KR20200109977A - Smartphone-based identity verification method using fingerprints and facial images - Google Patents

Abstract

The present invention relates to a smartphone-based identity verification method using fingerprint and face images, which performs identification by acquiring a fingerprint image and a face image of a subject from a camera built in a police smartphone, while making easy and convenient for a frontline police to use, and increasing the speed and accuracy of identification. The smartphone-based identity verification method of the present invention comprises: a subject fingerprint obtaining step photographing a fingerprint image of the subject to transmit the image to a police smartphone by a fingerprint obtaining assistant device operated in conjunction with the police smartphone and having a camera; a fingerprint feature detection step of the police smartphone detecting a fingerprint feature point from the fingerprint image of the subject received from the subject fingerprint obtaining step and transmitting the fingerprint image of the subject and the fingerprint feature point to a matching server through a gateway server by the police smartphone; and an identity analysis step of the matching server matching the fingerprint feature point of the subject received from the fingerprint detection step and matching fingerprint feature points of previously stored plurality of comparison targets in a database unit to perform identification verification by the matching server. In addition, the fingerprint obtaining assistant device has a thimble shape, and includes a camera at the contact portion of the bottom surface of the finger.

Description

Smartphone-based identity verification method using fingerprints and facial images}

The present invention performs identification by acquiring a fingerprint image and an image of a subject's face from a camera built into a police smartphone, but it is easy and convenient for the frontline police to use, while increasing the identification speed and accuracy, fingerprint and face images. It relates to a smartphone-based identification method used.

If a police officer working at the front line finds a drunkard, an elderly person with dementia, and a person suspected of being a lost child, a criminal, illegal alien, wanted person, criminal, etc., at present, they will accompany them to the police station and then use the computer at the police station. Through the process, the identification process is performed.

In some cases, it is inconvenient to have people who do not care to go to the police station to verify their identity.

Problems that continue to be an issue in the field of fingerprints are the deterioration of fingerprint recognition and too low image quality, and image improvement problems occur when there is no direction information due to too low image quality. Although there are several types of fingerprints, all fingerprints have a certain pattern, so it is possible to improve the image quality by analyzing the pattern of the image and improving direction information.

Therefore, by using a smartphone or mobile device (laptop, PDF, etc.) carried by the police officer himself, it is possible to obtain a fingerprint of the subject and immediately verify the identity at the site, but an identification system with improved fingerprint image quality is desired.

In addition, fingerprint recognition technology and face recognition technology can be employed to verify the identity of the subject using a police officer's smartphone (hereinafter referred to as a police smartphone).

Fingerprint recognition technology is a technology that recognizes a user by acquiring a digital image of a fingerprint using a dedicated sensor, and includes optical fingerprint recognition, capacitive fingerprint recognition, and ultrasonic fingerprint recognition.

In the case of the national DB, since fingerprints are stored for all 10 fingers, approximately 500 million fingerprints must be matched, so even if 100,000 matches per second, it takes a large amount of 80-100 minutes, so speed is important. Element.

Face recognition technology is a technology that analyzes and recognizes about 50 areas where the facial skeleton changes, such as nose, mouth, eyebrows, and chin, rather than the entire face. Facial images are sensitive to lighting, environment, and angle of images, and have disadvantages in distinguishing disguises, facial changes that occur over time, plastic surgery, and similar facial features of twins.

The present invention requires an identification system that can verify the identity of the subject more quickly and accurately in the field by acquiring an image of the subject's fingerprint and face using a smartphone or mobile device (laptop, PDF, etc.) carried by the police officer himself. do.

As a prior art, Korean Utility Model Publication No. 20-0283571 relates to a portable identification device that verifies the identity of the person by scanning the biometric fingerprint of the ID card or ID card holder and comparing it with information registered in the administrative computer network. In this case, since it is recognized only as a fingerprint, it is impossible to check if the fingerprint is erased.

As another prior art, Korean Unexamined Patent Publication No. 2003-0086527 relates to a police radio that can verify whether the identity checker is a party by comparing the fingerprint corresponding to the identity checker's resident registration number with the fingerprint input by the identity checker. , Even in this case, since the fingerprint is recognized only as a fingerprint, it is impossible to check if the fingerprint is erased.

As another prior art, Korean Laid-Open Patent Publication No. 2002-0011577 stores the identity of users on the Internet, data such as fingerprints and facial photos in a verification database in advance, and then inputs fingerprints and facial photos of users, etc. It relates to a user identification system on the Internet, which is configured to compare and check the data of a user. In general, it takes a lot of time to compare and analyze a user's fingerprint and facial photos.

The problem to be solved by the present invention is to perform identification by acquiring a fingerprint image and an image of a subject's face from a camera built into a police smartphone, but it is easy for the front line police to use, while increasing the identification speed and accuracy, It is to provide a smart phone-based identification method using fingerprint and face images.

