KR20040078290A - Extraction and matching method of ridge number in finger identification system - Google Patents

Abstract

PURPOSE: A method for extracting/matching a ridge number information on a fingerprint recognition system is provided to extract the ridge number information, which is a feature value independent to the rotation, the transition, and the resolution change of a fingerprint image, and perform fingerprint matching by using the extracted information. CONSTITUTION: A virtual line connecting the first minutia and the second minutia is generated(303). Labeling is performed at the first and the second minutia by following a ridge(305). The virtual line is divided into segments(307). The segment is selected from the segments in a sequential order(311). It is checked that the selected segment is labeled by examining a pixel value(315). In case that the segment is not labeled, it is judged that the segment is the ridge by examining the pixel value(317). In the case of the ridge, the coordinates of a segment position is decided as an intersecting point and a ridge number is increased(319). After completely examining all segments, all labels are removed and the accumulated ridge number is decided as the ridge number existing between two minutiae(329).

Description

Extraction and Matching of Ridge Count Information in Fingerprint Recognition System {EXTRACTION AND MATCHING METHOD OF RIDGE NUMBER IN FINGER IDENTIFICATION SYSTEM}

The present invention relates to a fingerprint recognition system, and particularly, extracts ridge number information, which is a feature amount independent of rotation, transition, and resolution change of a fingerprint image, from a fingerprint image, and performs fingerprint matching (or matching) using the ridge number information. It relates to a method for doing so.

In general, biometrics (Biometrics) refers to a technology that uses the unique characteristics of each person as a unit of measure for identification based on human physiological or behavioral characteristics. In other words, it is a technical field that uses human biological characteristics for identification through an automated device. Biometric technology is safer and more convenient than conventional password or password identification methods because the recognized person must actually exist at the time of recognition and eliminate the need to remember passwords or carry tokens. The bio-parts used for biometrics are diverse such as fingerprint, retina, iris, face, hand, vein, voice, signature, body smell, DNA, etc. However, the most popular parts are bio-prints using fingerprint, voice, iris, and face. It is awareness.

In particular, fingerprint recognition is one of the oldest technologies that has been used successfully in many applications. Fingerprints are raised in the sweat glands to form a constant flow. Retina and Iris are more reliable in terms of reliability and stability due to their unique characteristics that are different from person to person and that do not change for life. It is estimated to be higher than the vein, facial, and the like, and is used as an efficient personal authentication method.

In general, conventional fingerprint recognition methods include thinning, Fourier transform or wavelet transform in frequency space, or neural network or fuzzy logic. Needed. It is also common to use a combination of methods such as syntactic, statistical, rule-based, and neural network construction. Chain code or ridge tracking is a common feature.

The fingerprint recognition algorithm based on the conventional feature points described above is suitable for the geometrical deformation of the extracted feature points even if the resolution of the fingerprint image, the distribution of pixel values, and the sensing method of the fingerprint input sensor are different. There is a problem in that the recognition rate cannot be guaranteed, and the aforementioned ridge number information extraction and use method is vulnerable to rotation and local deformation of the fingerprint, and has a problem of poor reproducibility.

For example, the conventional method of using the number of ridges is as follows.

First, referring to the technology of NEC (US Patent No. 4,944, 021), one reference feature point is selected and four feature points present on the quadrant based on the direction of the feature point are selected and existed in each quadrant. We propose a method of searching for the existence of a similar group by using ridge number information between feature points as a group. This technique can be matched even in the remaining fingerprints where only a part of the fingerprint remains, but if the direction of the reference feature is rotated, the distribution of the feature points present in the quadrant may change, and it is difficult to extract the same five feature points from the input fingerprint during matching. Has

On the other hand, looking at the technology of IBM (US Patent No. 6,266,433), selecting two feature points and connecting them with a virtual straight line consisting of five or three pixels, and each of five or three pixels as one segment We propose a method to extract the number of ridges by checking whether the segment of the ridge is a ridge or a valley. Since this technique calculates 5 or 3 pixels as one segment, the ridge number information can be extracted only in an area in which the ridges existing between the two feature points are orthogonal to the virtual straight line connecting the two feature points. That is, since the exact number of ridges is extracted in a region where the flow of the ridge is high, the reliability is high, but the ridge information extraction fails even if some deformation occurs in the flow of the ridge.

The above-described fingerprint recognition methods of NEC and IBM are technologies used for the Automatic Fingerprint Identification System (AFIS). Since the fingerprint image obtained by using the fingerprint input device is relatively smaller than the fingerprint region of the rotating fingerprint image, there is a problem in using the NEC / IBM ridge number extraction method used in the AFIS system. Therefore, there is a need for a new method suitable for live scan images and a method for extracting a sufficient number of ridge number information in consideration of registration.

In order to solve the above problems, an object of the present invention is to provide a method for extracting ridge number information, which is a characteristic amount independent of rotation, transition, and resolution change of a fingerprint image in a fingerprint recognition system. .

