WO2023124745A1

WO2023124745A1 - Fingerprint comparison method and apparatus, storage medium, and device

Info

Publication number: WO2023124745A1
Application number: PCT/CN2022/135830
Authority: WO
Inventors: 孔勇; 周军
Original assignee: 北京眼神智能科技有限公司; 北京眼神科技有限公司; 深圳爱酷智能科技有限公司
Priority date: 2021-12-27
Filing date: 2022-12-01
Publication date: 2023-07-06

Abstract

The present application discloses a fingerprint comparison method and apparatus, a storage medium, and a device, relating to the field of fingerprint recognition. The method comprises: acquiring a first group of minutiae of a first fingerprint image and a second group of minutiae of a second fingerprint image; acquiring matching minutiae pairs of the first group of minutiae and the second group of minutiae; for each matching minutiae pair, calculating a similarity median of the each matching minutiae pair according to a distance and an angle difference between the two minutiae of the each matching minutiae pair and neighborhood minutiae information; and calculating a matching score of the first fingerprint image and the second fingerprint image according to the similarity median of the each matching minutiae pair.

Description

Fingerprint comparison method, device, storage medium and equipment

related application

This application requires that the Chinese patent application with the application number 202111612355.2 and the name "fingerprint comparison method, device, storage medium and equipment" be submitted to the China Patent Office on December 27, 2021, and the Chinese patent application filed on January 22, 2022. The patent office's application number is 202210075001.7, and the title is "fingerprint comparison method, device, storage medium and equipment". The entire content of the aforementioned two patents is incorporated in this application by reference.

technical field

The present application relates to the field of fingerprint identification, in particular to a fingerprint comparison method, device, storage medium and equipment.

Background technique

Fingerprint identification technology is one of the most mature and common technologies in the field of biometric identification. Fingerprints are made up of alternating ridges and valleys on the surface of the finger. Because of the convenient fingerprint collection process, low hardware cost, reliable performance and convenient use, it has been widely used in many aspects such as identity verification and access control, such as mobile phone unlocking, access control equipment, fingerprint collection when applying for ID cards, and entry and exit. Fast pass system, etc.

Contents of the invention

In a first aspect, the present application provides a fingerprint comparison method, the method comprising:

Obtain the first set of minutiae points of the first fingerprint image and the second set of minutiae points of the second fingerprint image;

Obtain the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points;

For each matching minutiae point pair, calculate the median similarity value of the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair and the neighborhood minutiae point information; and

Calculate the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each matching pair of minutiae points.

In some embodiments, for each matching minutiae point pair, the distance and angle difference between two minutiae points of the matching minutiae point pair and the neighborhood minutiae point information are used to calculate the matching minutiae point pair Similarity median value; and according to the similarity median value of each minutiae matching pair, calculate the matching score of the first fingerprint image and the second fingerprint image, including:

Selecting a matching minutiae point pair as a reference minutiae point pair, and performing a rigid transformation on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide;

For each matching minutiae point pair, calculate the contribution value of the matching minutiae point pair according to the matching minutiae point pair and the corresponding neighborhood minutiae point pair, wherein the contribution value is the median value of the similarity;

Wherein, the neighborhood minutiae point pair sets all matching minutiae point pairs within the neighborhood range for the matching minutiae point pair;

The matching score between the first fingerprint image and the second fingerprint image is calculated according to the contribution values of all matching minutiae point pairs.

Further, for each matching detail point pair, calculating the contribution value of the matching detail point pair according to the matching detail point pair and the corresponding neighborhood detail point pair includes:

calculating the first contribution value of the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair;

performing a rigid transformation on the second group of minutiae points of each neighborhood minutiae point pair within the set neighborhood range of the matching minutiae point pair so that the two minutiae points of the matching minutiae point pair coincide;

For each neighborhood minutiae point pair, calculate the second contribution value of each neighborhood minutiae point pair according to the distance and angle difference between the two minutiae points of the neighborhood minutiae point pair, and the corresponding interior angle value;

Wherein, the inner angle value is an angle value of an inner angle formed by two minutiae points of the neighborhood minutiae point pair and the matching minutiae point pair;

The contribution value of the matching minutiae point pair is calculated according to the first contribution value of the matching minutiae point pair and the second contribution values of all neighborhood minutiae point pairs.

Further, the first contribution value of the matching minutiae point pair is calculated by the following formula;

sd＝Z(d _i ,μ ₁ ,τ ₁ )*Z(Δθ _i ,μ ₂ ,τ ₂ )

Wherein, sd is the first contribution value of the pair of matching minutiae points, Z(,,) is a sigmoid function, d _i and Δθ _i are respectively the distance and angle difference between two minutiae points of i matching minutiae point pairs, i =1,2,...,N, N is the number of matching minutiae point pairs, μ ₁ , τ ₁ , μ ₂ , τ ₂ are the set parameters;

The second contribution value of each neighborhood detail point pair is calculated by the following formula;

Among them, sn _j is the second contribution value of the jth neighborhood detail point pair, j=1, 2,..., M, M is the neighborhood detail point pair within the neighborhood range of the matching detail point pair number, sn _j = Z(d _j ,μ ₁ ,τ ₁ )*Z(Δθ _j ,μ ₂ ,τ ₂ )*Z(α _j ,μ ₃ ,τ ₃ ), d _j and Δθ _j are respectively The distance and angle difference between the two minutiae points of the j neighborhood minutiae point pair, α _j is the angle value of the interior angle formed by the two minutiae points of the jth neighborhood minutiae point pair and the matching minutiae point pair, μ ₃ , τ ₃ is the set parameter;

Calculate the contribution value of the matching detail point pair by the following formula;

Among them, pairscore _i is the contribution value of the i-th matching minutiae point pair, np and nq are respectively the number of the first group of minutiae points and the number of the second group of minutiae points within the set neighborhood of the matching minutiae point ;

.