In order to solve the above problems, the smart phone-based identification method of the present invention, in the fingerprint image of the subject photographed by the camera of the police smartphone, the police smartphone detects a fingerprint feature point and connects the fingerprint image and the fingerprint feature point of the subject to the gateway. The fingerprint information detection step of the police smartphone transmitted to the matching server through the server; An identity analysis step of a matching server, wherein the matching server performs identification by matching the fingerprint feature points of the subject received in the fingerprint detection step with the fingerprint feature points of the subject to be compared previously stored in the database unit; It characterized in that it comprises a.

In addition, the smart phone-based identification method of the present invention is operated in conjunction with a police smart phone and a fingerprint acquisition auxiliary device equipped with a camera takes a fingerprint image of the subject and transmits it to the police smart phone, the subject fingerprint acquisition step; In the fingerprint image of the subject received from the subject's fingerprint acquisition step, the police smartphone detects a fingerprint feature point and transmits the subject's fingerprint image and fingerprint feature point to a matching server through a gateway server, the fingerprint information detection step of the police smartphone; And an identity analysis step of the matching server, in which the matching server performs identification by matching the fingerprint feature points of the subject received in the fingerprint detection step and a plurality of fingerprint feature points to be compared previously stored in the database unit. do.

The fingerprint acquisition auxiliary device has a thimble shape, and includes a camera at the contact portion of the bottom of the finger.

Fingerprint feature points include end and branch points of the ridge of the fingerprint.

The fingerprint information detection step of the police smartphone is a finger fingerprint image receiving step in which the operation processing unit of the police smartphone receives the finger fingerprint image captured from the camera of the police smartphone or a fingerprint acquisition auxiliary device linked to the police smartphone. ; The police smartphone determines whether the finger fingerprint image captured in the finger fingerprint image reception step is a finger fingerprint or not, and compares it with the color of a previously stored finger, and determines whether it is a finger fingerprint, and if it is not Or a finger color comparison step of notifying that it is not a finger print by voice and returning to the step of receiving a finger print image; After the finger color comparison step, the operation processing unit of the police smartphone extracts the outline of the finger from the fingerprint image of the finger and determines whether the outline of the finger is smaller than the finger photographing guideline and within the finger photographing guideline and the limit point, A step of determining whether to input a fingerprint image, informing that the deviation from the finger photographing guideline is notified by text or voice and returning to the step of receiving a fingerprint image of the finger when it is out of the threshold or is smaller than the finger photographing guideline; In the step of determining whether to input a fingerprint image, the operation processing unit of the police smartphone, if the outline of the finger is greater than or equal to the finger photographing guideline, but within the limit of the finger photographing guideline, the fingerprint image is 100px×100px around the fingerprint guideline. It is extracted as an image of and detects horizontal and vertical ridges from the fingerprint image, and determines whether the horizontal and vertical ridges form a pattern by repeating a certain portion of each, and if it is determined that the pattern is not formed, text or voice And, informing that it is not a finger fingerprint image, and returning to the finger fingerprint image receiving step, a ridge check step.

The fingerprint information detection step of the police smartphone is, after the ridge check step, the operation processing unit of the police smartphone detects fingerprint feature points including end points and branch points of the ridge of the fingerprint from the fingerprint image of the subject. It further includes a; detecting step.

The fingerprint feature detection step includes: a pre-processing step of thinning the fingerprint image of the subject; The operation processing unit of the police smartphone comprises: a candidate fingerprint feature point detection step of detecting a candidate fingerprint feature point from the fingerprint image thinned in the pre-processing step; The operation processing unit of the police smartphone includes a post-processing step of removing the pseudo feature point from the feature point detected in the candidate fingerprint feature point detection step.

In the step of analyzing the identity of the matching server, the matching server calculates the rotation and transition amount of the fingerprint feature points of the subject and the fingerprint feature points to be compared previously stored in the database. An alignment step of selecting an alignment reference point, converting coordinates of the feature points to match the alignment reference point, and determining a corresponding feature point pair; The matching server calculates a similarity by using coordinates, types, and angle information of the pair of corresponding feature points determined in the alignment step; The matching server includes a score acquisition step of outputting the matching degree of the fingerprint feature point of the subject and the fingerprint feature point to be compared previously stored in the database as a score, according to preset rule data, from the similarity of the corresponding feature point pair. .

In the identity analysis step of the matching server, before the alignment step, the matching server obtains the ridge spacing for the fingerprint image of the subject received from the operation processing unit of the police smartphone and normalizes the ridge spacing using the maximum and minimum values of the ridge spacing. By doing, standardizing, ridge spacing standardization step; After the ridge interval standardization step, the matching server detects the ridge direction information from the subject's fingerprint image, and determines that the ridge direction information is incorrect when there are no two or more consecutive ridge directions in any one of the front, back, left and right. And, by correcting the wrong ridge direction information in the direction of any one of the successive ridges before, after, left, and right of the ridge having the erroneous ridge direction information, correcting the fingerprint feature points of the subject; further Include.