It is another object of the present invention to provide a method for extracting ridge number information, which is a feature amount independent of rotation, transition, and resolution of a fingerprint image in a fingerprint recognition system, and performing fingerprint matching using the extracted ridge number information. Is in.

Another object of the present invention is to detect the same click (Clique) between the enrolled fingerprint and the input fingerprint in the fingerprint recognition system using the ridge number information, and to find the same feature point outside the detected same click to expand the click In addition, the present invention provides a method for determining whether fingerprint matching is successful based on the number of identical feature points acquired in this process.

According to the first aspect of the present invention to achieve the above objects, a method for extracting the number of ridges existing between two feature points in the fingerprint image, generates a virtual straight line connecting the first feature point and the second feature point A second process of labeling a predetermined length along the ridge at the first and second feature points, a third process of dividing the virtual straight line into a plurality of segments, and the plurality of processes A fourth step of selecting a specific segment in a sequential order among the segments, a fifth step of determining whether a label is checked by checking a pixel value of the selected segment, and a ridge line by checking the pixel value if it is not labeled A sixth process of determining whether the image is ridged, a seventh process of determining the coordinates of the corresponding segment position as an intersection and increasing the number of ridges; The eighth step of returning to the fourth step after labeling along the ridge, and upon completion of the inspection of the plurality of segments, all labeling on the ridge is removed, and the number of accumulated ridges so far is calculated between the two feature points. It characterized in that it comprises a ninth process to determine the number of ridges present in.

According to a second aspect of the present invention, a fingerprint recognition method includes: converting an analog fingerprint image from a fingerprint input sensor into a digital fingerprint image, preprocessing the digital fingerprint image to generate a thinned image; Extracting feature points from the thinned image, extracting the number of ridges existing between the feature points, inputting a specific user with information of the extracted feature points and information of the number of ridges Generating characteristic data of a fingerprint, reading characteristic data of a registered fingerprint corresponding to the specific user, characteristic data of the input fingerprint and characteristic data of the registered fingerprint Create groups of three feature points with each, and three feature points for each group (B) generating feature data (click) including shape information on a triangle and ridge number information between the three feature points; and comparing the feature data of the input fingerprint with the feature data of the registered fingerprint. Checking whether there are identical clicks between fingerprints, and when the same click exists, searching for the same feature point present in both the enrolled fingerprint and the input fingerprint based on the same click, and using the searched same feature point as a vertex. Repeating an operation of expanding a triangle formed by three feature points into a polygon, obtaining a number of identical feature points, and determining whether the registration fingerprint and the input fingerprint are the same using the obtained number of identical feature points. Characterized in that.

1 is a block diagram of a fingerprint recognition system according to an embodiment of the present invention.

2 is a view for explaining a method for extracting the number of ridges in a thinned fingerprint image.

3 is a diagram illustrating a procedure for extracting the number of ridges existing between two feature points according to an embodiment of the present invention.

4 is a diagram illustrating a fingerprint image comparison procedure using the ridge number information according to an embodiment of the present invention.

5 is a diagram illustrating a detailed procedure of the same group determination process (step 425) described with reference to FIG.

FIG. 6 is a diagram showing the detailed procedure of the polygon extension section (step 427) described in FIG.

FIG. 7 is a diagram showing a detailed procedure of the same feature point detection process (step 607) described with reference to FIG.

8 is a view for explaining feature data (click).

9 is a diagram showing a relationship between a click on a registered fingerprint and a click on an input fingerprint;

10 is a view for explaining a polygon expansion method having the same feature point as a vertex.

<Description of Symbols for Main Parts of Drawings>

10: fingerprint image input device 20: A / D converter

30: Image memory 40: I / O interface

50: console 60: external storage device

70: main controller 80: working memory

90: program memory 100: output device

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, if it is determined that the detailed description of the related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

Hereinafter, a description will be given of a technique of extracting ridge number information between two feature points independent of rotation, transition, and resolution change of a fingerprint image, and determining similarity between a registered fingerprint and an input fingerprint using the ridge number information. will be. The present invention can provide a reliable recognition rate regardless of the type of fingerprint input sensor used for fingerprint image acquisition.

First, look at the symbols used in the present invention before the description as follows.

P1, P2 ---> Two Feature Points in Thinning Images

L1 ---> Segment connecting two feature points

CP1, CP2 ---> Intersection of L1 with Thinned Ridge

M1.d ---> direction angle of the first feature point

M2.d ---> direction angle of the second feature point

M3.d ---> Direction angle of the third feature point

RC ₁ ---> Number of ridges between the first and second feature points

RC ₂ ---> Number of ridges between 2nd and 3rd feature points

RC ₃ ---> Number of ridges between the third and the first feature point

θ _{B --->} first angle between angles formed by three feature points from the center of the circumscribed circle

θ _{S --->} Second angle between angles formed by three feature points from the center of circumscribed circle

θ _{R --->} Reference angle (or third angle) for rotation compensation during registration

M1 ---> 1st feature

M2 ---> 2nd feature

M3 ---> 3rd Features

--A ---> M1, M2, C The first of the triangular cabinets

--B ---> M1, M2, C The second of the triangle cabinets that make up the three feature points

--C ---> M1, M2, C The third of the triangle cabinets

Here, a predetermined angle between two of the angles except for the angle between the center of the triangle circumscribed by the three feature points, including the 0 ° reference axis of the three angles formed by the three feature points first This is called the inter-angle, and the remaining inter-angle is defined as the second inter-angle.

