JP4592921B2 - Fingerprint verification device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a fingerprint collation apparatus that collates an input fingerprint with a registered fingerprint and determines whether or not the person is the person himself / herself.
[0002]
[Prior art]
FIG. 9 is a functional block diagram showing a conventional fingerprint collation apparatus. Fingerprint image data is read by the image capturing means 1 and fingerprint collation feature data 3 is created by the feature extraction means 2. The collation feature data 3 and the registered feature data 4 are collated by the collating means 5 to calculate a collation rate. If the collation rate is equal to or greater than the threshold value, the result is OK. If the collation rate is lower than the threshold value, the collation is performed again.
[0003]
[Problems to be solved by the invention]
In the conventional fingerprint collation apparatus as described above, the pass / fail is determined by comparing the collation rate with a threshold value. Therefore, when there is a person who is difficult to collate due to a thin fingerprint, the threshold value (OK criterion) ) Must be set low. However, in this case, there is a problem that the probability of accepting another person (other person's spoofing rate) is high and the identification rate is low.
[0004]
The present invention was made to solve the above problems, and an object of the present invention is to provide a fingerprint collation apparatus that can improve the identification rate while keeping the other person's spoofing rate constant even when the identification rate is low. And
[0005]
[Means for Solving the Problems]
The fingerprint collation device according to the first aspect of the present invention performs collation a prescribed number of times when the collation rate is lower than the threshold value, calculates the appearance frequency of each feature point from the collation feature data, and assigns the frequency. Create matching feature data, compare this frequency-matched matching feature data with pre-created frequency-registered feature data, calculate the number of matching feature points for each frequency, calculate the determine specific function using the collation rate, this discriminant function value, is obtained so as to determine the acceptability of the input fingerprint.
[0007]
The fingerprint collation device according to the second invention is the one according to the first invention, wherein if the fingerprint is determined to be negative, it is determined whether the discriminant function value is in the holding range, and if it is determined that the fingerprint function is in the holding range, The frequency-matching feature data and the frequency-registered feature data are collated again, and the pass / fail of the fingerprint is determined again based on the result.
[0008]
The fingerprint collation device according to the third invention is the one according to the second invention, wherein when the fingerprint is re-determined, the number of collations between the frequency-matched feature data and the frequency-registered feature data exceeds the specified number. , The repetition of verification is permitted.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
Embodiment 1 FIG.
1 to 5 are diagrams showing an embodiment of the first and second inventions of the present invention. FIG. 1 is a functional configuration diagram, FIGS. 2 and 3 are verification operation flowcharts, and FIG. 4 is an operation of frequency calculation means. FIG. 5 and FIG. 5 are explanatory diagrams of an example of creating feature data with frequency. (The following embodiments are also the same)
[0010]
In FIG. 1, 1 is an image capturing means for reading fingerprint image data of a finger placed at a predetermined position, and 2 is a fingerprint feature data 3 (end point, branch point, etc.) from the fingerprint image data read by the image capturing means 1. Feature extraction means to extract, 4 is registered feature data with frequency registered in advance, 5 is collation means for collating feature data 3 and registered feature data 4, and 6 is a collation result for displaying the result of the above-mentioned collation OK / NG A display 7 is a frequency calculation means for calculating the appearance frequency of feature points by the above collation, and 8 is frequency-equipped feature data calculated by the frequency calculation means 7.
[0011]
Next, the collating operation of this embodiment will be described with reference to FIGS. In this embodiment, the user's operation is instructed by an appropriate display (not shown), and the number of fingerprint data used for frequency calculation, that is, the specified value Nmax = 3.
In step S1, first, the counter N is initialized to zero. In step S2, the fingerprint image data is read by the image capture means 1, and the fingerprint feature data 3 is created by the feature extraction means 2.
[0012]
Next, in step S3, the collation unit 5 collates the registered feature data 4 with the feature data 3 that is collation data, and calculates a collation rate F. In step S4, it is determined whether the verification rate F is equal to or greater than a threshold value Th. If it is equal to or greater than the threshold value Th, the person is recognized, and in step S5, OK processing is performed and the collation result display unit 6 displays the collation result. If the verification rate F is smaller than the threshold value Th, the counter N is incremented by 1 in step S6, and the process proceeds to step S7.