Further, the matching score between the first fingerprint image and the second fingerprint image is calculated by the following formula:

Wherein, s is the matching score between the first fingerprint image and the second fingerprint image, N1 and N2 are the numbers of the first group of minutiae points and the number of the second group of minutiae points respectively.

Further, the selection of a matching minutiae point pair as a reference minutiae point pair includes:

Calculate the center of gravity of the first set of minutiae points of all matching minutiae point pairs;

Find a first group of minutiae points closest to the center of gravity from the first group of minutiae points of all matching minutiae point pairs, and use the matching minutiae point pair to which the found first group of minutiae points belong as the reference minutiae point pair .

Further, performing a rigid transformation on all the minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincides, including:

Calculate the angle difference Δθ and the coordinate difference Δx, Δy of the two minutiae points of the reference minutiae point pair;

Δθ=θ2-θ1

Among them, (x2, y2, θ2) are the coordinates and direction angles of the first group of minutiae points of the reference minutiae point pair, and (x1, y1, θ1) are the coordinates of the second group of minutiae points of the reference minutiae point pair and orientation angle;

Calculate the coordinates and orientation angles of the second group of minutiae points after the rigid transformation according to the angle difference Δθ and the coordinate difference Δx, Δy of the two details of the reference minutiae point pair;

θ'=θ-Δθ

Among them, (x, y, θ) are the coordinates and orientation angles of the second group of minutiae points before rigid transformation, and (x′, y′, θ′) are the coordinates and orientation angles of the second group of minutiae points after rigid transformation.

Further, the method also includes:

A fingerprint comparison result is obtained according to the matching score between the first fingerprint image and the second fingerprint image.

In some embodiments, the obtaining the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points includes:

calculating a 2-nearest neighbor structure representation of each minutiae in the first set of minutiae points and a 2-nearest neighbor structure representation of each minutiae point in the second set of minutiae points;

Calculate the first group of minutiae points and the second group of minutiae points according to the 2-nearest neighbor structure representation of each minutiae point in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points The local similarity matrix of ;

Select several elements from the local similarity matrix in descending order, and obtain two minutiae points corresponding to each selected element as a matching pair.

For each matching pair, calculate the matching certainty similarity matrix of the matching pair according to the local similarity matrix and the distance difference and angle difference between two minutiae points of the matching pair, wherein the matching determination The sexual similarity matrix is the median value of the similarity; and

Select the matching detail point pairs of each matching deterministic similarity matrix and calculate the matching score of each matching deterministic similarity matrix; and calculate the total matching score according to all matching scores.

Further, the matching certainty similarity matrix of each matching pair is calculated by the following formula:

Wherein, the matching pair includes the a-th minutiae point in the first group of minutiae points and the b-th minutiae point in the second group of minutiae points, ml(i,j) is the i-th minutiae of the matching certainty similarity matrix The element in the jth column of the row, i=1, 2,..., M, j=1, 2,..., N, sl(i, j) is the element in the jth row of the local similarity matrix, d , α, β are the absolute values of the three elements of the three-dimensional vector v0,

is the eigenvector obtained by performing the polar coordinate operation on the i-th minutiae point in the first group of minutiae points with reference to the a-th minutiae point in the first group of minutiae points,

is a feature vector obtained by performing polar coordinate operation on the jth minutiae point in the second group of minutiae points with reference to the bth minutiae point in the second group of minutiae points.

Further, the 2-nearest neighbor structure representation of each minutiae point in the first group of minutiae points is calculated by the following formula:

Among them, Fl _i is the 2-nearest neighbor structure representation of the i-th minutiae F _i in the first group of minutiae points, p and q are the two minutiae points F _p and F _q closest to the i-th minutiae F _i respectively serial number,

t _l , (x _l ,y _l ) and

Respectively, the type, coordinates and angle of the lth detail point F _i , l takes i, p, q;

n _ip is the number of ridge lines on the line between the i-th detail point and the p-th detail point in the first group of detail points, and n _iq is the i-th detail point and the q-th detail point in the first group of detail points The number of ridges on the connection;

dφ(t ₁ ,t ₂ ) is the angle difference function, which is defined as:

Calculate the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points by the following formula:

Among them, Gl _j is the 2-nearest neighbor structure representation of the jth minutiae G _j in the second group of minutiae points, m and n are the two minutiae points G _m and G _n closest to the jth minutiae G _j respectively serial number,

t _h , (x _h ,y _h ) and

Respectively, the type, coordinates and angle of the hth detail point G _h , where h takes j, m, n;

n _jm is the number of ridge lines on the line between the jth minutiae point and the mth minutiae point in the second group of minutiae points, and n _jn is the jth minutiae point and the nth minutiae point in the second group of minutiae points The number of ridges on the link.