It features a recording medium storing a smartphone program source for the smartphone-based identification method of the present invention.

The smart phone-based identification method of the present invention performs identification by acquiring a fingerprint image and an image of the subject's face from a camera built into the police smart phone, but it is easy to use for the front line police, and the speed and accuracy of identification are more improved. It is made to be raised. That is, in the present invention, by using a smartphone or mobile device (laptop, PDF, etc.) carried by the police officer himself, it is possible to acquire an image of the subject's fingerprint and face, and to verify the identity more quickly and accurately in the field.

In particular, it is possible to establish a database for drunkards, lost children, or dementia elderly, and to check the identity of the subject in one stop by linking with the database.

1A is a schematic diagram schematically illustrating the configuration of an identification system to which a smartphone-based identification method using a fingerprint and a face image of the present invention is applied.
1B is a flowchart schematically illustrating a smart phone-based identification method using a fingerprint and a face image of the present invention.
2 is an example of a fingerprint acquisition auxiliary device applied to the smartphone of FIG. 1.
3 is an explanatory diagram for explaining image improvement according to direction information.
4 is an explanatory diagram for explaining distortion of a general fingerprint image.
5 is an explanatory diagram for explaining change of coordinates of feature information according to determination of a rotation angle.
6 is an example of outputting coincident feature points between fingerprints to be compared and connection information between feature points.
7 is an explanatory diagram for explaining using the similarity of feature information around the feature information when checking the similarity of feature information.
8 is an explanatory diagram schematically illustrating a fingerprint recognition process in a matching server.
9 is an example of photographing the index finger of the right hand on the fingerprint screen of the police smartphone of the present invention.
10 is a diagram illustrating whether a finger selected on the fingerprint photographing screen of FIG. 9 matches a finger photographing guideline.
11 is a flowchart schematically showing a driving process of a police smart phone in the smart phone-based identification method of the present invention.
12 is a flowchart illustrating a feature information detection step of FIG. 11.
13 is a flowchart schematically illustrating an identification process in the matching server 300 in the smart phone-based identification method of the present invention.

Hereinafter, an identification system using a smart phone of the present invention will be described in detail with reference to the accompanying drawings.

In the present invention, a smartphone or mobile device (laptop, PDF, etc.) carried by the police officer himself is referred to as a'police smartphone'.

1A is a schematic diagram for schematically explaining the configuration of an identification system applying a smart phone-based identification method using a fingerprint and a face image of the present invention, and FIG. 1B is a smartphone-based diagram using a fingerprint and a face image of the present invention. It is a flow chart schematically explaining the identification method.

The police smartphone (police officer's smartphone) 100 is a smartphone in which a predetermined application is installed through a specific registration process such as authentication, and captures an image of the subject's fingerprint or face on the police smartphone 100 ( S10), detecting a feature point of a fingerprint or a face from an image of a fingerprint or face (S20), selecting an identification group, and selecting an identification request, detects a feature point in the fingerprint image or image of the face of the subject, and The confirmation request signal, the identification group signal, the fingerprint feature points, and the facial feature points are transmitted to the gateway (G/W) server 200 (S30).

Here, the identification group can be input by selecting an age group, a gender, and the like, and also, a drunkard, a dementia elderly, a lost child, a criminal, a foreigner, a wanted person, a criminal, and the like can be set.

Although the present invention can acquire a fingerprint without a fingerprint acquisition auxiliary device, a fingerprint acquisition auxiliary device may be used in some cases.

In the case of fingerprint recognition, in the case of contact technology, the resolution is fixed at 500 dpi, but in general, in the case of a smartphone, since the resolution varies according to the distance by the point of time at which the subject is photographed, there may be a high rate of misrecognition during matching.

In general, when acquiring a fingerprint on the fingerprint sensor, the shape of the fingerprint is deformed according to the pressure at the position where the force of the finger is applied.In order to solve this, the fingerprint is acquired through a camera, but if a specific fingerprint assist device is used, the acquired image has a certain resolution. Resolution compatibility can be secured by making it available.

In some cases, instead of the police smartphone 100 and the fingerprint acquisition auxiliary device, the police smartphone 100 may receive and use a fingerprint image acquired using a live scanner, which is an existing fingerprint input device. .

The gateway (G/W) server 200 receives the identification request signal, the identification group signal, the fingerprint characteristic points, and the facial characteristic points received from the police smartphone 100 and transmits the received to the matching server 300 (S35). ).

