JPH05205034A - Fingerprint collating device - Google Patents

Abstract

PURPOSE:To provide a fingerprint collating device which can quickly perform the matching of position between fingerprints collated with each other. CONSTITUTION:A fingerprint collating device consists of a fingerprint sensor 1 which reads the fingerprint images, a binarization processing part 10 which applies the binarization processing, the line thinning processing, etc., to the fingerprint image read by the sensor 1, a feature extracting circuit 20 which extracts the feature points out of the part 10, a window selecting circuit which receives the output of the circuit 20 and segments the feature points and their peripheries as the windows, a position matching pattern extracting circuit which tracks the lines out of the fingerprint images read out of the part 10 to calculate the curvature of each fingerprint image and stores such a line part with which the curvature satisfies the prescribed conditions as the shape information on the fingerprints, and a fingerprint dictionary output circuit 22 which outputs the outputs of both circuits 21 and 30 as fingerprint dictionary.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fingerprint collation device.
With the spread of electronic computers in society in recent years, public attention has been focused on how to secure security. I have been used so far as a means of verifying identity when entering a computer room or using a terminal.
Many questions have been raised regarding the security of D cards and personal identification numbers. On the other hand, fingerprints are "universal",
Since it has the two major characteristics of "lifetime immutability", it is considered to be the most effective means for personal identification, and much research and development has been conducted on a simple personal identification system using fingerprints.

[0002]

2. Description of the Related Art FIG. 11 is a conceptual diagram showing the construction of a conventional fingerprint collation device. First, the operation at the time of registration will be described. The fingerprint sensor 1 is pressed against a finger to detect a fingerprint pattern, and is converted into digital data by an A / D converter (not shown) in the fingerprint sensor 1. The converted digital data (fingerprint data) is processed by the subsequent binarization circuit 2.
It is converted into binary data of “0” and “1” and stored in the binarization memory 3.

The fingerprint data stored in the binarized memory 3 is sequentially read out and then the characteristic information is extracted. Here, the characteristic information means, for example, a branch point as shown in FIG. 12A, an end point as shown in FIG. The fingerprint can be specified by the number of such branch points and end points at which positions. The extracted feature information is stored in the fingerprint dictionary storage unit 5
Stored in. The above operation is repeated for a plurality of individuals, and the characteristic information of the individuals is stored in the fingerprint dictionary storage unit 5.

In this case, the characteristic information is extracted as a window including each characteristic point and its vicinity. The extracted window and its coordinates are stored in the fingerprint dictionary storage unit 5. FIG. 13 is an explanatory diagram of a window registration method. When registering the fingerprint dictionary, the characteristic information is extracted from the input image A and registered as a plurality of windows (1 to 6) and a positioning window (thick □ window in the figure) as shown in the figure. A window near the center of the image is selected as the alignment window.

When the registration of the characteristic information in the fingerprint dictionary storage unit 5 is completed in this way, the collation operation of the individual fingerprint is started this time. In the case of collation, a finger (a finger used for registration in advance; for example, an index finger) is placed on the fingerprint sensor 1, and then an ID number of the user is input with a ten-key pad (not shown). As a result,
The collation unit 6 can narrow the search range by determining the search range of the fingerprint dictionary storage unit 6 based on the ID number and reading it out at the time of matching.

Similar to the registration, the fingerprint pattern of the input image B as shown in FIG. 12 is input and converted into digital data by the A / D converter in the fingerprint sensor 1. The converted digital data is converted into binary data of “0” and “1” by the subsequent binarization circuit 2 and stored in the binarization memory 3.

The collation unit 6 reads out the fingerprint image for collation stored in the binarization memory 3 and the characteristic information of each individual stored in the fingerprint dictionary storage unit 5, and collates both (pattern matching). I do.

Specifically, as shown in FIG. 14, the input image B is scanned by the alignment window, and the portions where the patterns match are used as alignment candidate points. In the figure, points A, B, and C are candidate points. A registration window is finally selected from these candidate points. Next, the matching unit 6 calculates a movement vector between the position matching window and each candidate point, thereby moving each matching window in parallel, further scanning the periphery of the moving destination, and pattern matching at each position. To perform.