Further, the local similarity matrix is calculated by the following formula:

Among them, A is the dot product of the weighting coefficient vector W and the vector V;

w _d ＝1,w _θ ＝0.3*180/π,

w _n =3, w _t =3, V=|V0|, V0=Fl _i -Gl _i .

Further, the eigenvectors are calculated by the following formula

and

Among them, (x _a ,y _a ) and

are respectively the coordinates and angles of the ath minutiae point in the first group of minutiae points, (x _b , y _b ) and

are the coordinates and angles of the bth minutiae point in the second group of minutiae points, respectively.

Further, the selecting matching detail point pairs of each matching deterministic similarity matrix and calculating the matching score of each matching deterministic similarity matrix includes:

For each matching deterministic similarity matrix, if the element ml(i,j) in the i-th row and j-th column is not the maximum value in its i-th row or not the maximum value in its j-th column, then ml( i, j) is set to 0;

The matching score s of each matching deterministic similarity matrix is calculated by the following formula:

Further, the method also includes:

After the total matching score is calculated according to all the matching scores, the fingerprint comparison result is obtained according to the total matching score.

In a second aspect, the present application provides a fingerprint comparison device, the device comprising:

The data preparation module is used to obtain the first group of minutiae points of the first fingerprint image and the second group of minutiae points of the second fingerprint image;

An acquisition module, configured to acquire the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points;

An intermediate value calculation module, for each matching detail point pair, according to the distance and angle difference between two detail points of the matching detail point pair, and the neighborhood detail point information to calculate the similarity of the matching detail point pair degree median; and

A matching score calculation module, configured to calculate a matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each matching pair of minutiae points.

In some embodiments, the intermediate value calculation module includes a rigid transformation module and a contribution value calculation module:

The rigid transformation module is configured to select a matching minutiae point pair as a reference minutiae point pair, and perform a rigid transformation on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide ;

The contribution value calculation module is used to calculate the contribution value of the matching detail point pair according to the matching detail point pair and the corresponding neighborhood detail point pair for each matching detail point pair;

The matching score calculation module is used to calculate the matching score between the first fingerprint image and the second fingerprint image according to the contribution values of all matching minutiae point pairs.

In some embodiments, the acquisition module includes a 2-nearest neighbor structure representation calculation module, a local similarity matrix calculation module, and a matching pair determination module, and the intermediate value calculation module includes a matching certainty similarity matrix calculation module:

The 2-nearest neighbor structure representation calculation module is used to calculate the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points ;

The local similarity matrix calculation module is used to calculate according to the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points The local similarity matrix of the first group of minutiae points and the second group of minutiae points;

The matching pair determination module is configured to select several elements from the local similarity matrix in descending order, and obtain two minutiae points corresponding to each selected element as a matching minutiae point pair;

The matching deterministic similarity matrix calculation module, for each matching pair, calculates the matching determination of the matching pair according to the distance difference and angle difference between the local similarity matrix and the two minutiae points of the matching pair sex similarity matrix;

The matching score calculation module is used to select the matching detail point pairs of each matching deterministic similarity matrix and calculate the matching score of each matching deterministic similarity matrix; and calculate the total matching score according to all matching scores.

In a third aspect, the present application provides a computer-readable storage medium for fingerprint comparison, including a memory for storing processor-executable instructions. The steps of the fingerprint comparison method.

In a fourth aspect, the present application provides a device for fingerprint comparison, including at least one processor and a memory storing computer-executable instructions, and the processor implements the fingerprint comparison described in the first aspect when executing the instructions method steps.

The details of one or more embodiments of the application are set forth in the accompanying drawings and the description below. Other features, objects and advantages of the present application will be apparent from the description, drawings and claims.

Description of drawings

In order to better describe and illustrate the embodiments of the application, reference may be made to one or more drawings, but additional details or examples used to describe the drawings should not be regarded as an invention of the application, the presently described implementation limitations on the scope of any of the examples or preferred modes.

Fig. 1 is an example diagram of terminal points and bifurcation points on the fingerprint;

Figure 2a is an example diagram of the first fingerprint image;

Fig. 2b is an example diagram of the second fingerprint image;

Figure 3a is an example diagram of the first fingerprint image and its first group of minutiae points;

Figure 3b is an example diagram of the second fingerprint image and its second group of minutiae points;

Fig. 4 is an example diagram of the first group of minutiae points and their numbers and the second group of minutiae points and their numbers;

Fig. 5 is an example diagram of matching minutiae point-to-connection lines;

Figure 6a and Figure 6b are schematic diagrams of the situation where two minutiae points match;

Figure 7a and Figure 7b are schematic diagrams of the situation where two minutiae do not match;

FIG. 8 is a flowchart of a fingerprint comparison method according to an embodiment of the present application;

FIG. 9 is a flowchart of a fingerprint comparison method according to another embodiment of the present application;

10 is a schematic diagram of an interior angle α _j formed by two minutiae points of a neighborhood minutiae point pair and a matching minutiae point pair in an embodiment of the present application;

Fig. 11 is a flowchart of a fingerprint comparison method according to another embodiment of the present application;

Fig. 12 is a schematic diagram of an embodiment of the present application when there are two minutiae points to be matched that may match the minutiae points to be matched in the limit box;

13 is a schematic diagram of the 2-nearest neighbor structure representation of the i-th minutiae point in the first group of minutiae points in an embodiment of the present application;

FIG. 14 is a schematic diagram of a fingerprint matching device in an embodiment of the present application.