1 is a block diagram of a fingerprint recognition system according to an embodiment of the present invention.

As shown, the fingerprint recognition system, the A / D converter 20, the image memory 30, the I / O interface 40, the console 50, the external storage device 60, the main control unit 70 , A working memory 80, a program memory 90, and an output device 1d00. Here, reference numeral 10 denotes a heterogeneous fingerprint input sensor (different types of fingerprint input sensors) that can be coupled to the fingerprint recognition system.

Referring to FIG. 1, first, the fingerprint input sensor 10 scans a fingerprint of a finger contacting an upper surface of a panel and outputs an analog signal according to the contrast of the fingerprint image to the A / D converter 20. The A / D converter 20 converts an analog signal applied from the fingerprint image input apparatus 10 into a digital signal and outputs the digital signal to the image memory 30. That is, the A / D converter 20 converts an analog signal corresponding to each pixel (or pixel) into a bit string having a predetermined number of bits and outputs the converted bit signal. The image memory 30 stores digital data about a fingerprint image applied from the heterogeneous fingerprint input sensor 10. The I / O interface 40 matches the image memory 30 with the console 50, the output device 100, and the main controller 70.

The console 50 includes a plurality of operation keys, and outputs a key signal according to a user's operation to the main controller 70. The external memory device 60 stores a plurality of pre-input fingerprint image data (characteristic amount data (feature point information, ridge information, etc.) of the registered fingerprint). The main control unit 70 performs an operation mode according to the key signal input from the console 50. When the operation mode is the authentication mode, the main control unit 70 reads the registered local culture image (characteristic data) according to the identification information (eg, ID, employee number, resident number, etc.) input by the user through the console 50. It reads from the external storage device 60, compares the read registration culture image and the input paper culture image currently input from the fingerprint input sensor 10 and outputs the result through the output device 100.

The work memory 80 is a memory used when comparing a currently input fingerprint image with a pre-registered fingerprint image. The program memory 90 stores program data and initial data for controlling the overall operation of the fingerprint recognition system. In particular, an algorithm is programmed for extracting and matching features of a fingerprint according to the invention. The output device 100 displays an overall state of the fingerprint recognition system and input user input information and the like on a display window (eg, a liquid crystal display (LCD)) under the control of the main controller 70. In particular, according to the present invention, a message or a corresponding fingerprint image according to the comparison result of the input fingerprint image and the comparison target fingerprint image is displayed.

The operation of the fingerprint recognition system configured as described above will be described in detail with reference to the accompanying drawings.

First, the ridge number extraction method will be briefly described with reference to FIG. 2. A straight line L1 connecting two feature points P1 and P2 in a thinned image is generated, starting from the feature point P1 and proceeding to P2 along a straight line L1. The change of the pixel value of L1 is examined to detect the points CP1 and CP2 that intersect with the ridge, and the number of detected intersections is determined as the number of ridges. In order to prevent false ridge number information extraction, the ridge is tracked and labeled from the coordinates of the feature points P1 and P2 to a distance n times the average ridge interval RD (RIDGE DISTANCE), and the same method from the coordinates of the intersection points CP1 and CP2. Track and label the ridges. N is an integer of 2 or more.

3 illustrates a procedure for extracting the number of ridges existing between two feature points according to an embodiment of the present invention.

Referring to FIG. 3, first, the main controller 70 extracts feature points from a thinned fingerprint image in step 301 and determines two feature points P1 and P2 among the extracted feature points. In operation 303, the main controller 70 generates an imaginary straight line L1 connecting the two feature points. In operation 305, the main control unit 70 performs labeling along the ridges of the determined two feature points P1 and P2. Perform. Here, the virtual straight line L1 is generated using a "bresenham" algorithm, and the labeling distance is set to 20 pixels, which is twice the average ridge spacing, for example, for a 500 dpi fingerprint image.

In operation 307, the main controller 70 divides the straight line L1 into a predetermined length (a predetermined number of pixels) to generate a plurality of segments, and numbers each of the segments from a predetermined reference feature point. Assign sequentially.

In operation 309, the main controller 70 initializes segment number N (N = 1). In operation 311, the main controller 70 examines the pixel value of the N-th segment in the fingerprint image, and checks whether the pixel value checked in step 313 exists within a predetermined range and cannot be determined. If it is not determined, the main controller 70 proceeds to step 333 to remove the labeling, displays an error message in step 335, and then proceeds to step 331. If the state can be determined, the main controller 70 checks whether the N-th segment is labeled by comparing the inspected pixel value with a predetermined reference value in step 315.

