edge points and four branch points are selected as feature points in order of decreasing distance from the processing screen center, and a total of eight feature points are selected as common point candidates.

Next, select a direction difference of ± 1 from the common point candidate,
It is set as a corresponding point candidate (see FIG. 4). Where ○,
● indicates the positions of the registered fingerprints registered in the IC card and the feature points (end points) of the input fingerprints. △ and ▲ indicate the positions of the registered fingerprints registered in the IC card and the feature points (branch points) of the input fingerprints. And the line segment extending from it indicates the direction.

Next, in order to perform position matching, the registered fingerprint is moved in parallel so that one shared feature point (common point candidate) selected now and its corresponding feature point (corresponding point candidate) overlap.

Further, in order to correct the deviation in the rotation direction, the registered fingerprint is rotated by 5 degrees up to ± 20 degrees around the overlapped point, and the number of feature points whose position and direction match each time is obtained.

Therefore, the corresponding point candidate having the highest score is set as the true corresponding point.

In this case, if the objects belong to the same person, the end point a shown in FIG.
Focusing on the fact that the position and the direction of the branch point b and the branch point b have a constant relationship, in the present invention, the movement processing in the end point processing is performed by the main CPU 6 as shown in FIG. The data is notified to the sub CPU 7 as movement amount data in the branch point processing. Therefore, the sub CPU 7 searches for corresponding points using the movement amount data. Similarly,
The rotation process of the registered fingerprint is also performed based on the direction data already obtained, the score is calculated and the determination is performed, and the true corresponding point is determined.

Therefore, as shown in FIG. 1, from the main CPU 6,
A command signal Sc of movement amount data is transmitted to the sub CPU 7, and the main CPU 6 and the sub CPU 7 are configured to perform movement amount processing and rotation processing in synchronization with each other.

The present invention is not limited to the above-mentioned embodiment, and various modifications can be made based on the spirit of the present invention.
They are not excluded from the scope of the present invention.

(Effects of the Invention) As described in detail above, according to the present invention, the following effects can be achieved.

(1) In the conventional feature point extraction method, the end points and the branch points of the fingerprint are matched separately, but in the present invention, the movement amount and the rotation amount in that case are shared between the CPUs to be synchronized. With this, it is possible to perform highly accurate fingerprint collation with very few misrecognitions.

(2) The amount of data processed by the CPU per unit time, that is, the throughput can be improved.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a fingerprint collation device to which the present invention is applied, FIG. 2 is a flow chart showing a fingerprint collation method of the present invention, FIG. 3 is an explanatory diagram of extraction of a fingerprint image and its characteristic points, and FIG. The figure shows a matching method for fingerprint matching,
FIG. 6 is a schematic configuration diagram of a conventional fingerprint collation device, FIG. 6 is an overall schematic processing flowchart in conventional fingerprint collation, and FIG.
The figure is a flow chart of the end point processing of the conventional fingerprint matching. 1 ... Prism, 2 ... Finger, 3 ... TY camera, 4 ... A /
D converter, 5 ... Computer, 6 ... Main CPU, 7 ...
... Sub CPU, 8 ... Bus selection circuit, 9 ... Shared memory (R
AM), 10 ... IC card reader / writer, 11 ... interface, 12 ... CRT with input / output device.

Claims (2)

[Claims] 1. A feature point is extracted from an input fingerprint, position data and direction data of the feature point are stored, collation with a registered fingerprint is performed, and feature point extraction processing is performed using two central processing units. In the fingerprint collation device, (a) the end point and the branch point as the feature points are extracted from the input fingerprint, and the movement processing of the position data in the processing of the first feature point of either the end point processing or the branch point processing is performed. A first central processing unit to perform, and (b) a position of the first central processing unit from the first central processing unit when executing processing of the second characteristic point that is not processed by the first central processing unit. A second central processing unit to which the amount of movement of data is transmitted; (c) the first central processing unit adjusts the amount of movement of position data of the first feature point to the second central processing unit. The second central processing unit is matched with the processing unit. In,
The movement processing of the second feature point corresponding to the movement amount of the position data of the first feature point transmitted from the first central processing unit is performed, and the first and second central processing units perform the movement processing. After the movement processing of the position data of the first and second feature points, the first central processing unit rotates the direction data of the first feature point, and the second central processing unit performs the rotation processing of the second central processing unit. 2. A fingerprint collation device, characterized in that the direction data of the two feature points is respectively rotated to correspond to the registered fingerprint and the number of matching feature points is obtained. 2. A feature point is extracted from an input fingerprint, position data and direction data of the feature point are stored, collation with a registered fingerprint is performed, and feature point extraction processing is performed using two central processing units. In the fingerprint collation method, (a) an end point and a branch point as feature points are extracted from the input fingerprint, and a first central processing unit processes one of the end point processing and the branch point processing of the first feature point. (B) in executing the processing of the second feature point that is not processed by the first central processing unit, from the first central processing unit to the second central processing unit. The amount of movement of the position data of the first feature point can be transmitted, and (c) the amount of movement of the position data of the first feature point of the first central processing unit is matched to the second central processing unit. By the transmission from the first central processing unit Wherein the first
Performing the movement processing of the position data of the second feature point corresponding to the movement amount of the position data of the feature point, and (d) the rotation processing of the direction data of the first feature point by the first central processing unit. (E) The second central processing unit executes the rotation processing of the direction data of the second feature points, and (f) Corresponds to the registered fingerprint to obtain the number of matching feature points. A fingerprint matching method, characterized in that steps are sequentially performed.