FIG. 15 is a schematic diagram of a fingerprint matching device in another embodiment of the present application.

Fig. 16 is a schematic diagram of a fingerprint matching device in another embodiment of the present application.

FIG. 17 is a schematic structural diagram of a device for fingerprint comparison in an embodiment of the present application.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the application with reference to the drawings in the embodiments of the application. Apparently, the described embodiments are only some of the embodiments of the application, not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

Fingerprint recognition technology mainly includes two algorithmic processes of fingerprint feature extraction and fingerprint matching. Among the extracted fingerprint features, the most commonly used and basic features are minutiae, which include two types of terminal points and bifurcation points, which correspond to the end of a ridge line and the division of a ridge line into The case of two ridges. As shown in Figure 1, the circles represent the minutiae points of the bifurcation point type, and the boxes represent the minutiae points of the terminal point type. Each detail point can be specified with type t, coordinates (x, y) and angle

Composed of four-dimensional coordinates (x, y,

t) to represent.

At present, fingerprint minutiae is recognized as the most discriminative and reliable local feature of fingerprints, and the local structure of minutiae has good adaptability to nonlinear deformation and noise. The fingerprint matching algorithm based on minutiae is the most important method in the research of fingerprint matching algorithm. Its core problem is to find the matching minutiae pair and calculate a matching score or similarity value according to certain rules.

The minutiae points of the first fingerprint image are called the first group of minutiae points, and the minutiae points of the second fingerprint image are called the second group of minutiae points. During the research process, the applicant found that in some related technologies, the main steps of the minutiae-based fingerprint matching algorithm are: calculating the local descriptor of each minutiae, calculating each The local similarity of two minutiae points, and find those matching minutiae point pairs that can be matched through the set rules, and finally get a matching score based on the number of these matching minutiae point pairs or the matching certainty similarity value between them .

For example, Fig. 2a and Fig. 2b respectively give an example of the first fingerprint image and the second fingerprint image. Figure 2a is the first fingerprint image, Figure 2b is the second fingerprint image, the first fingerprint image and the second fingerprint image are from the same fingerprint. The minutiae points of the two are shown in Figure 3a and Figure 3b respectively, where Figure 3a is the first fingerprint image and its first group of minutiae points, and Figure 3b is the second fingerprint image and its second group of minutiae points. Each detail point is numbered as shown in Figure 4, the left side of Figure 4 is the first group of detail points (50 in total), and the right side is the second group of detail points (38 in total). There are 36 pairs of matching detail points obtained by using the algorithm, and the matching detail point pairs are marked with connecting lines, as shown in Figure 5.

After the matching detail point pair is found, the matching score can be calculated according to the matching detail point pair. Assume that the number of detail points in the first group of detail points is N1, and the number of detail points in the second group of detail points is N2. The number is N (that is, there are N pairs of minutiae points matched), then the formula for calculating the matching score s in the related art generally has the following types:

Some studies have pointed out that if there are 13 "completely" matching detail point pairs, it means that the two fingerprint images participating in the comparison come from the same fingerprint, but the "completely" here is artificially expressed by experts, and it is difficult to pass A concept expressed by a specific measure. The applicant found through experiments that two fingerprints of non-same fingers can also reach 13 or even 29 matching detail point pairs, while two fingerprints of the same finger only have 8 matching detail point pairs due to the small overlapping area and poor quality. Therefore, due to the different judging criteria of matching minutiae pairs, the similarity between two fingerprint images cannot be well evaluated only by the number of matching minutiae pairs or their similarity.

In other related technologies, the main steps of the minutiae-based fingerprint matching algorithm are: calculate the local similarity matrix of the first group of minutiae points and the second group of minutiae points, select a minutiae point pair according to the local similarity, and refer to this Make a rigid change of the detail point pairs, that is, let the two detail points in this detail point pair completely overlap, and then count the matching detail point pairs that can be paired after the rigidity change, and finally based on the number of these matching detail point pairs and the distance between them The match certainty similarity value for a given match yields a match score.

Among them, the Bounding box technology is used when counting those matching detail point pairs that can be matched after the rigidity changes. The bounding box, also known as the bounding box, is used to limit whether two detail points are close and match. When two groups of fingerprint minutiae are rotated and translated according to a matching pair, they are used to find more matching minutiae pairs. For example, when two minutiae points meet the condition of the bounding box, that is, when the distance between them is very close and the direction is also very close, then the two minutiae points can be considered as matching minutiae point pairs (candidates to be selected).