[0013]
In step S7, it is determined whether the counter N has reached a specified value Nmax (= 3). If the specified value Nmax has not been reached, the process returns to step S2. When the counter N reaches the specified value Nmax, the process proceeds to step S8. In step S8, the appearance frequency of each feature point is calculated from the three sets of collation data so far including this time, and collation feature data with frequency is created.
Next, creation of frequency matching feature data will be described with reference to FIGS.
[0014]
In FIG. 4, the frequency calculation means 7 creates frequency-equipped feature data 8 based on the three-time collation data of the first to third feature data 3-1 to 3-3.
FIG. 5 shows a specific example. Based on the feature data 3-3, the features A1 and A2 are frequency 3 because they match each other in the feature data 3-1 to 3-3, and the feature A3 is feature data. Since the frequency 3-1 and 3-3 match, the frequency is 2, and the other features A4 and A5 appear only in the feature data 3-3. Although the frequency is calculated as described above, collation is performed several times when the registered feature data is created, and the frequency is calculated in the same manner and stored as the frequency-added registered feature data 4.
[0015]
In step S9, the collated feature data with frequency created as described above and the registered feature data with frequency are collated, the number of feature points matched for each frequency is calculated, and the number of feature points matched and collated for each maximum frequency. The discriminant function S is calculated using the rate.
Next, the number of matches between the frequency 3 of registered data and the frequency 3 of collation data, which is the maximum frequency in this case, will be described as M33.
Discriminant function S = f (matching rate F, number of matches M33 for each frequency 3)
However, (f: function)
[0016]
A specific discriminant function S is shown below.
S = p1 × F + p2 (M33 / number of feature points) −p3 | (a3 / a) − (b3 / b) | + p4
Here, a and b represent the number of feature points of registration data and collation data, respectively, and a3 and b3 represent the number of feature points of weight 3 of registration data and collation data, respectively. The coefficients p1 to p4 are coefficients that are statistically calculated in advance, and are coefficients that discriminate between the person and others. The probability of the person increases as the discriminant coefficient S increases, and the probability of the other person increases as the value decreases.
[0017]
In step S10, it is determined whether the discriminant function S is greater than or equal to a certain value Sh. If it is greater than or equal to the certain value Sh, it is determined that the person is the person, and OK processing is performed in step S11. NG processing is performed at S12.
Here, S4, S6 to S8 constitute frequency calculation means, S9 constitutes discriminant function calculation means, and S10 constitutes pass / fail judgment means.
[0018]
In this way, even in the case of a person with a low matching rate at the time of matching, by using a plurality of feature data obtained by re-input and using a high-precision discriminant function based on stable feature point information, Since the collation is facilitated without lowering the threshold value, it is possible to improve the collation accuracy for the person while maintaining the misrepresentation rate of others (preventing frequent misrecognition). In addition, since a new external device is not added, an inexpensive and reliable device can be realized.
[0019]
Embodiment 2. FIG.
FIG. 6 and FIG. 7 are collation operation flowcharts showing one embodiment of the third invention of the present invention. 1, 2, 4, and 5 are shared by the second embodiment.
Next, the operation of this embodiment will be described.
Steps S1 to S11 are the same as those in the first embodiment.
If the discriminant function S is smaller than the fixed value Sh in step S10, the process proceeds to step S15, and it is determined whether the discriminant function S is in the holding range (S1 to Sh). If it is outside the holding range, NG processing is performed in step S16.
[0020]
If it is within the reserved range, there is a possibility of the person, so the process proceeds to step S17, and the image data is taken in again and the feature data is extracted. Evaluate S. Then, depending on the result, in step S21 or step S22, OK processing or NG processing is performed.
Here, step S15 constitutes a hold determination means, and steps S17 to S20 constitute pass / fail redetermination means.
[0021]
In this way, even if the fingerprint verification fails, if the fingerprint data is in the holding range, the pass / fail is re-determined by re-verification. Opportunities are given and convenience can be improved.
[0022]
Embodiment 3 FIG.
FIG. 8 is a collation operation flowchart showing one embodiment of the fourth invention of the present invention. 1, 2, and 4 to 6 are shared by the third embodiment.
FIG. 8 is obtained by adding step S23 to FIG.
Steps S1 to S11 and S15 to S21 are the same as those in the second embodiment.
If the discriminant function S is smaller than the fixed value Sh in step S20, the process proceeds to step S23, and it is determined whether the number of repetitions exceeds the specified number. If the specified number is not exceeded, the process proceeds to step S15. If the specified number is exceeded, NG processing is performed in step S22.