If the degree of non-coincidence in the pattern matching is equal to or less than the threshold value, if the pattern matches and a certain number or more of the patterns in the entire matching window, it is determined that the person is the true person (referred to as moving window method).

[0010]

In the above-mentioned conventional apparatus, since many similar patterns exist, it is necessary to check many candidate points before reaching the matching position. There was a problem that it took time.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a fingerprint collation device that can quickly perform position alignment during fingerprint collation.

[0012]

FIG. 1 is a block diagram showing the principle of the present invention. The same parts as those in FIG. 11 are designated by the same reference numerals. In the figure, 1 is a fingerprint sensor that reads a fingerprint image, 10 is a binarization processing unit that performs binarization processing, thinning processing, and the like on the fingerprint image read by the fingerprint sensor 1.
Reference numeral 20 is a feature extraction circuit that extracts feature points from the binarization processing unit, 21 is a window selection circuit that receives the output of the feature extraction circuit 20 and cuts out the feature points and their vicinity as windows, and 30 is the binarization unit. An alignment pattern extraction circuit for tracing the line from the fingerprint image read from the processing unit 10 to calculate the curvature of the fingerprint image, and storing the line portion where the curvature satisfies a predetermined condition as fingerprint shape information, 22 Is a fingerprint dictionary output circuit for outputting the outputs of the window selection circuit 21 and the alignment pattern extraction circuit 30 as a fingerprint dictionary.

[0013]

In many cases, the center of the fingerprint image has a large curvature. Therefore, if the fingerprint image of the area having the large curvature is stored in the fingerprint dictionary, the position of the fingerprint image for the collation can be swiftly adjusted by first aligning the fingerprint image with the fingerprint image for collation. it can.

[0014]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 2 is a configuration block diagram showing an embodiment of the present invention. The same parts as those in FIG. 1 are designated by the same reference numerals. In the figure, the binarization processing unit 10 includes a fingerprint image storage unit 11, a binarization circuit 12, a binarized image storage unit 13, a thinning circuit 14, and a thinned image storage unit 15.
The operation at the time of registration of the device configured as described above will be described below.

The fingerprint sensor 1 reads a fingerprint image, converts it into digital data (multi-gradation data), and stores it in the fingerprint image storage unit 11. The binarization circuit 12 reads the multi-tone image stored in the fingerprint image storage unit 11 and converts it into a binary image of "1" and "0" with a certain threshold. For example,
The ridge portion is converted into "1" and the valley portion is converted into "0". The binarized fingerprint image data is stored in the binarized image storage unit 14.

The thinning circuit 13 reads the binarized image stored in the binarized image storage unit 14 and converts it into a thin line.
The binarized image data converted by the thinning circuit 13 is stored in the binarized image storage unit 15. The feature extraction circuit 20 is shown in FIG.
The feature points as shown in 2 are extracted and output. The window selection circuit 21 cuts out the feature points extracted by the feature extraction circuit 20 and the vicinity thereof as windows.

On the other hand, the alignment pattern extraction circuit 30
Calculates the curvature of the fingerprint image by sequentially tracking the lines of the thinned image stored in the thinned image storage unit 15, and uses the line portion whose curvature satisfies a predetermined condition as fingerprint shape information. Remember. Here, the operation of the alignment pattern extraction circuit 30 will be described in detail.

Description will be made with reference to FIG. Line segment L in the figure
Is, for example, a ridge line, a valley line, or a boundary line between these lines in the fingerprint image. It is assumed that the plane curve y = f (x) represents the line segment L. 3 points on L P, Q,
Considering R, if the center of the circle passing through the PQR is K and the radius of this circle is ρ, the following equation holds.

[0019]

[Equation 1]

Here, 1 / ρ is the curvature and ρ is the radius of curvature. FIG. 4 is an explanatory diagram of calculating the average curvature. As shown in FIG. 10A, the curvature is calculated according to the equation (1) while tracing the line segment L from the starting point X = 0 to the ending point X = S. Then, the curved portion having the largest curvature in the entire fingerprint image is detected. (B) is a calculated curvature characteristic diagram. The vertical axis represents the curvature R and the horizontal axis represents the tracking position x. In (b), A
It is assumed that (a) is the average value of the curvature in the line width b (width from x = a to x = a + b).