As shown in Figure 6a-7b, a black circle plus a short line (the direction of this short line can represent the angle of this detail point) represents a detail point, which is recorded as the matched detail point, and the surrounding box is the matched detail point. Click to determine the limited box. The white circle plus a short line is a minutiae point to be matched to judge whether it matches the minutiae point to be matched. The two matching situations are shown in Fig. 6a and Fig. 6b respectively. The two cases of mismatch are shown in Fig. 7a and Fig. 7b respectively, in which, the first case is that the minutiae point to be matched is not in the bounding box, and the second case is that the minutiae point to be matched is in the bounding box but is matched with The direction of the detail point is too different.

The minutiae-based fingerprint matching algorithm of this kind of related technology adopts the method of limiting the box when counting those matching minutiae pairs that can be matched after the rigidity changes, but this method cannot obtain better matching minutiae pairs, resulting in fingerprint The reduction of the comparison accuracy reduces the evaluation indicators such as the equal error rate.

Based on the above problems, after further in-depth research by the applicant, a fingerprint comparison method is designed to better evaluate the similarity of two fingerprint images and improve the accuracy of fingerprint comparison.

Example 1:

In some embodiments of the present application, as shown in Figure 8, the fingerprint comparison method includes:

S1: Obtain the first set of minutiae points of the first fingerprint image and the second set of minutiae points of the second fingerprint image.

The first group of minutiae points and the second group of minutiae points are obtained by extracting the first fingerprint image and the second fingerprint image participating in the comparison through a fingerprint feature extraction algorithm. This application does not limit the specific implementation of the fingerprint feature extraction algorithm.

S2: Obtain matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points.

The method for obtaining the pair of matching detail points will be described in detail in the following embodiments.

S3: For each matching minutiae point pair, calculate the median similarity value of the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair and the neighborhood minutiae point information.

The method for calculating the intermediate value of the similarity will be described in detail in conjunction with the following embodiments.

S4: Calculate the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each matching pair of minutiae points.

The method for calculating the matching score will be described in detail in conjunction with the following embodiments.

Example 2:

This embodiment provides a fingerprint comparison method, as shown in Figure 9, the method includes:

The first group of minutiae points and the second group of minutiae points in this embodiment are obtained by extracting the first fingerprint image and the second fingerprint image participating in the comparison through the fingerprint feature extraction algorithm, and this application does not limit the specific implementation of the fingerprint feature extraction algorithm Way. Moreover, the present application does not limit the manner of obtaining the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points, and various methods in the prior art may be used.

For example, the number of minutiae points of the first group of minutiae points is N1, the number of minutiae points of the second group of minutiae points is N2, and the number of matching minutiae points of the first group of minutiae points and the second group of minutiae points is N( N<=N1, and N<=N2), the N matching detail point pairs are represented by (p _i , q _i ), i=1, 2,..., N, where p _i , i=1, 2, ..., N are N minutiae points belonging to the first group of minutiae points, q _i , i=1, 2, ..., N are N minutiae points belonging to the second group of minutiae points.

In this embodiment, the aforementioned step S3: For each matching minutiae point pair, calculate the matching minutiae according to the distance difference and angle difference between two minutiae points of the matching minutiae point pair, and the neighborhood minutiae point information The intermediate value of similarity between point pairs specifically includes the following steps S31a and S32a.

S31a: Select a matching minutiae point pair as a reference minutiae point pair, and perform a rigid transformation on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide.

The selected reference minutiae point pair can be recorded as (p _a , q _a ), where _pa is the minutiae point belonging to the first group of minutiae points, and q _a is the minutiae point belonging to the second group of minutiae points. Then q _a is subjected to rigid transformation, so that q _a coincides with p _a , and other minutiae points in the second group of minutiae points are also subjected to the same rigid transformation. Convert the second group of minutiae points to the coordinates of the coordinate system of the first group of minutiae points to facilitate subsequent operations.

S32a: For each matching minutiae point pair, calculate the contribution value of the matching minutiae point pair according to the distance and angle difference between two minutiae points in the matching minutiae point pair and the corresponding neighborhood minutiae point pair.

The contribution value is the median similarity value described in Example 1.

Wherein, the neighborhood minutiae point pair sets all the matching minutiae point pairs within the neighborhood range for the matching minutiae point pair.

This application improves the method of calculating the contribution value of the matching minutiae point pair. When calculating the contribution value of the matching minutiae point pair, it is not only calculated according to the two minutiae points of the matching minutiae point pair, but also according to the neighbors in the neighborhood of the matching minutiae point pair. The domain minutiae point pair calculation makes full use of the useful information contained in the matching minutiae point pair, which can improve the accuracy of fingerprint comparison.

In this embodiment, the foregoing step S4: calculate the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each minutiae matching pair, specifically the following step S41a .

S41a: Calculate the matching score between the first fingerprint image and the second fingerprint image according to the contribution values of all matching minutiae point pairs.

The matching score is the statistical value of the contribution value of all minutiae point pairs. After the matching score is calculated, the fingerprint comparison result can be obtained according to the matching score.

