JP2557487B2 - Personal verification device - Google Patents

Description

DETAILED DESCRIPTION [Table of Contents] Outline Industrial field of application Conventional technology Problem to be solved by the invention Means for solving the problem Action Example Explanation of principle of the present invention (FIGS. 1 to 3) Example (Figs. 4 to 10) Effect of the invention [Overview] Regarding a personal collation device, bridges and breaks in a binary image of a fingerprint can be accurately detected,
The purpose of the present invention is to provide a personal collation device capable of appropriately removing bridges / breaks, inputting a fingerprint of a specific individual, thinning the fingerprint image, and then removing bridges / breaks on ridges. In the personal verification device that extracts a feature point including a branch point and registers and verifies the fingerprint of the individual based on the feature point, the bridge
A bridge / break removing means for removing a break is provided, and the bridge / break removing means detects a pair of branch points of a ridge line and a valley line and a pair of end points in the vicinity thereof, and connects the pair of branch points to each other. And a line connecting the pair of end points intersects in a predetermined cross shape, it is regarded as a bridge / break, and the pattern near the end point is copied to remove the bridge / break. .

[Industrial applications]

The present invention relates to a personal verification device, and more particularly, to a personal verification device for identifying an individual using a fingerprint.

With the spread of computers throughout society in recent years, public attention has been focused on how to secure safety. Many questions have been raised regarding the security of ID cards and personal identification numbers that have been used up to now as means of personal identification when entering the computer room or using terminals. On the other hand, fingerprints are "universal"
-Since it has the two major characteristics of "lifetime immutability", it is considered to be the most powerful means of personal identification, and much research and development has been conducted on a simple personal identification system using fingerprints.

[Conventional technology]

In the development of a personal identification device using fingerprints, a “real-time fingerprint sensor” that can clearly input the fingerprint image of the individual in real time, and the registered feature information and the input fingerprint image are sufficient for practical use. The key is both the "algorithm" that performs high-speed matching while maintaining high reliability.

With the "real-time fingerprint sensor", it is necessary to input a clear fingerprint image without distortion without being affected by the residual fingerprints of the previous user. The present inventor has developed a "holographic fingerprint sensor" having a new structure in which a reflective light guide plate is combined.

On the other hand, in the development of the "algorithm", it is necessary to flexibly deal with the difference in how the force is applied at the time of input and the distortion of the fingerprint image caused by moving with the finger on the glass. The present inventor has newly developed "moving wind algorithm" that has both the redundancy of the so-called "partial pattern matching method" and the structural analysis of the "feature extraction method".

As a concrete prior art by the above development, the present inventor prototyped a simple personal collation device consisting of a 16-bit personal computer with a frame memory and a matching circuit and an independently developed "holographic fingerprint sensor" to confirm the identity. We have confirmed the high-speed collation of 0.7 seconds later and about 1.5 seconds for excluding others.

By the way, when a finger is placed on the glass surface, if sweat adheres to the fingerprint, a portion of the ridge where the meat is raised may connect to the adjacent ridge to form a “bridge”. In addition, fingerprints are not always perfect and may be broken, which is called a “break”. In any case, such bridges / breaks are not true feature points, so to speak, they correspond to noise, and these bridges / breaks need to be properly removed in order to improve the accuracy of fingerprint registration and verification. There is.
For example, in the conventional method for detecting a bridge, first, there is an H-shaped ridge structure in which only the ridge is thinned and two bifurcation points (points that are three-pronged) are formed. Is regarded as a bridge.

[Problems to be Solved by the Invention]

However, in such a conventional individual collation device, when there is another branch point in the vicinity of a certain branch point in the thinned image for which feature extraction has been completed, two branch points in the vicinity of each other are connected to each other. Since the condition is adopted, there is a problem in that even if the bridge is not actually a bridge, it may be erroneously recognized as a bridge, and the accuracy of fingerprint registration and collation decreases.