If it is labeled, the main control unit 70 proceeds to step 323, otherwise, the main control unit 70 compares the inspected pixel value with predetermined reference values in step 317 to determine the N-th segment position. Determine if it is or ridge. Here, in the case of the goal, the main control unit 70 proceeds to step 323, and if it is determined that the ridge is the main control unit 70 proceeds to step 319 to coordinate (x, y) of the N-th segment position. Is determined by the coordinate of the intersection point CP, and the number of ridges is increased (rc = rc + 1). In operation 321, the main controller 70 performs labeling on the intersection point. The reason for labeling as described above is to prevent the case of erroneously recognizing one ridge as several ridges when the direction of the ridge changes abruptly or when the imaginary straight line connecting the two feature points and the ridge contact each other.

Thereafter, the main controller 70 increases the segment number by '1' in step 323 and checks whether the increased segment number exceeds the number of segments M in step 325. If the segment number exceeds the number of segments, the main controller 70 proceeds to step 327. Otherwise, the main controller 70 returns to step 311 to check the next segment and repeats the following process.

On the other hand, the main controller 70 removes all the labeling performed on the fingerprint image in step 327, and stores and outputs the number of ridges corresponding to the two feature points in step 329. In operation 331, the main control unit 70 determines whether the number of ridges existing between all the feature points of the fingerprint image is extracted. If the ridge number extraction has been performed for all the feature points, the main control unit 70 terminates the present program. Otherwise, if it is determined that there are remaining feature points to perform the ridge number extraction, the process returns to step 301 and the following process is performed. Rerun

As described above, the ridge number information is extracted for all combinations of all two feature points that can be generated from the feature points extracted from the fingerprint image, and the extracted ridge number information is obtained by indexing the feature points as shown in Table 1 below. ROW) and columns (COLUMN) are stored in a triangular matrix data structure. The triangular matrix may be an upper triangular matrix (UPPER TRIANGULAR MATRIX) or a lower triangular matrix (LOWER TRIANGULAR MATRIX).

INDEX One 2 3 ... N 0 3 2 4 ... 6 One X 2 5 ... 4 2 X X 5 ... 2 ... X X X ... One N-1 X X X X 3

(X: Don't care)

4 illustrates a fingerprint image comparison procedure using ridge number information according to an embodiment of the present invention.

Referring to FIG. 4, first, in operation 401, the main control unit 70 checks whether an authentication request key according to a user input is input. When the authentication request key is input, the main control unit 70 displays identification information (eg, ID, employee number, social security number, etc.) input request message in step 403, and checks whether the identification information is input in step 405. do.

When the identification information is input, the main control unit 70 searches the database of the external storage device 60 with the identification information in step 407, and checks whether the same identification information is registered in step 409. If the same identification information is registered, the main control unit 70 proceeds to step 411 and displays a fingerprint input request message on the display window. If the same identification information does not exist, the main control unit 70 proceeds to step 435. Proceed to display the error message and exit.

After the fingerprint input request message is displayed, the main controller 70 checks whether a fingerprint image is input from the fingerprint input sensor 10 in step 413. When the fingerprint image is input, the main controller 70 extracts feature data (feature point information, ridge number information) from the input fingerprint image in step 415. After acquiring the feature data for the input fingerprint image, the main control unit 70 stores the feature data (characteristic data of the registered fingerprint image) stored with the input identification information in step 417. Read out from the device 60.

In operation 419, the main control unit 70 generates groups of three feature points for the feature points of the read feature data, and generates feature data (or clicks) for each of the groups. Here, all combinations of groups that can be generated with three feature points are vivid, except for those whose radius of the circumscribed circle is outside the threshold, and the difference between the largest angle θ _M3 and the smallest angle θ _R is the threshold value. Exclude the following groups. For example, as shown in FIG. 8, when three feature points are selected, information (coordinates and types) of the three feature points and between the three feature points M ₁ , M ₂ , and M _{3 are} shown. The sum of the squares of the ridge number information RC ₁ , RC ₂ , and RC ₃ at SUM_RC ² , and the three angles (θ _B , θ _S , θ that form three feature points from the center of the circumscribed circle of the three feature points) _R ) and the direction of the three feature points (M1.d, M2.d, M3.d) with the feature data (or Clique) is defined.

After generating the feature data, the main control unit 70 generates groups of three feature points from the feature data of the input fingerprint image in the manner described above in step 421, and the feature data for each of the groups. Create As described above, after generating the feature data for each of the registered fingerprint and the input fingerprint, the main control unit 70 in step 423 sums the feature data for each of the registered fingerprint and the input fingerprint sum of the squares of the ridge number information. Sort by.