This embodiment first obtains the first group of minutiae points of the first fingerprint image and the second group of minutiae points of the second fingerprint image, and obtains the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points; then select A matching minutiae point pair is used as a reference minutiae point pair, and a rigid transformation is performed on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide; for each matching minutiae point pair, according to the matching The two minutiae points of the minutiae point pair and the minutiae points of the minutiae point pair within the set neighborhood range of the matching minutiae point pair calculate the contribution value of the matching minutiae point pair; finally, the matching score is calculated according to the contribution values of all minutiae point pairs .

In this embodiment, when calculating the contribution value of a matching minutiae point pair, it is not only calculated based on the two minutiae points of the matching minutiae point pair, but also calculated based on the neighborhood minutiae point pair within the neighborhood of the matching minutiae point pair, making full use of the matching minutiae point pair The useful information contained within a certain range of itself and its surroundings can well evaluate the similarity of two fingerprint images and improve the accuracy of fingerprint comparison.

As an improvement of this embodiment, the aforementioned step S32a includes:

S320: Calculate the first contribution value of the matching minutiae point pair according to the distance and angle difference between the two minutiae points of the matching minutiae point pair

In one of the examples, the first contribution value of the matching minutiae point pair can be calculated by the following formula;

sd=Z(di,μ ₁ ,τ ₁ )*Z(Δθ _i ,μ ₂ ,τ ₂ )

Among them, sd is the first contribution value of the pair of matching minutiae points, Z(,,) is a sigmoid function, and d _i and Δθ _i are two minutiae points of i matching minutiae point pairs (p _i , q _i ) , i=1,2,...,N, N is the number of matching minutiae point pairs, μ ₁ , τ ₁ , μ ₂ , τ ₂ are set parameters.

The formula of the sigmoid function is as follows:

Where v is a variable, μ and τ are two parameters, and the value range of the sigmoid function is [0,1]. The distance d _i and the angle difference Δθ _i are respectively passed into the sigmoid function as variables, and a value between 0 and 1 can be calculated according to the set μ, τ parameters to represent a certain similarity, such as the closer the distance The larger the value, the smaller the angle difference, the larger the value, and so on.

S321: For the matching minutiae point pair (p _i , q _i ), set the second group of minutiae points of each neighborhood minutiae point pair within the neighborhood range so that the two minutiae points of the matching minutiae point pair Coincident rigid transformation.

Assume that the number of the first group of minutiae points in the neighborhood of pi with a radius of 90 is np, the number of the second _group of minutiae points in the neighborhood of q _i with a radius of 90 is nq, and p _i and q _i are within a radius of There are M matching pairs of neighborhood minutiae points in the neighborhood range of 90, M<=np and M<=nq, because not every minutiae point in the neighborhood range has a matching pair.

Denote the M neighborhood detail point pairs within the neighborhood range as (pNeighbor _j ,qNeighbor _j ), j=1,2,...,M, use a method similar to that in S31a to make rigid changes so that p _i and q _i coincide, and at the same time , the qNeighbor _j of the M neighborhood detail point pairs in the neighborhood also undergoes the same rigid transformation.

S322: For each neighborhood minutiae point pair, calculate the second contribution value of each neighborhood minutiae point pair according to the distance and angle difference between the two minutiae points of the neighborhood minutiae point pair and the corresponding inner angle value.

Wherein, the internal angle value is an angle value of an internal angle formed by two minutiae points of the neighborhood minutiae point pair and the matching minutiae point pair.

In one example, the second contribution value of each neighborhood detail point pair can be calculated by the following formula;

Among them, sn _j is the second contribution value of the jth neighborhood detail point pair (pNeighbor _j , qNeighbor _j ), j=1, 2,..., M, M is the set neighborhood range of the matching detail point pair The number of detail point pairs in the neighborhood of sn _j =Z(d _j ,μ ₁ ,τ ₁ )*Z(Δθ _j ,μ ₂ ,τ ₂ )*Z(α _j ,μ ₃ ,τ ₃ ), d _j and Δθ _j are the distance and angle difference between the two detail points of the jth neighborhood detail point pair (pNeighbor _j , qNeighbor _j ), and α _j is the jth neighborhood detail point pair (pNeighbor _j , qNeighbor _j ) The angle value of the internal angle formed by the two minutiae points and the pair of matching minutiae points (p _i , q _i ), α _j is shown in Figure 10, and μ ₃ , τ ₃ are set parameters.

S323: Calculate the contribution value of the matching minutiae point pair according to the first contribution value of the matching minutiae point pair and the second contribution values of all neighborhood minutiae point pairs.

In one example, the contribution value of the matching detail point pair can be calculated by the following formula;

.

In this embodiment, when calculating the contribution value of a matching minutiae point pair, not only the distance and angle difference between the two minutiae points of the matching minutiae point pair, but also the neighborhood minutiae point pairs within a certain neighborhood of the matching minutiae point pair are considered. The distance between two minutiae points, the angle difference, and the value of the internal angle formed with the pair of matching minutiae points. It can make full use of information such as the distance and angle of the matching minutiae point pair itself and the matching minutiae point pair within a certain range around it, thereby improving the accuracy of fingerprint comparison.