The matching server 300 performs identification by matching and comparing the fingerprint feature points of the identification group previously stored in the database unit 370 with the fingerprint feature points of the subject received through the gateway server 200, and , The facial feature points of the identification group previously stored in the database unit 370 and the facial feature points of the subject received through the gateway server 200 are matched and compared to perform identification (S50), and the result is a gateway It transmits to the police smartphone 100 through the server 200 (S60).

2 is an example of a fingerprint acquisition auxiliary device applied to the smartphone of FIG. 1.

The fingerprint acquisition auxiliary device 170 of FIG. 2 has a finger clip, that is, in the form of a thimble, and includes a fingerprint detection sensor 177 at a portion where the bottom of the finger is closed.

The housing portion of the fingerprint acquisition auxiliary device 170 is formed in a finger clip that applies horizontal pressure to the finger by surrounding the front and rear surfaces of the subject's finger, that is, in the form of a thimble, and when the finger is inserted into the acquisition auxiliary device, a fingerprint detection sensor A finger print is formed by 177.

The fingerprint detection sensor 177 is a means for acquiring image information of a finger fingerprint, and detects the fingerprint image and transmits it to an operation processing unit (not shown) of the police smartphone 100. The fingerprint detection sensor 177 It can be a camera or an image sensor.

3 is an explanatory diagram for explaining image improvement according to direction information, FIG. 4 is an explanatory diagram for explaining distortion of a general fingerprint image, and FIG. 5 is a description for explaining change of feature information coordinates according to determination of a rotation angle 6 is an example of outputting matching feature points between fingerprints to be compared and connection information between feature points, and FIG. 7 is an explanatory diagram for explaining the use of feature information similarity around corresponding feature information when checking feature information similarity. .

FIG. 3(a) shows an example of low image quality, and in FIG. 3(a), the green line is direction information that should actually be displayed (ie, direction information displayed by an expert), and FIG. 3(b) is As a conventional analysis method, as an example of analyzing the direction information in the fingerprint image of FIG. 3(a), it can be seen that the direction information is properly wrong in the part within the square of FIG. 3(a). That is, there is a problem with the direction information. FIG. 3C is an example of correcting the fingerprint image to improve direction information in the fingerprint image of FIG. 3A.

That is, the present invention analyzes a pattern of an image and improves the image quality through direction information improvement.

Figure 4 (a) is an example of downward twist, Figure 4 (b) is an example of upward twist, Figure 4 (c) is an example of right twist, Figure 4 (d) is an example of left twist to be.

That is, when the subject takes a fingerprint, image distortion occurs according to the pressure of the finger, that is, the direction in which the force of the finger is applied. Therefore, the matching server 300 detects the distortion direction of the image when comparing the fingerprint and corrects it so that the image fits a square.

In the present invention, the matching server 300 multiplexes rotation and displacement selection to improve the deformed image matching performance.

FIG. 5(a) shows an example of setting the angle change of the input fingerprint image, and FIG. 5(b) shows the fingerprint image whose coordinates are changed according to the angle change of FIG. 5(a).

The matching server 300 attempts to match using a plurality of rotation values because the rotation angle may be different depending on the image position when the rotation angle determination has a large effect on the feature information coordinates and the image deformation is large.

Figure 6 (a) is an example of a fingerprint stored in the database, and Figure 6 (b) is an example of a fingerprint of the subject, and the matching feature points between the two fingerprints are represented by a red line, and the connection information between the feature points is blue It is represented by a line.

Visual fingerprint recognition result expression information is obtained from the police smartphone 100, and the matching server 300 receives it and uses it as an aid when determining whether the fingerprint is the same. That is, when it is difficult to determine whether the matching server 300 is the same fingerprint, it may determine by using visual fingerprint recognition result display information.

As shown in FIG. 6, it is possible to know whether or not the fingerprint is the same by outputting the matching feature points between the fingerprints to be compared and the connection information between the feature points. Even if the distance between the feature information or the value of the feature point direction is changed due to the fingerprint deformation, the deviation is small because the inspection feature point and the surrounding information move together. Therefore, as shown in FIG. 7, when checking the similarity of the feature information, the similarity of feature information around the corresponding feature information is considered.

7(a) is the peripheral feature points (M1 to M5) of the feature point (Mr) in the fingerprint stored in the database, and FIG. 7(b) shows the peripheral feature points (M1 to Ms) of the subject fingerprint to be corresponded. In M5), the similarity of the feature points around the feature point to be compared is reviewed, and the similarity of feature information is detected using the similarity review result.

In the present invention, the matching server 300 uses index information to compare 1:N to improve the fingerprint recognition speed, so the speed is faster than the case of applying other algorithms (for example, analysis of about 100,000 people per second). Since the fingerprint type, which is a static classification, has a limited type, the present invention performs a high-speed search by dynamically selecting an object.

8 is an explanatory diagram schematically illustrating a fingerprint recognition process in a matching server.

In the fingerprint feature information processing step (S200), the fingerprint feature information is processed for high-speed search.