After arranging the feature data, the main control unit 70 performs the predetermined sequence of feature data (group) among the feature data (or group) of the registration fingerprint and the feature data of the input fingerprint in step 425. It is checked whether or not the same group is compared by comparing any one feature data. That is, the information constituting the feature data (sum of squares of ridge number information, three inter-angle angles (θ _B , θ _S , θ _R ) forming three feature points from the center of the circumscribed circle of the triangle and directions of three feature points ( M1.d, M2.d, M3.d)) are compared to each other to determine whether they are identical. This will be described in detail later with reference to FIG. 5.

If it is determined that the two groups are identical to each other, the main control unit 70 proceeds to step 427, and if it is determined to be different from each other, proceeds to step 429. Meanwhile, the main controller 70 searches for the same feature point in addition to the three feature points forming the group in step 427, and repeats the operation of extending the polygon by using the detected feature point as a vertex when detecting the same feature point. The feature point is detected. This will be described in detail later with reference to FIG. 6.

When the polygon expansion (same feature point detection) is completed, the main controller 70 checks whether there is a group to be examined any more in step 429. If there are no more groups to test, the main control unit 70 returns to step 425 to perform the following process again to check the next order combination (two groups). If the group to be examined no longer exists, the main controller 70 proceeds to step 431 and determines whether the registered fingerprint and the input fingerprint are the same based on the number of identical feature points obtained through the polygon expansion.

For example, a value obtained by dividing the number of the same feature points by the minimum of the number of the feature points of the registered fingerprint and the number of the feature points of the input fingerprint as a similarity is determined as similarity, and comparing the similarity with a predetermined reference value to determine whether they are the same. Here, the number of the same feature point is obtained as many as the number of the same group detected as described above, and the similarity is calculated by taking the largest value among the obtained number of the same feature point. The similarity may be calculated by other probabilistic calculation methods in addition to the above-described method.

Similarity = number of identical feature points / MIN (number of registration points, number of input points)

If it is determined that the registered fingerprint and the input fingerprint is the same, the main control unit 70 proceeds to step 433 and displays the authentication success message and ends. If it is determined that the registered fingerprint and the input fingerprint are different, the process proceeds to step 433. Proceed and display the authentication failure message and exit.

FIG. 5 shows a detailed procedure of the same group determination process (step 425) described with reference to FIG. First, prior to description, it is assumed that the group of the enrolled fingerprint and the group of the input fingerprint among the two groups to be examined are as shown in FIGS. 9A and 9B, respectively.

Referring to FIG. 5, first, in operation 501, the main control unit 70 adds up the sum of squares of the number of pieces of ridge number information among the feature data of the registered fingerprint. Sum of squares of ridge number information among features and input fingerprint features Calculate the difference between and check if the difference is less than a certain tolerance (eg 2). If it is greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 517 to determine another group and returns to step 429 of FIG. 4.

If it is smaller than the predetermined tolerance, the main control unit 70 proceeds to step 503 in which the first interstitial angle T.θ _B of the feature data of the registered fingerprint and the first of the feature data of the input fingerprint are performed. The difference between the angles A.θ _B is calculated and it is checked whether the calculated difference is less than a predetermined tolerance (eg 30 °). If less than the tolerance, the main control unit 70 proceeds to step 505, and if greater than or equal to the tolerance, proceeds to step 517 to determine a different group and returns to step 429 of FIG.

On the other hand, the main controller 70 calculates the difference between the respective (A.θ _S) between the second of the second between each (T.θ _S) and the input feature data of the fingerprint of the registered fingerprint in the step 505 In addition, it is checked whether the calculated difference is smaller than a predetermined tolerance (eg, 30 °). If the tolerance is smaller than the tolerance, the main control unit 70 proceeds to step 507 such that the angle θ _R between the predetermined reference (0 °) and the specific feature point (M1 in the drawing) is equal to (or specified characteristic point). The input fingerprint is rotated (feature data update) so that the direction angles are the same. The rotation angle may be obtained by subtracting the third inter-angle angle A.θ _R of the input fingerprint from the third inter-angle angle T.θ _R of the registered fingerprint. On the other hand, if it is greater than or equal to the tolerance, the main controller 70 proceeds to step 517 to determine a different group and returns to step 429 of FIG. 4.

After correcting the feature data of the input fingerprint such that the registration fingerprint and the input fingerprint are as geometrically similar as described above, the main control unit 70 in step 509 the direction angle of the first feature point of the feature data of the registration fingerprint (TM ₁₎ .d) and the difference between the direction angle AM ₁ .d of the first feature point among the feature data of the corrected input fingerprint, and check whether the calculated difference is smaller than a predetermined tolerance (eg, 30 °). . If it is greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 517 to determine the two groups as different groups and returns to step 429 of FIG. 4.

If it is smaller than the predetermined tolerance, the main control unit 70 may determine a direction angle TM ₂ .d of the second feature point among the feature data of the registered fingerprint and a second feature of the corrected input fingerprint in step 511. The difference in the direction angle AM ₂ .d of the feature point is calculated, and it is checked whether the calculated difference is smaller than a predetermined tolerance (eg 30 °). If it is greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 517 to determine the two groups as different groups and returns to step 429 of FIG. 4.