Then, the matching score between the first fingerprint image and the second fingerprint image is calculated by the following formula:

Among them, s is the matching score between the first fingerprint image and the second fingerprint image, N1 and N2 are the number of minutiae points in the first group and the number of minutiae points in the second group respectively.

In this embodiment, multiple methods can be used to select the reference detail point pair. In one example, the method for selecting the reference detail point pair includes:

S310: Calculate the center of gravity of the first group of minutiae points of all matching minutiae point pairs.

The matching minutiae points are (p _i ,q _i ), i=1,2,...,N, where p _i ,i=1,2,...,N are N minutiae points belonging to the first group of minutiae points, and It is the first group of minutiae points of all matching minutiae point pairs, and the coordinates (xc, yc) of its center of gravity C are:

S311: Find a first group of minutiae points closest to the center of gravity from the first group of minutiae points of all matching minutiae point pairs, and use the matching minutiae point pair to which the found first group of minutiae points belong as a reference minutiae point pair.

Among the above N minutiae points p _i , assuming that the minutiae point closest to the center of gravity C (such as Euclidean distance, etc.) is p _a , and its corresponding pair of matching minutiae points is (p _a , q _a ), the (p _a , q _a ) as the reference minutiae point pair.

After obtaining the reference detail point pair, rigid transformation can be performed. This application does not limit the specific implementation of rigid transformation. An example includes:

S312: Calculate the angle difference Δθ and the coordinate difference Δx, Δy of the two minutiae points of the reference minutiae point pair;

Δθ=θ2-θ1

Among them, (x2, y2, θ2) are the coordinates and orientation angles of the first group of minutiae points of the reference minutiae point pair, and (x1, y1, θ1) are the coordinates and orientation angles of the second group of minutiae points of the reference minutiae point pair.

Assuming that the coordinates and orientation angle of q _a in the reference detail point pair (p _a , q _a ) are (x2, y2, θ2), and the coordinates and orientation angle of p _a are (x1, y1, θ1), calculate q _a and p _a Angle difference Δθ and coordinate difference Δx, Δy.

S313: Calculate the coordinates and orientation angles of the second group of minutiae after rigid transformation according to the angle difference Δθ and the coordinate difference Δx, Δy between the two minutiae of the reference minutiae point pair.

θ'=θ-Δθ

The rigid transformation that makes q _a and p _a coincident includes a rotation and translation operation T, and the angles of the second set of minutiae points are all subtracted by Δθ.

In order to illustrate the effect of the present application more clearly, the following test is carried out: a fingerprint test set is constructed, and the fingerprint matching algorithm based on the minutiae of the prior art is completely tested, and its equal error rate is 5.5%, while the fingerprint of the present application The equal error rate of the comparison method is 3.5%. Compared with the prior art, the equal error rate is reduced by about 36.3%, and the result has been greatly improved.

To sum up, this application optimizes the method of calculating the fingerprint comparison score. When calculating the contribution value of the matching minutiae point pair, it not only considers the distance and angle difference between the two minutiae points of the matching minutiae point itself, Also consider information such as the distance between the two minutiae points of the neighborhood minutiae point pair within a certain neighborhood of the matching minutiae point pair, the angle difference, and the inner angle value formed with the matching minutiae point pair. The information that is helpful for fingerprint comparison, such as the distance and angle contained in the matching minutiae point pair itself and the matching minutiae point pair within a certain range around it, can be fully utilized, and the accuracy of fingerprint comparison is improved. Compared with the existing technology, the test results on the test set have been greatly improved, the accuracy of fingerprint comparison has been improved, and the evaluation indicators such as the equal error rate have been reduced.

Example 3:

This embodiment provides a fingerprint comparison method, as shown in Figure 11, the method includes:

The first group of minutiae points and the second group of minutiae points in this step are obtained by extracting the first fingerprint image and the second fingerprint image participating in the comparison through the fingerprint feature extraction algorithm. This application does not limit the specific implementation of the fingerprint feature extraction algorithm. .

An example of the first group of minutiae points and the second group of feature points is: the first group of minutiae points corresponding to the first fingerprint image includes M minutiae points, denoted as

i=1,2,...,M, t _i , (xi _, y _i ) and

are respectively the type, coordinates and angle of the i-th minutiae F _i in the first group of minutiae points; the second group of minutiae points corresponding to the second fingerprint image includes N minutiae points, denoted as

j=1,2,...,N, t _j , (x _j ,y _j ) and

are respectively the type, coordinates and angle of the jth minutiae point G _j in the second group of minutiae points.

The comparison between the first fingerprint image and the second fingerprint image can be called the matching between the first set of minutiae points and the second set of minutiae points.

In this embodiment, the foregoing step S2: obtaining matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points specifically includes the following steps S21b-S23b.

S21b: Calculate the 2-nearest neighbor structure representation of each minutiae point in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points.