In the similarity test and preliminary target selection step (S300), the similarity is checked from the feature information processed in the fingerprint feature information processing step (S200), and the fingerprint feature information stored in the database unit is sorted according to the similarity score, and the sorted database unit Among the fingerprint feature information stored in, a preliminary target determined to be similar to the subject's fingerprint feature information is selected. In the similarity test and preliminary target selection step (S300), the number (size) of preliminary targets (candidate) is determined in consideration of the DB size of the test target group stored in the database unit.

9 is an example of photographing the index finger of the right hand on the fingerprint photographing screen of the police smartphone of the present invention, and FIG. 10 is a diagram illustrating whether a finger selected on the fingerprint photographing screen of FIG. 9 matches the finger photographing guidelines.

9A shows an example of selecting the right index finger on the fingerprint photographing screen, and clicking the right index finger selection button in the right and left hand drawings at the bottom of the fingerprint photographing screen. In the middle of the fingerprint photographing screen, a finger photographing guideline is displayed, and a fingerprint guideline (in the case of FIG. 9A, a blue circle) is displayed in the finger photographing guideline. Mark the limits (triangles) of the range that can deviate from this guideline. In addition, a message is displayed on the upper part of the fingerprint photographing screen, instructing to proceed with photographing by placing a selected finger in the finger photographing guideline.

9B is an example of a photograph of a fingerprint photographing screen by touching the right index finger to the screen.

Figure 10 (a) is an example of a case in which the right index finger selected on the fingerprint photographing screen deviates from the finger photographing guidelines, and FIG. 10 (b) shows a case where the right index finger is smaller than the finger photographing guidelines in the fingerprint photographing screen. As an example, (c) of FIG. 10 is an example of a case in which the index finger of the right hand coincides with the finger photographing guideline on the fingerprint photographing screen.

As shown in (c) of FIG. 10, if the boundary line of the finger is positioned between the finger photographing guideline and the limit point, the fingerprint to be photographed is properly positioned within the fingerprint guideline.

In the present invention, a mobile-only app was developed in order to acquire a fingerprint image from the camera of the police smart phone 100 or the fingerprint acquisition assistant 177. When acquiring a fingerprint image through a mobile app, the app presents a finger photographing guideline for the fingerprint image input part, and checks three items for the finger input in the guideline, that is, a finger fingerprint image through color comparison. A good quality finger fingerprint image can be obtained by checking whether or not a fingerprint image has been properly input in the guideline, and checking whether a fingerprint image is used through a ridge check.

11 is a flowchart schematically illustrating a driving process of a police smartphone in the smart phone-based identification method of the present invention, and FIG. 12 is a flowchart illustrating a feature information detection step of FIG. 11. Here, it can be said that FIG. 11 shows a fingerprint photographing and feature point detection process in a police smartphone.

In the finger fingerprint image receiving step, the operation processing unit of the police smartphone 100 receives the finger fingerprint image captured from the camera of the police smartphone or the fingerprint acquisition auxiliary device 170 interlocked with the police smartphone (S110). ).

In the finger color comparison step, the calculation processing unit of the police smartphone 100 compares whether the finger fingerprint image photographed in the finger fingerprint image reception step is a fingerprint of a finger, and compares the color of a previously stored finger to determine whether it is a finger print or not. When it is determined that it is not a fingerprint of the finger (S120), it notifies that it is not a finger by text or voice (S125), and returns to the step of receiving a fingerprint image of the finger (S110).

In the step of determining whether to input a fingerprint image, after the finger color comparison step, the operation processing unit of the police smartphone 100 extracts the outline of the finger from the finger fingerprint image, and the corresponding finger outline is smaller than the finger photographing guideline, and It is determined whether it is within the shooting guideline and the limit point (S130), and if it is out of the limit point or is smaller than the finger shooting guideline, it is notified by text or voice (S135), and the process returns to the step of receiving a finger fingerprint image (S110). At this time, the operation processing unit of the police smartphone 100 determines that an appropriate fingerprint image is input within the fingerprint guideline if the outline of the finger is greater than or equal to the finger capture guideline, but is within the limit point with the finger capture guideline, and If it is out of or smaller than the finger photographing guideline, it is determined that an appropriate fingerprint image is not input in the fingerprint guideline.

In the ridge check step, in the step of determining whether to input a fingerprint image, if it is determined that the outline of the finger is greater than or equal to the finger shooting guideline, but is within the limit of the finger shooting guideline, and an appropriate fingerprint image is input within the fingerprint guideline, the police The operation processing unit of the smartphone 100 extracts the fingerprint image as a 100px×100px image centering on the fingerprint guideline, detects a horizontal ridge and a vertical ridge from the fingerprint image (S140), and detects the horizontal ridge and the vertical ridge. It is determined whether the directional ridges form a pattern by repeating a certain portion of each (S150), and if the horizontal ridges and the vertical ridges form a pattern by repeating a certain part, it is determined that it is a fingerprint image of the finger, and if the pattern is not formed, the finger It is determined that it is not a fingerprint image, it notifies that it is not a finger image by text or voice (S155), and returns to the step of receiving a finger fingerprint image (S110).