The average curvature value is plotted over all tracking positions, as shown in (c). The vertical axis represents the average value A (x) of curvature, and the horizontal axis represents the tracking position x. here,
With a threshold value of TH, a section where A (x) ≧ TH (p,
If an arbitrary point r in q) is adopted, the curve portion from x = r to x = r + b will be adopted as an alignment pattern. The part where the curvature is large is near the center of the finger.

The alignment pattern thus adopted by the alignment pattern extraction circuit 30 is stored in the fingerprint dictionary (not shown) from the dictionary output circuit 22. The pattern registered in this way becomes a pattern near the center of the finger.Therefore, at the time of fingerprint matching, by using this alignment pattern to match the fingerprint dictionary pattern with the matching pattern, Matching will be done promptly.

Next, a method of performing the alignment at a higher speed will be described. In order to perform the alignment at a higher speed, the number of pixels in the curved portion of the line selected as the window may be reduced. In the conventional method, as shown in FIG. 13, a rectangular partial image is cut out from a binarized image as a window. Therefore, pattern matching is performed for both the ridge portion (black pixel) and the valley portion (white pixel) to calculate the degree of mismatch.

On the other hand, according to the present invention, as shown in FIG. 5, from a ridge or a valley having a high curvature in a binarized image to a thin line, a part of a line of a thin line image, or about 1 or 2 pixels. A thickened line is generated and the coordinates of these white pixels or black pixels are stored in the dictionary. In FIG. 5, (a) is a thin-line image having a high curvature, (b) is an image obtained by reducing the number of pixels from the thin-line image shown in (a), and (c) is one or two pixels added to the image shown in (a). It is a fat statue. In this way, if you use the original thin line processed as the alignment dictionary,
Alignment becomes faster.

Further, in preparation for the case where the collation fingerprint is rotated and inputted, the image shown in FIG. 6 obtained by rotating the image shown in FIG. 5A by ± θ ° is registered as the alignment dictionary pattern. I'll do it. As a result, even if the collation fingerprint image is rotated and input, the alignment can be quickly performed.

FIG. 7 is a block diagram showing a specific example of the configuration of the alignment pattern extraction circuit 30. The same parts as those in FIG. 2 are designated by the same reference numerals. A solid arrow indicates a data flow, and a broken arrow indicates a signal flow. In the alignment pattern extraction circuit 30, 31 is a control circuit for performing the overall control operation, and 32 is a fine line tracking circuit for reading a thinned image from the thinned image storage unit 15 and tracing a certain point as a starting point. In this case, the type of thin line stored in the thinned image storage unit 15 may be a thin line of either a ridge line or a valley line.

Reference numeral 33 denotes a tracking pixel number storage unit for receiving the output of the fine line tracking circuit 32 and storing the number of tracking pixels, and 34 denotes a fine line end detection circuit for reading the fine line image from the fine line image storage unit 15 and detecting the end thereof. Is. The output of the thin line end detection circuit 34 is input to the control unit 31. Reference numeral 35 is a curvature calculation circuit for calculating the curvature in a section of the fine line traced by the fine line tracking circuit 32 according to the equation (1), and 36 is a curvature calculation for receiving the output of the curvature calculation circuit 35 and storing the calculated curvature. The value storage unit 37 is a partial pixel detection circuit that receives the curvature data from the curvature calculation value storage unit 36 and outputs a portion satisfying a predetermined condition to the dictionary output circuit 22 as an alignment pattern. The operation of the circuit thus configured will be described below with reference to the flowchart of FIG.

When a fingerprint image is input from the fingerprint sensor 1 (S1), the fingerprint image is binarized by the binarization circuit 12 (S2) and stored in the binarized image storage unit 14. The thinning circuit 13 reads the binarized image stored in the binarized image storage unit 14 and performs thinning processing (S3). The binarized image that has undergone the thinning processing is stored in the thinned image storage unit 15. The feature extraction circuit 20 reads the thinned image stored in the thinned image storage unit 15 and extracts the features (S4).
Next, the window selection circuit 21 selects and cuts out the window including the feature points extracted by the feature extraction circuit 20 and the image in the vicinity thereof (S5).