[0023]
Here, step S23 constitutes a collation repetition permission means.
In this way, in the case of NG in the second determination, since re-matching is permitted up to the specified number of times, it is possible to provide a rescue opportunity for a person with a bad fingerprint state and further improve convenience. It becomes possible.
[0024]
【The invention's effect】
Or as described in the first aspect of the invention, the collation rate is carried out matching the prescribed number is lower than the threshold value, matching with the frequency and calculates the appearance frequency of each feature point from their matching feature data Create feature data, compare this frequency matching feature data with frequency registered feature data, calculate the number of matches for each frequency, and use the number of feature points for each maximum frequency and the matching rate to determine the discriminant function. Since it is calculated and the pass / fail of the fingerprint is determined based on the discriminant function value, it is easy to perform collation without lowering the threshold value, and the identity identification rate can be improved while keeping the other person's spoof rate constant. .
[0025]
In the second invention, if it is determined that the fingerprint is negative, it is determined whether the discriminant function value is in the holding range, and if it is determined that it is in the holding range, the frequency-matching feature data and the frequency-registered feature data are again displayed. Since the result of the fingerprint is re-determined based on the result, even a person with a poor fingerprint status can be given a re-determination opportunity to improve convenience.
[0026]
Further, in the third invention, when the fingerprint is re-determined to be negative, the collation is allowed to be repeated until the number of collations between the frequency-matched feature data and the frequency-registered feature data exceeds the specified number. A person who has a bad fingerprint status can be given relief, and convenience can be further improved.
[Brief description of the drawings]
FIG. 1 is a functional configuration diagram showing a first embodiment of the present invention.
FIG. 2 is a collation operation flowchart showing the first embodiment of the present invention;
FIG. 3 is a collation operation flowchart showing the continuation of FIG. 2;
FIG. 4 is an operation explanatory diagram of frequency calculation means showing Embodiment 1 of the present invention.
FIG. 5 is an explanatory diagram of an example of creating frequency feature data according to the first embodiment of the present invention.
FIG. 6 is a collation operation flowchart showing Embodiment 2 of the present invention.
FIG. 7 is a collation operation flowchart showing the continuation of FIG. 6;
FIG. 8 is a collation operation flowchart showing the third embodiment of the present invention (continuation of FIG. 6).
FIG. 9 is a functional configuration diagram showing a conventional fingerprint collation apparatus.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Image capture means, 2 Feature extraction means, 3 Feature data, 3-1 to 3-3 1st-3rd feature data, 4 Frequency registration feature data, 5 Matching means, 6 Matching result display, 7 Frequency calculation means 8 Feature data with frequency.
S4, S6 to S8 Frequency calculation means, S9 discriminant function calculation means, S10 pass / fail determination means, S15 hold determination means, S17 to S20 pass / fail re-determination means, S23 collation repeat permission means.

Claims (3)

The collation feature data extracted from the input fingerprint image data is collated with the pre-registered feature data, and if the collation rate indicating the degree of collation is equal to or greater than a threshold value, the collation result is determined to be coincident. In the device
A frequency calculation means for performing the matching a predetermined number of times when the matching rate is lower than the threshold value, calculating the appearance frequency of each feature point from the matching feature data, and creating frequency matching feature data; ,
The matching feature data with frequency and the registered feature data with frequency created in advance are collated, the number of matching feature points for each frequency is calculated, and the number of matching feature points for each maximum frequency and the matching rate are used. A discriminant function calculating means for calculating a discriminant function;
A fingerprint collation apparatus comprising: pass / fail judgment means for judging pass / fail of the inputted fingerprint based on the discriminant function value. A hold determination unit that determines whether the discriminant function value is in the hold range when it is determined as NO by the pass / fail determination unit;
Claims characterized in that there is provided pass / fail re-determination means for collating frequency-matching feature data with frequency-registered feature data again when it is determined that the position is within the holding range, and judging pass / fail of the fingerprint based on the result. Item 2. The fingerprint collation device according to Item 1 . A verification repetition permission unit is provided that permits the verification to be repeated until the number of verifications between the frequency-matched feature data and the frequency-registered feature data exceeds a specified number when the pass / fail re-determination unit determines NO. The fingerprint collation apparatus according to claim 2 .

Images