In the feature information detection step (S160), the operation processing unit of the police smartphone 100 extracts feature information from the finger fingerprint image, and the feature information detection step is a pre-processing step (S161), a feature point extraction step The process proceeds to step 3 of (S163) and post-processing (S167).

The pre-processing step (S161) is a step of thinning the fingerprint image, image enhancement is performed, the image-improved fingerprint image is binarized, and thinned. That is, the preprocessing step goes through a process of converting a block directional image, a binarized image, and a thinning image. The fingerprint image is divided into blocks of a certain size, and the direction representing the flow of the ridge is determined for each block and converted into a direction image for each block. The obtained block-by-block ridge direction information is used for binarization and smoothing processes. In the process of binarization, the fingerprint image is changed to a binary image that is expressed only in black and white, and the binary image is again smoothed to remove noise and emphasize the fusion, thereby improving the connectivity of the jungseon and turning the jungseon into a line of 1 pixel thickness. It is expressed and converted into a thinned image,

The feature point extraction step S163 is a step of extracting a candidate (preliminary target) feature point from the fingerprint image thinned in the preprocessing step. Here, the term “Minutiae” refers to a point in which a change occurs in the flow of a ridge, which is called a disadvantage (a point where the flow of a ridge is cut off) and a branch point (a point where one ridge splits into two). The feature quantity (T={m1, m2,, mm}) is composed of information on the feature point (mi) existing in the fingerprint image. In most cases, the information on the feature point used is the type of feature point that indicates whether it is a branch point or a weakness. , There is information on the position in the fingerprint image and the direction of the ridge where the feature point is located.

That is, the feature point extraction step is a process of extracting candidate feature points, and stores information on the disadvantages and branch points in which a change in the flow of ridges occurs from the thinning image. Candidate feature points including pseudo feature points are extracted from the ridge information of the thinning image. In this case, a pseudo feature point may be generated due to an incorrect ridge portion of the thinning image. A pseudo feature point refers to a fake feature point that occurs during the thinning process due to noise during fingerprint acquisition. The pseudo feature point increases the amount of insignificant computation and increases the error, which degrades the performance of the system.

The post-processing step (S167) is a step of removing a fake feature point from the feature point extracted in the feature point extraction step. That is, in the post-processing step, the ridge portion that is the cause of the pseudo feature point is modified to remove the pseudo feature point, and the missing feature point is added to extract the final feature point.

In the feature information transmission step, the calculation processing unit of the police smart phone 100 transmits the feature information consisting of candidate feature points detected in the feature information detection step and a fingerprint image to the matching server 300 through the gateway server 200 ( S170).

13 is a flowchart schematically illustrating an identification process in the matching server 300.

In the ridge spacing standardization step (S210), the ridge spacing is obtained from the fingerprint image received from the operation processing unit of the police smartphone 100 and the ridge spacing is normalized by using the maximum and minimum ridge spacing values. To have.

In the correction step (S220), by analyzing the ridge direction information from the fingerprint image received from the calculation processing unit of the police smartphone 100, the part in which the direction information is incorrect (for example, two or more consecutive ridge directions are the same. When there is no), the direction information is corrected (for example, in the direction of the previous ridge of the wrong part).

In the fingerprint matching step (S310), the matching server 300 uses a feature amount defined from feature point information extracted from the operation processing unit of the police smartphone 100 using a matching algorithm of a fingerprint recognition method using feature points. The step of determining the similarity between the two fingerprint images includes three steps of an image alignment step S320, a Mnutiae Matching step S330, and a scoring step S350.

In the alignment step (S320), an alignment reference point of the image is selected by calculating the amount of rotation and transition in which the feature points of the two fingerprint images overlap the most. Subsequently, the coordinates of the feature points are transformed to match the alignment reference point, and then the corresponding feature point pairs are determined. At this time, in order to find the ideal rotational transition amount, it is necessary to perform as many candidate alignments as possible.

In the matching step (S330), a measure vector is calculated using the coordinates, types, and angle information of the corresponding feature point pair determined in the alignment step. In order to realistically reflect the similarity of the pair of corresponding feature points, it can be considered from various angles with various statistical techniques in the method of determining the similarity.

In the scoring step (S350), the degree of coincidence between the two fingerprint images from the measure vector of the corresponding feature point pair is represented as a score according to preset rule data. Rule data is data previously determined by using statistical basis and mathematical modeling in the decision making of score calculation rules in order to accurately reflect the degree of matching of images as in the matching step.