Then, the operation of the alignment pattern extraction circuit 30 starts. The control circuit 31 checks whether or not the tracing of all thin lines has been completed (S6). If the tracking is not completed, the thin line end detection circuit 34 detects the end of the thin line (S7). Here, the initial value of the number N of tracking pixels is set to N = 1 (S8). Next, N = N + 1 is calculated (S9).

When the update calculation of N is completed, the control circuit 31
Compares the tracking pixel number stored in the tracking pixel number storage unit 33 with the threshold value (S10). Here, the threshold value is set for the number of tracking pixels in order to prevent the fine line tracking circuit 32 from continuing the tracking operation indefinitely. When the number of tracking pixels becomes larger than the threshold value, the process returns to step S6 to determine whether the tracking of all thin lines has been completed.

When the number of tracking pixels is smaller than the threshold value,
Continue the tracking operation. That is, the curvature calculation circuit 35 performs the curvature calculation in the thin line area traced by the thin line tracking circuit 32 according to the equation (1) (S11), and the result is stored in the curvature calculation value storage unit 36 (S12). .. After that, the process returns to step S9, and curvature calculation processing is performed until the number of tracking pixels exceeds the threshold value.

If the number of tracking pixels exceeds the threshold value in step S10, the process returns to step S6. Then, it is checked whether the tracking processing has been completed for all the thin lines. When the tracking is completed, the process proceeds to step S13, and the control circuit 31 checks whether or not there is a portion where curvature average ≧ threshold (S13). If so, the partial pixel detection circuit 37 calculates the position of the portion (S1).
4), output to the dictionary output circuit 22 (S15). At this time, the window image selected by the window selection circuit 21 is also output to the dictionary output circuit 22 at the same time. In step S13, if there is no portion where curvature average ≧ threshold value, nothing is done.

FIG. 9 is a block diagram showing the configuration of another embodiment of the present invention. The same parts as those in FIG. 2 are designated by the same reference numerals. In the figure, 1 is a fingerprint sensor, 11 is a fingerprint image storage unit, 12 is a binarization circuit, and 14 is a binarized image storage unit. Reference numeral 40 is a dictionary storage unit that stores a window pattern including feature points and a partial pixel having a large curvature by the above-described method, and 41 is a scanning circuit that scans the partial pixel and the window stored in the dictionary storage unit 40. Is.

Reference numeral 42 denotes the image read by the scanning circuit 41 and 2
A pattern matching circuit that performs pattern matching with the matching image read from the binarized image storage unit 14, 43 is a matching determination circuit that determines the matching result of the pattern matching circuit 42, and 44 is the matching determination circuit 43.
Is a determination circuit for receiving the output of the above and making a final collation determination. FIG. 10 shows an operation at the time of collation of the device configured as described above.
The following is a description using.

First, a fingerprint image for collation is input from the fingerprint sensor 1 (S1). The input fingerprint image is stored in the fingerprint image storage unit 11. The fingerprint image stored in the fingerprint image storage unit 11 is read out by the binarization circuit 12 and
It is digitized (S2). The binarized fingerprint image is stored in the binarized image storage unit 14. Next, the device scan circuit 40
It is checked whether or not the scanning has been completed (S3). If the scanning is completed, the collation is rejected (S
4).

When the scanning is not completed, the pattern matching circuit 42 performs the alignment of the rotation angle θ = 0 °. At this time, the scanning circuit 41 reads a partial pixel having a large curvature stored in the dictionary storage unit 40 and a binarized image stored in the binarized image storage unit 14 so that both patterns match. Is positioned (S5). As described above, since the pattern of the portion having a large curvature exists near the center of the finger, positioning can be performed quickly.

Then, it is checked whether or not the alignment is performed at this position (S6). If the alignment at the rotation angle θ = 0 ° is not possible, the rotation angle θ = −
Positioning is performed with 5 ° partial pixels (S7). And
It is checked whether or not the alignment can be performed with the partial pixel having the rotation angle θ = −5 ° (S8). If the alignment cannot be performed, the alignment is performed on the partial pixel at the position of the rotation angle θ = 5 ° (S9). Then, it is checked whether or not alignment can be performed at this position (S10). If the alignment fails at this position, the process returns to step S3 and the alignment is performed again.