If it is smaller than the predetermined tolerance, the main control unit 70 generates a third of the feature data of the corrected input fingerprint and the direction angle TM ₃ .d of the third feature point among the feature data of the registered fingerprint in step 513. The difference in the direction angle AM ₃ .d of the feature point is calculated, and it is checked whether the calculated difference is smaller than a predetermined tolerance (eg, 30 °). If greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 517 to determine the two groups as different groups, and returns to step 429 of FIG. 4. If less than the predetermined tolerance, the main control unit 70 proceeds to step 515 and determines the two groups as the same group, and returns to step 427 of FIG. 4.

FIG. 6 shows a detailed procedure of the polygon expanding section (step 427) described in FIG. First, prior to the description of the algorithm, the number m of the same feature points is set to '3'. The reason why m is set to '3' is because three identical feature points (same click) are currently detected.

Referring to FIG. 6, when detecting the same group, the main control unit 70 sets the current state (STATE) to NR in step 601, and the main control unit 70 checks in a polygon in which the same feature points are the vertices in step 603. Check to see if there is a segment to do. Here, the segment inspection is to select a segment consisting of two vertices, and the registration fingerprint and the input fingerprint among the feature points existing on the outer portion of the selected segment (the opposite side of the polygon with respect to the extension line after extending the segment: see FIG. 10). It shows the operation of finding the same feature point between them.

If there is a line segment to be inspected, the main control unit 70 proceeds to step 605 to select a particular line segment from the remaining line segments according to a predetermined rule (clockwise or counterclockwise). In operation 607, the main control unit 70 detects the same feature point existing in both the registration fingerprint and the input fingerprint at the outer portion of the selected line segment. This will be described in detail later with reference to FIG. 7.

If the same feature point is not found, the main control unit 70 returns to step 603 to perform the following process again. If the same feature point exists, the main control unit 70 registers the found feature point as a vertex of the polygon in step 609. In operation 611, the main controller 70 sets the current state to R, increases the number m of the same feature points, and returns to step 603 to perform the following process again.

On the other hand, if it is determined in step 603 that there are no more line segments of the polygon to be searched, the main control unit 70 proceeds to step 613 to check the current state, and in step 615, the current state is NR. Determine if it is or R. If the current state is R, the main controller 70 makes the current polygon a convex polygon based on the found identical feature point in step 619, and returns to step 601 to perform the following process again. If the current state is NR, the main control unit 70 proceeds to step 617 and outputs the same number of feature points m, and returns to step 429 of FIG. 4.

FIG. 7 illustrates a detailed procedure of the same feature point detection process (step 607) described with reference to FIG. The main controller 70 checks whether the conditions described in FIG. 7 are satisfied from the feature point (combination) closest to the selected line segment. If a feature point (or the same feature point) that satisfies all the conditions of FIG. 7 is detected, the process returns to step 609 of FIG. 6.

Referring to FIG. 7, first, in operation 701, the main control unit 70 includes the number of ridges between a predetermined vertex (one of two vertices connecting the selected line segment: a starting point) and an arbitrary first feature point of the registration fingerprint polygon. It is searched whether the difference in the number of ridge number information A.RC ₁ between RC ₁ ) and a predetermined vertex of the input fingerprint polygon and an optional second feature point is smaller than a predetermined tolerance (eg, 1). If greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 715 and determines the first feature point and the second feature point as different feature points, and then proceeds to step 717.

If it is smaller than the predetermined tolerance, the main control unit 70 proceeds to step 703 in which a predetermined vertex of the registration fingerprint polygon (the other one of the two vertices connecting the selected line segment: an end point) and the first feature point. the test is less than: ridge count information (T.RC ₂₎ the difference between the ridge count information (A.RC ₂₎ between the vertices of a predetermined input fingerprint polygon and the second feature point is a predetermined tolerance (n = 1). If greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 715 and determines the two feature points as different feature points, and then proceeds to step 717.

If it is smaller than the predetermined tolerance, the main control unit 70 proceeds to step 705 where two feature points forming a line segment (the selected line segment) of the registered fingerprint polygon and triangles of the triangle forming the first feature point (T). A, T. B, T. B: Two feature points forming a first predetermined internal angle T. XA and a line segment of the input fingerprint polygon (the selected line segment) and The difference between the predetermined first angle A.∠A among the triangle angles (A.∠A, A.∠B, A.∠C) of the second characteristic point is greater than the predetermined tolerance (eg, 30 °). If it is greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 715 and determines the two feature points as different feature points, and then proceeds to step 717.

If it is smaller than the predetermined tolerance, the main control unit 70 proceeds to step 707 in which two feature points forming one line of the registered fingerprint polygon and a second cabinet angle of the triangles forming the first feature point. 차이 B), the difference between the two feature points constituting a line segment of the input fingerprint polygon and the second inner angle A. ∠B among the angles of the triangle constituting the second feature point is less than a predetermined tolerance (eg, 30 °). Check for small If greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 715 and determines the two feature points as different feature points, and then proceeds to step 717.