In the preliminary target selection step (S360), when the score obtained according to the similarity of the corresponding feature point pair in the scoring step exceeds a predetermined threshold, the fingerprint feature information of the database unit is selected as a preliminary target, or the database with the highest score The negative fingerprint feature information is selected as a preliminary target, and it is delivered to the police smartphone through the gateway server.

The conventional device for obtaining a fingerprint image was acquired by using a live scanner, but in the present invention, the fingerprint image acquired from the camera of the existing live scanner or the police smartphone 100 or the camera of the fingerprint acquisition assistant 177 is compared. Confirm your identity.

However, in general, when a fingerprint image is acquired with a camera, if the distance from the camera to the finger is not constant, compatibility with the fingerprint image acquired with the existing equipment cannot be guaranteed. Therefore, the present invention obtains the interval of the ridge for the acquired fingerprint image in order to standardize the fingerprint image differently acquired according to the live scanner, the camera of the police smartphone 100, etc., and improves the image based on the standardized interval. A fingerprint template is created from the image.

Matching server 300 is MCU (matching controller units, matching control unit) (not shown), MPU (matching process units, matching processing unit) (not shown), MMU (Matching Monitoring Uni, matching monitoring unit) (not shown) Includes.

The MCU is a means for controlling the matching server 300 as a whole, and its main functions are fingerprint matching task management and MPU management.

The fingerprint matching task management function of the MCU is to fetch the fingerprint identification request information received from the G/W server 200, pre-stored data from the DB unit 370, and request a matching task to the MPU connected to the MCU. It collects the matching results of, determines the candidate, and delivers the result to the G/W server.

The MPU management function of the MCU monitors the state of the MPU through the MMU and manages the start and stop of the MPU. When a specific MPU process is stopped or a deadlock has occurred, the MCU can detect the status through MPU monitoring and then restart the process of the MPU.

The MPU analyzes the fingerprint matching task requested from the MCU to determine the matching target, compares the feature points of the requested fingerprint with the feature points of the target fingerprint, and transmits the target fingerprint information with a high priority to the MCU.

The MMU monitors the state of the MPU interlocked with the MCU and retrieves the operation log of the MPU. The MMU sends a status check message to the MPU, and the received MPU checks its status and sends it to the MMU whether it is operating normally. When the MMU does not normally receive a message from the MPU or receives a message in an abnormal state, the MMU transmits the operation status of the MPU to the MCU, and the MCU sends a message to restart the process to the MPU to restart the MPU function. .

In the present invention, the matching server and the database unit may be referred to as a fingerprint matching system.

The fingerprint data of the fingerprint matching system is managed in a distributed environment to optimize the performance of the MPU. Distributed processing of fingerprint data is performed in two ways: real-time distributed processing and batch distributed processing.

In real-time distributed processing, the fingerprint data that has been changed while the fingerprint matching system is operating is distributed to each MPU. When fingerprint data is changed due to operations such as input, modification, and deletion in the main fingerprint database, change monitoring is performed and fingerprint data is stored. When a change is detected, the changed fingerprint data is distributedly processed and audited to ensure data consistency.

Batch distributed processing can be performed when an initial fingerprint matching system is established, or when fingerprint data in the main database is batch-distributed to MPU when there are many changes in fingerprint data. Distributed processing of basic data and performing an audit ensures the integrity of fingerprint data.

As described above, although the present invention has been described by the limited embodiments and drawings, the present invention is not limited to the above embodiments, which is various modifications and variations from these descriptions to those of ordinary skill in the field to which the present invention belongs. Transformation is possible. Accordingly, the spirit of the present invention should be grasped only by the claims set forth below, and all equivalent or equivalent modifications thereof will be said to belong to the scope of the present invention.

100: police smartphone 170: fingerprint acquisition auxiliary device 177: fingerprint detection sensor
200: gateway server 300: matching server 370: database unit

Claims (10)