Therefore, an object of the present invention is to provide a personal collation device capable of accurately detecting a bridge / break in a binary image of a fingerprint and appropriately removing the bridge / break.

[Means for solving the problem]

In order to achieve the above-mentioned object, the personal collation device according to the present invention inputs a fingerprint of a specific individual, thins the fingerprint image, and then extracts feature points including branch points while removing bridges / breaks in ridges. , A personal collation device for registering and collating the fingerprint of the individual based on the feature points is provided with a bridge / discontinuity removing means for eliminating the bridge / discontinuity, and the bridge / discontinuity removing means includes a ridge line and a valley line. A pair of branch points and a pair of end points in the vicinity thereof are detected, and if a line connecting the pair of branch points and a line connecting the pair of end points intersect in a predetermined cross shape, it is regarded as a bridge / break, Set the pattern near the end point to the bridge
It is configured to copy bridges and remove bridges and breaks.

[Action]

In the present invention, a pair of ridge and valley branch points and a pair of end points in the vicinity thereof are detected, and if a line connecting the pair of branch points and a line connecting the pair of end points intersect in a predetermined cross shape. bridge·
Considering a break, the pattern near the end point is copied to the bridge / break.

Therefore, the bridge / break is accurately detected, and the bridge / break is appropriately removed.

〔Example〕

 Hereinafter, the present invention will be described with reference to the drawings.

1 to 3 are diagrams for explaining the basic principle of the present invention, in which the hatched portion surrounded by a solid line indicates a ridge line of a fingerprint (so-called convex portion corresponding to a mountain), and is surrounded by a solid line. The white part shows a valley line (concave and corresponds to a valley). First
The figure shows the vicinity of the feature points of the fingerprint image in order to explain the bridge detection method. In this figure, B represents the branch point when the ridge is thinned (see the broken line in the figure), E represents an end point when the valley line is thinned (see the broken line portion in the figure). Now, a pair of branch points B are connected by a straight line, a pair of end points E are connected by a straight line, and both straight lines intersect in a cross shape are detected as a bridge. Further, FIG. 2 shows the vicinity of the feature points of the fingerprint image in order to explain the method of detecting the break, and in this figure, B is a branch when the valley line is thinned (see the broken line portion in the figure). A point is shown, and E is an end point when the ridge is thinned (see the broken line in the figure). Now, a pair of branch points B are connected by a straight line, and a pair of end points E are connected by a straight line, and when both straight lines intersect in a cross shape, it is detected as a break. That is,
In the present invention, unlike the conventional example in which a bridge / break is unconditionally detected from two branch points in the vicinity of each other,
Both ridges and valleys are thinned to detect pairs of end points and branch points in ridges and valleys, and to detect bridges and breaks only when the line connecting them intersects in a cross. ing. Therefore, it is possible to avoid the situation of erroneously recognizing a non-bridge / break as pointed out in the conventional problems, and to accurately detect the bridge / break.

The bridges / breaks thus detected are removed by the method shown in FIG. In FIG. 3 (a), when a bridge is detected at a point connecting the branch points B and the end points E, the end point E
A pattern surrounded by a broken line in the vicinity is cut out, and this pattern is transplanted onto the bridge as shown in FIG. 3 (b) to remove the bridge. The cuts shown in FIG. 2 can be removed in the same manner.

An embodiment will be described below based on the above-mentioned basic principle. 4 to 10 are views showing an embodiment of the personal collation device according to the present invention. First, the configuration will be described. FIG. 4 is a block diagram showing the function of the personal collation device. In this figure, the personal collation device is constructed by using, for example, a 16-bit personal computer, and is roughly classified into an image input system 1, an image storage circuit 2, and a two-pixel storage circuit. It has a function represented by the binarization circuit 3, the thinning circuit 4, the feature point detection circuit 5, the intersection angle intersection point inspection circuit 6, the neighborhood pattern transplantation circuit 7, and the operation control circuit 8.