If the alignment is successful in steps S6, S8 and S10, the pattern matching circuit 42
Performs collation using the collation window pattern read by the scanning circuit 41 this time (S11). Matching determination circuit 43
Checks whether or not both patterns match each other while the pattern matching circuit 42 has matched them. The judgment circuit 44 makes a final judgment based on the judgment of the matching judgment circuit 43 and checks the matching condition (S12). If it is determined that they do not match as a result of the collation, the process returns to step S3. If it is determined that they match, the person is confirmed (S13).

[0039]

As described above in detail, according to the present invention, it is possible to provide a fingerprint collation device which can quickly perform position alignment during fingerprint collation.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

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

FIG. 3 is an explanatory diagram of curvature calculation.

FIG. 4 is an explanatory diagram of calculating an average curvature.

FIG. 5 is a diagram showing partial pixels of a curved portion.

FIG. 6 is an explanatory diagram of curved portion rotation.

FIG. 7 is a block diagram showing a specific configuration example of an alignment pattern extraction circuit.

FIG. 8 is a flowchart showing an operation at the time of registration of the present invention.

FIG. 9 is a configuration block diagram showing another embodiment of the present invention.

FIG. 10 is a flowchart showing an operation at the time of matching according to the present invention.

FIG. 11 is a conceptual diagram showing a configuration of a conventional fingerprint matching device.

FIG. 12 is a diagram showing an example of fingerprint characteristic information.

FIG. 13 is an explanatory diagram of a window registration method.

FIG. 14 is a diagram showing a relationship between alignment and verification.

[Explanation of symbols]

 1 Fingerprint Sensor 10 Binarization Circuit 20 Feature Extraction Circuit 21 Window Selection Circuit 22 Fingerprint Dictionary Output Circuit 30 Registration Pattern Extraction Circuit

Claims (7)

[Claims] 1. A fingerprint sensor (1) for reading a fingerprint image
And 2 for the fingerprint image read from the fingerprint sensor (1).
A binarization processing unit (10) that performs binarization processing, thinning processing, etc.
A feature extraction circuit (20) for extracting feature points from the binarization processing unit (10); and a window selection circuit () that receives the output of the feature extraction circuit (20) and cuts out the feature points and the vicinity thereof as windows ( 21), the curvature of the fingerprint image is calculated by tracing the line from the fingerprint image read from the binarization processing unit (10), and the line portion where the curvature satisfies a predetermined condition is fingerprint shape information. A fingerprint including a registration pattern extraction circuit (30) for storing as a fingerprint and a fingerprint dictionary output circuit (22) for outputting the output of the window selection circuit (21) and the registration pattern extraction circuit (30) as a fingerprint dictionary. Matching device. 2. The alignment pattern extraction circuit (3)
The fingerprint collation device according to claim 1, wherein 0) is adapted to trace the thinned image of the ridge to calculate the curvature of the fingerprint image. 3. The alignment pattern extraction circuit (3)
2. The fingerprint collation device according to claim 1, wherein 0) is adapted to trace the thinned image of the valley line to calculate the curvature of the fingerprint image. 4. The alignment pattern extraction circuit (3)
The fingerprint collation device according to claim 1, wherein 0) is adapted to detect a line portion having an average curvature equal to or more than a threshold value and register it as a fingerprint dictionary. 5. The alignment pattern extraction circuit (3)
The fingerprint collation device according to claim 1, wherein 0) is adapted to detect a line portion where the minimum value of the curvature of a certain section width is equal to or more than a threshold value and register it as a fingerprint dictionary. 6. The fingerprint according to claim 4, wherein pixels are extracted from the thinned image of the fingerprint at predetermined intervals, and these extracted pixels are registered as a fingerprint dictionary. Matching device. 7. The fingerprint collation apparatus according to claim 4, wherein, in addition to the registration portion satisfying the condition, an image obtained by rotating the portion by a predetermined angle is also registered as a fingerprint dictionary.