In the fingerprint image of the subject photographed by the camera of the police smartphone, the police smartphone detects the fingerprint feature point and transmits the fingerprint image and the fingerprint feature point of the subject to the matching server through the gateway server, the fingerprint information detection step of the police smartphone;
An identity analysis step of a matching server, wherein the matching server performs identification by matching the fingerprint feature points of the subject received in the fingerprint detection step with the fingerprint feature points of the subject to be compared previously stored in the database unit;
Characterized in that it comprises a, smart phone-based identification method. A fingerprint acquisition step of a subject, which is operated in conjunction with a police smartphone and in which a fingerprint acquisition assistant device equipped with a camera takes a fingerprint image of the subject and transmits it to the police smartphone;
In the fingerprint image of the subject received from the subject's fingerprint acquisition step, the police smartphone detects a fingerprint feature point and transmits the subject's fingerprint image and fingerprint feature point to a matching server through a gateway server, the fingerprint information detection step of the police smartphone;
An identity analysis step of a matching server, wherein the matching server performs identification by matching the fingerprint feature points of the subject received in the fingerprint detection step with the fingerprint feature points of the subject to be compared previously stored in the database unit;
Characterized in that it comprises a, smart phone-based identification method. The method of claim 2,
Fingerprint acquisition auxiliary device is formed in the form of a thimble, characterized in that the camera is provided in the contact portion of the bottom of the finger, smart phone-based identification method. The method according to any one of claims 1 or 2,
The fingerprint feature point comprises an end point and a branch point of a ridge of the fingerprint. The method of any one of claims 1 or 2, wherein the step of detecting fingerprint information of a police smartphone,
A finger fingerprint image receiving step of receiving a finger fingerprint image captured from a camera of a police smartphone or a fingerprint acquisition auxiliary device interlocked with the police smartphone;
The police smartphone determines whether the finger fingerprint image captured in the finger fingerprint image reception step is a finger fingerprint or not, and compares it with the color of a previously stored finger, and determines whether it is a finger fingerprint, and if it is not Or a finger color comparison step of notifying that it is not a finger print by voice and returning to the step of receiving a finger print image;
After the finger color comparison step, the operation processing unit of the police smartphone extracts the outline of the finger from the fingerprint image of the finger and determines whether the outline of the finger is smaller than the finger photographing guideline and within the finger photographing guideline and the limit point, A step of determining whether to input a fingerprint image, informing that the deviation from the finger photographing guideline is notified by text or voice and returning to the step of receiving a fingerprint image of the finger when it is out of the threshold or is smaller than the finger photographing guideline;
In the step of determining whether to input a fingerprint image, the operation processing unit of the police smartphone, if the outline of the finger is greater than or equal to the finger photographing guideline, but within the limit of the finger photographing guideline, the fingerprint image is 100px×100px around the fingerprint guideline. It is extracted as an image of and detects horizontal and vertical ridges from the fingerprint image, and determines whether the horizontal and vertical ridges form a pattern by repeating a certain portion of each, and if it is determined that the pattern is not formed, text or voice A ridge check step of notifying that it is not a fingerprint image of the finger and returning to the receiving step of a finger fingerprint image;
Characterized in that it comprises a, smart phone-based identification method. The method of claim 5, wherein the step of detecting fingerprint information of a police smartphone,
After the ridge check step, the operation processing unit of the police smartphone includes a fingerprint feature detection step of detecting a fingerprint feature point including an end point and a branch point of the ridge of the fingerprint from the fingerprint image of the subject;
Characterized in that it further comprises a, smart phone-based identification method. The method of claim 6, wherein the step of detecting a fingerprint feature,
The calculation processing unit of the police smartphone includes a pre-processing step of thinning the fingerprint image of the subject;
The operation processing unit of the police smartphone comprises: a candidate fingerprint feature point detection step of detecting a candidate fingerprint feature point from the fingerprint image thinned in the pre-processing step;
The operation processing unit of the police smartphone includes a post-processing step of removing pseudo feature points from the feature points detected in the candidate fingerprint feature point detection step;
Characterized in that made, including, a smart phone-based identification method. The method of claim 7, wherein the step of analyzing the identity of the matching server,
The matching server selects the alignment reference point of the image by calculating the rotation and transition amount at which the characteristic points of the two fingerprint images overlap the maximum number of the fingerprint characteristic points of the subject and the fingerprint characteristic points to be compared previously stored in the database, and the alignment reference point An alignment step of determining a corresponding pair of feature points after transforming the coordinates of the feature points so as to be aligned with each other;
The matching server calculates a similarity using coordinates, types, and angle information of the corresponding feature point pair determined in the alignment step, a matching step;
The matching server includes a score acquisition step of outputting a match degree between the subject's fingerprint feature point and the fingerprint feature point to be compared previously stored in the database as a score, according to preset rule data, from the similarity of the corresponding feature point pair;
Characterized in that made, including, a smart phone-based identification method. The method of claim 8, wherein the step of analyzing the identity of the matching server comprises:
Before the alignment step, the matching server obtains the ridge interval for the fingerprint image of the subject received from the operation processing unit of the police smartphone and normalizes the ridge intervals by using the maximum and minimum values of the ridge interval to standardize the ridge interval. Standardization stage;
After the ridge interval standardization step, the matching server detects the ridge direction information from the subject's fingerprint image, and determines that the ridge direction information is incorrect when there are no two or more consecutive ridge directions in any one of the front, back, left and right. And correcting the wrong ridge direction information in the direction of any one of successive ridges before, after, left, and right of the ridge having the erroneous ridge direction information, thereby correcting the fingerprint feature point of the subject;
Characterized in that it further comprises a, smart phone-based identification method. A recording medium storing a source of a smartphone program for the method of claim 1 or claim 2 for identification.

Images