The image input system 1 includes, for example, a fingerprint sensor of a type in which a hologram is used as a light source and a hologram and a flat light guide plate are combined, and the fingerprint is received by an image sensor and converted into image input. The image storage circuit 2 receives the image input from the image input system 1
The image is stored by performing D conversion or the like, and the stored data is taken out to the outside as needed and is also sent to the binarization circuit 3 to exchange data with the binarization circuit 3. Be seen.

The binarization circuit 3 binarizes the image data, and the thinning circuit 4
Thins ridges and valleys based on the binarized digital data. The feature point detection circuit 5 extracts the feature points from the thinned image, and the intersection / intersection point inspection circuit 6 inspects the intersection points / intersection points of the line connecting the branch points and the end points among the feature points. The neighborhood pattern transplanting circuit 7 is the intersection angle / intersection point inspection circuit 6
Based on the output of, the binarization pattern near the end point is cut out and transplanted near the bridge / break. The operation control circuit 8 registers an individual fingerprint and compares the registration data with image input data to collate the fingerprint.

The above image processing data is sent from the image storage circuit 2 or the operation control circuit 8 to an external CRT display or the like for monitoring. Also, in image processing, for example, using a numeric keypad or a mouse to create a window or feature points,
Processing necessary for searching for a branch point is performed.

 Next, the operation will be described.

FIG. 5 is a flow chart showing a program for removing bridges / breaks of a fingerprint image, among the personal identification devices.
P _{1 to} P ₁₅ indicate program steps.

First, enter the fingerprint image of the specific individual to be registered in the P _1. In this embodiment, an image size of 256 × 256 pixels is used. Then, binarizing the input image P _2. This is, for example, in order to deal with the grayscale distribution of the image due to the imbalance of finger pressing at the time of input, the image is divided into small regions of 16 × 16 pixels, and the average density of the grayscale image is obtained for each small region. It is binarized with that as a threshold. This is called local binarization. The image is decomposed into two digital values of "1" and "0" by this binarization process, but the image is output as an actual image on the CRT. Next, in P ₃ , ridge thinning processing is performed, and in P ₄ , it is determined whether or not all branch points have been investigated as feature points.
Go to P ₅ , and if YES go to P ₁₀ .

At P ₅ , as shown in FIG. 6A, the vicinity of the branch point 11 of the ridge is cut out as a dotted line frame (square area) 10 and another branch point 12 is detected in the dotted line frame 10. Then, at P ₆ , Fig. 6 (b)
As shown in, the valley line portion of this dotted line frame (square area) 10 is thinned, and the end points 13 and 14 of the valley line portion are detected. Then P ₇
Then, as shown in FIG. 6 (c), the intersection point and intersection angle of the vector connecting the branch points 11 and 12 of the ridge and the vector connecting the end points 13 and 14 of the valley are investigated, and P _{At 8} it is decided whether or not it is a bridge. Specifically, the intersection / intersection angle is investigated by checking whether or not the two vectors intersect in a cross, as shown in FIG. That is, in FIG. 7, the start point of the vector connecting the branch points 11 and 12 is B, the end point is B ', the start point of the vector connecting the end points 13 and 14 is E, the end point is E', and the intersection point of the two vectors is If C and the intersection angle are ψ, the intersection angle ψ is around 90 ° (ψ≈90 °), and the intersection point C is approximately at the center of the two vectors (BC≈CB ′, EC≈
If the conditions such as CE ') are met, it is determined to be a bridge. If it is a bridge, cut out the binarization pattern near the end point 13 at P ₉ , transplant it near the bridge, and return to P ₄ , and if not, return to P _{4 as} it is. For transplantation of the neighborhood pattern, for example, as shown in FIG. 8, a rectangular PQRS region that is parallel to the vector with the end point 13E as the center is cut out and copied so as to cover the same vector. This will remove the bridge.