If it is smaller than the predetermined tolerance, the main control unit 70 proceeds to step 709 in which a third interior angle T. of two triangles forming one line of the registration fingerprint polygon and the triangles forming the first feature point. 차이 C) and the difference between the two feature points that form a line segment of the input fingerprint polygon and the third inside angle A.∠C of the triangles that form the second feature point are less than a predetermined tolerance (eg, 30 °). Check for small If greater than or equal to the predetermined tolerance, the main control unit 70 proceeds to step 715 and determines the two feature points as different feature points, and then proceeds to step 717.

If it is smaller than the predetermined tolerance, the main control unit 70 proceeds to step 711 and the direction angle TCd of the first feature point and the input from the center of the circumscribed circle of the three feature points of the registration fingerprint. The difference of the direction angle ACd of the second feature point is calculated from the center of the circumscribed circle of the triangle formed by the three feature points of the fingerprint, and it is checked whether the calculated difference is smaller than a predetermined tolerance (eg, 30 °). If greater than or equal to the tolerance, the main controller 70 proceeds to step 715 and determines the two feature points as different feature points, and then proceeds to step 717.

If it is smaller than the predetermined tolerance, the main control unit 70 checks whether there is a feature point (combination of two feature points) to be further examined in step 717. If there are no more combinations to check, the main control unit 70 continues to return to step 701 to check again whether the predetermined feature point of the registered fingerprint and the predetermined feature point of the input fingerprint are the same. If there are no more combinations to check, the main control unit 70 proceeds to step 719 and checks whether at least one identical feature point is detected. If none of the same feature points are detected, the main control unit 70 returns to step 603 of FIG. 6. If at least one feature point is detected, the main control unit 70 proceeds to step 721 and sets the feature point having the least error accumulation among the detected feature points as the same feature point, and then returns to step 609 of FIG. 6.

As described above, the present invention compares two fingerprint images using ridge number information, which is a feature amount independent of rotation, transition, and resolution change of a fingerprint image. There is an advantage to guarantee. That is, the performance and reliability of fingerprint recognition can be improved.

Claims (11)