On the other hand, when all branch points have been investigated in P ₄ , it is determined whether all end points have been investigated in P ₁₀ , and when NO, P ₁₁
If YES, the process of this routine ends. P ₁₁
Then, as shown in FIG. 9A, the vicinity of the end point 16 of the ridge portion is cut out as a dotted line frame (square area) 15, and another end point 17 is detected in the dotted line frame 15. Then thinning the valley portion of the dotted line frame (square region) 15 as shown in Figure 9 at P ₁₂ (b),
The branch points 18 and 19 of the valley line portion are detected. Then, examine the intersection-intersection angle between the vector connecting a ninth diagram branch point 19 between the vector connecting the valley portions of the end points 16, 17 between the ridge portion as shown in (c) in P ₁₃ Then, at P ₁₄ , it is determined whether or not there is a break. The intersection and the intersection angle are investigated by checking whether or not the two vectors intersect in a ten-character pattern, as in the case of FIG. 7 described above. When it is cut back to P ₁₀ implanted in the vicinity of the cut by cutting the binarization pattern in the vicinity of the end point 16 at P _15, returns to it P ₁₀ and if not cut. As in the case of the above-mentioned bridge transplantation, cut out a rectangular PQRS region parallel to the vector with the end point 16E as the center, and cover it with the same vector as in the case of the bridge transplantation described above. (See Fig. 10 (b)). This removes the break.

In this way, both ridges and valleys are thinned and the pattern of valleys is also examined, so bridges can be detected more accurately than in the conventional example, and non-bridges can be erroneously recognized as bridges. Disappear. In addition, a ridge break can be detected by the same algorithm, and a bridge / break can be easily removed by copying a binary pattern in the vicinity.

As a result, the accuracy of fingerprint registration and verification can be improved.

〔effect〕

According to the present invention, it is possible to accurately detect bridges / breaks in a binary image of a fingerprint and appropriately remove the bridges / breaks.

[Brief description of drawings]

1 to 3 are diagrams for explaining the basic principle of the personal collation device according to the present invention, FIG. 1 is a diagram for explaining the method of detecting the bridge, and FIG. 2 is a diagram for explaining the method of detecting the break. FIG. 3, FIG. 3 is a diagram for explaining the removal of the bridge, FIGS. 4 to 10 are diagrams showing an embodiment of the personal collation device according to the present invention, FIG. 4 is a functional block diagram thereof, and FIG. Fig. 6 is a flowchart showing the bridge / break break removal program for the fingerprint image. Fig. 6 is a diagram for explaining the detection process of the bridge. Fig. 7 is a diagram for explaining the investigation of the intersection / intersection. Fig. 8 is the bridge. FIG. 9 is a diagram for explaining the removal of the break, FIG. 9 is a diagram for explaining the detection process of the break, and FIG. 10 is a diagram for explaining the removal of the break. 1 ... Image input system, 2 ... Image storage circuit, 3 ... Binarization circuit, 4 ... Thinning circuit, 5 ... Feature point detection circuit, 6 ... Crossing angle / crossing point inspection circuit, 7 ... Neighborhood Pattern transplantation circuit, 8 ... Operation control circuit, 10, 15 ... Dotted line frame, 11, 12, 18, 19 ... Branch point, 13, 14, 16, 17 ... End point, B, B ': Branch point, E, E ': end points, C: intersection angle, ψ: intersection point.

Claims (1)

(57) [Claims] 1. A fingerprint of a specific individual is input as an image, the fingerprint image is thinned, and then feature points including branch points are extracted while removing bridges and breaks in ridges, and based on the feature points, A personal collation device for registering and collating individual fingerprints is provided with a bridge / discontinuity removing means for eliminating the bridge / discontinuity, and the bridge / discontinuity removing means comprises a pair of branch points of a ridge line and a valley line and the vicinity thereof. If a line connecting the pair of branch points and a line connecting the pair of end points intersect in a predetermined cross shape, it is considered as a bridge / break, and the pattern near the end point is regarded as a bridge / break. An individual collation device characterized in that it is configured to copy a break and remove a bridge / break.