In the method for extracting the number of ridges existing between two feature points in the fingerprint image, A first process of generating an imaginary straight line connecting the first feature point and the second feature point, A second process of labeling a predetermined length along the ridge at the first and second feature points; A third process of dividing the virtual straight line into a plurality of segments; A fourth process of selecting a specific segment in a sequential order among the plurality of segments; A fifth step of determining whether the pixel value of the selected segment is labeled and labeled; A sixth step of determining whether the ridge is examined by inspecting the pixel value if the label is not labeled; In the case of the ridge, the seventh process of determining the coordinate of the segment position as the intersection, and increases the number of ridges, An eighth process of labeling along the ridge at the intersection and returning to the fourth process, And upon completion of the inspection of the plurality of segments, a ninth step of removing all labelings on the ridges and determining the number of accumulated ridges so far as the number of ridges existing between the two feature points. . The method of claim 1, Wherein said predetermined length is at least twice the mean ridge spacing. The method of claim 1, The determined number of ridges is stored in a triangular matrix data structure having the indexes of the feature points in rows and columns. The method of claim 1, A tenth step of checking whether the pixel value of the selected segment position is undetermined; And if it is undeterminable, further comprising eleventh removing all labeling on the ridge. In the fingerprint recognition method, Converting an analog fingerprint image from a fingerprint input sensor into a digital fingerprint image; Generating a thinned image by preprocessing the digital fingerprint image; Extracting minutia from the thinned image; Extracting the number of ridges existing between the feature points; Generating feature variable data of an input fingerprint of a specific user based on the extracted feature points and information on the number of ridges; Reading feature data on the registered fingerprint corresponding to the specific user; A group consisting of three feature points is generated with each of the feature data of the input fingerprint and the feature data of a registered fingerprint, and between the three feature points and the shape information of a triangle of three feature points for each group. Generating feature data including ridge number information of the; Comparing feature data of the input fingerprint with feature data of the registered fingerprint and checking whether there is an identical triangle between the two fingerprints; When the same triangle exists, searching for the same feature point existing in both the registration fingerprint and the input fingerprint based on the same triangle, and repeating the operation of extending the triangle into a polygon by using the searched same feature point as a vertex. Obtaining a number of feature points, And determining whether the registered fingerprint and the input fingerprint are the same with the obtained number of identical feature points. The method of claim 5, The feature data includes coordinates and types of three feature points, a sum of squares of ridge number information existing between the three feature points, and three points formed by three feature points from the center of a circumscribed circle of a triangle formed by the three feature points. And angle information of the three feature points. According to claim 5, The ridge count extraction process, A first process of generating an imaginary straight line connecting the first feature point and the second feature point, A second process of labeling a predetermined length along the ridge at the first and second feature points; A third process of dividing the virtual straight line into a plurality of segments; A fourth process of selecting a specific segment in a sequential order among the plurality of segments; A fifth step of determining whether a label is inspected by inspecting pixel values of the selected segment position; A sixth step of determining whether the ridge is examined by inspecting the pixel value if the label is not labeled; In the case of the ridge, the seventh process of determining the coordinate of the segment position as the intersection, and increases the number of ridges, An eighth process of labeling along the ridge at the intersection and returning to the fourth process, And upon completion of the inspection of the plurality of segments, a ninth step of removing all labelings on the ridges and determining the number of accumulated ridges so far as the number of ridges existing between the two feature points. . The method of claim 6, wherein the same triangle inspection process, Checking whether the sum of squares of the ridge number information in the arbitrary feature data of the registration fingerprint and the sum of the squares of the ridge number information in the arbitrary feature data of the input fingerprint is less than a first predetermined tolerance; Checking whether the difference between the first angle in the feature data of the registration fingerprint and the first angle in the feature data of the input fingerprint is smaller than the predetermined second allowable error, if the value is smaller than the predetermined first allowable error; If less than the predetermined second allowable error, checking whether the difference between the second angle between the feature data of the registration fingerprint and the second angle between the feature data of the input fingerprint is smaller than the predetermined third allowable error; If it is smaller than the predetermined third allowable error, the rotation angle of the input fingerprint is calculated by comparing the direction angle of a predetermined feature point in the feature data of the registered fingerprint and the direction angle of a predetermined feature point in the feature data of the input fingerprint, and the rotation angle Correcting the characteristic data of the input fingerprint based on; After correcting the feature data, checking whether a difference between a direction angle of a first feature point of the feature data of the registered fingerprint and a first feature point of the feature data of the corrected input fingerprint is smaller than a predetermined fourth allowable error; Checking whether the difference between the direction angle of the second feature point in the feature data of the registered fingerprint and the second feature point in the feature data of the corrected input fingerprint is smaller than the predetermined fifth allowable error, if it is smaller than the predetermined fourth allowable error. and, Checking whether the difference between the direction angle of the third feature point in the feature data of the registered fingerprint and the third feature point in the feature data of the corrected input fingerprint is smaller than the sixth allowable error, if it is smaller than the predetermined fifth allowable error. and, And determining that the triangle corresponding to the feature data of the registered fingerprint and the triangle corresponding to the feature data of the input fingerprint are the same when the predetermined sixth tolerance is smaller than the sixth allowable error. The method of claim 5, wherein the obtaining of the same feature point number comprises: A first process of initializing the number of identical feature points to '3', A second process of selecting specific line segments in a sequential order among polygon segments having the same feature points as vertices, A third process of detecting the same feature point present in both the registration fingerprint and the input fingerprint outside the selected line segment; A fourth process of increasing the number of identical feature points, registering the detected same feature points as vertices of a polygon, and returning to the second process if the same feature points exist; A fifth process of checking whether the same feature point registered as a polygon vertex is completed when the same feature point detection is completed for an arbitrary polygon; A sixth process of returning to the second process after expanding the arbitrary polygon by using the registered feature as a vertex, when the registered same feature point exists; And if the registered identical feature point does not exist, determining a number of identical feature points accumulated so far as the number of identical feature points detected based on the same triangle. The method of claim 9, wherein the fifth process, The difference between the ridge number information between the start point of the selected line segment and the arbitrary first feature point in the registration fingerprint polygon and the ridge number information between the start point of the selected line segment and the optional second feature point in the input fingerprint polygon is predetermined. To check if it is less than the error, If less than the seventh allowable error, the number of ridges between the end point of the selected line segment and the first feature point in the registration fingerprint polygon and the number of ridges between the end point of the selected line segment and the second feature point in the input fingerprint polygon. Checking whether the difference in information is less than a predetermined eighth error, Less than the predetermined eighth tolerance, a triangle formed by the first interior angle of the triangle formed by the selected line segment and the first feature point and the second feature point by the input fingerprint polygon Checking whether the difference between the first cabinet of the cabinets is smaller than the eighth tolerance; If less than the eighth allowable error, checking whether a difference between a second cabinet angle among the triangles of the registration fingerprint and a second cabinet angle of the triangles of the input fingerprint is smaller than a predetermined ninth tolerance; , If it is less than the predetermined ninth tolerance, checking whether a difference between a third cabinet angle among the triangles of the registration fingerprint and a third cabinet angle among the triangles of the input fingerprint is smaller than the tenth tolerance; , And determining the first feature point and the second feature point as the same feature point when the size is smaller than the predetermined tenth tolerance. The method of claim 5, wherein the determination process, Determining a value obtained by dividing the obtained number of identical feature points by the minimum of the number of feature points of a registration fingerprint and the number of feature points of an input fingerprint as a similarity; And comparing the similarity with a predetermined reference value to determine the same.