For the i-th minutiae point in the first group of minutiae points

Its 2-nearest neighbor structure representation is related to the two closest minutiae points F _p and F _q , where:

21bt _p , (x _p ,y _p ) and

are respectively the type, coordinates and angle of the pth minutiae point F _p in the first group of minutiae points.

t _q , (x _q ,y _q ) and

are respectively the type, coordinates and angle of the qth minutiae point F _q in the first group of minutiae points.

The present application does not limit the specific calculation method of the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points. In one example, it can be calculated by the following formula:

Among them, Fl _i is the 2-neighbor structure representation of the i-th minutiae F _i in the first group of minutiae points, Fl _i is an 11-dimensional vector, and p and q are the two nearest neighbors to the i-th minutiae F _i , respectively. The numbers of minutiae points F _p and F _q ,

t _l , (x _l ,y _l ) and

Respectively, the type, coordinates and angle of the lth minutiae point F _i in the first group of minutiae points, l takes i, p, q;

Indicates the distance between the i-th minutiae point and the p-th minutiae point;

Indicates the distance between the i-th minutiae point and the q-th minutiae point;

Indicates the difference between the angle between the line connecting the i-th detail point and the p-th detail point and the x-axis and the angle of the i-th detail point;

Indicates the difference between the angle between the line connecting the i-th detail point and the q-th detail point and the x-axis and the angle of the i-th detail point;

Indicates the difference between the angle of the i-th minutiae point and the angle of the p-th minutiae point;

Indicates the difference between the angle of the i-th detail point and the angle of the q-th detail point;

The aforementioned dφ(t ₁ ,t ₂ ) is an angle difference function, which is defined as:

n _ip is the number of ridge lines on the line between the i-th detail point and the p-th detail point in the first group of detail points, and n _iq is the i-th detail point and the q-th detail point in the first group of detail points The number of ridges on the link.

An example of the 2-neighbor structure representation of the ith minutiae F _i in the first group of minutiae points is shown in Fig. 13 .

Correspondingly, for the jth minutiae point in the second group of minutiae points

Its 2-nearest neighbor structure indicates that it is related to the two nearest minutiae points _Gm and _Gn , and the specific calculation formula is as follows:

t _h , (x _h ,y _h ) and

Respectively, the type, coordinates and angle of the hth minutiae point G _h in the second group of minutiae points, where h takes j, m, n;

S22b: According to the 2-nearest neighbor structure representation of each minutiae point in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points, calculate the local parts of the first group of minutiae points and the second group of minutiae points similarity matrix.

The local similarity matrix is a matrix of size M*N, and the element sl(i, j) in the i-th row and j-column is the i-th minutiae point in the first group of minutiae points and the j-th minutiae point in the second group of minutiae points The local similarity of minutiae points, the local similarity represents Fl _i through the 2-nearest neighbor structure of the i-th minutiae point in the first group of minutiae points and the 2-nearest neighbor of the j-th minutiae point in the second group of minutiae points The structure represents Gl _i calculated.

Exemplarily, a specific calculation method of the local similarity matrix is:

Calculate the difference between two 2-nearest neighbor structures representing Fl _i and Gl _j to obtain a vector V0=Fl _i -Gl _j , and take the absolute value of each element of the vector V0 to obtain a new vector V=|V0|.

Another 11-dimensional vector

W is a weighting coefficient vector, and its specific values are: w _d =1, w _θ =0.3*180/π,

w _n =3, w _t =3.

Calculate the dot product of the weighting coefficient vector W and the vector V to obtain the parameter A, namely:

Calculate the local similarity sl(i,j) between the i-th minutiae in the first group of minutiae and the j-th minutiae in the second group of minutiae:

i=1,2,...,M; j=1,2,...,N.

S23b: Select several elements from the local similarity matrix in descending order, and obtain two minutiae points corresponding to each selected element as a matching pair.

In this step, only one element may be selected, and at this time, the element is the maximum value of all elements in the local similarity matrix, which corresponds to a matching pair. Assume that the matching pair includes the a-th minutiae point in the first group of minutiae points and the b-th minutiae point in the second group of minutiae points. The matching pair is denoted as (a,b), which satisfies sl(a,b)=max _i,j (sl(i,j)), which means the ath minutiae point and the second minutiae point in the first group of minutiae points The local similarity value between the bth minutiae in the group minutiae is the largest.

The significance of the matching pair (a, b) is that the relative transformation of translation and rotation between the first two fingerprint images is unknown in advance, and operations such as translation and rotation need to be performed according to the matching pair (a, b) to find the two fingerprint images. Matching pairs of other minutiae between group fingerprint minutiae.

Of course, this step can also select multiple elements from large to small, for example, 5, and each element corresponds to a matching pair. Subsequent processing is performed separately for each matching pair.

In this embodiment, the foregoing step S3: For each matching minutiae point pair, calculate the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair and the neighborhood minutiae point information The intermediate value of the similarity, specifically includes the following step S31b.

S31b: For each pair, calculate a matching certainty similarity matrix of the matching pair according to the local similarity matrix and the distance difference and angle difference between two minutiae points of the matching pair.

Wherein, the matching deterministic similarity matrix is the median similarity value described in Embodiment 1.

The size of the matching deterministic similarity matrix is M*N, and the element ml(i, j) in the i-th row and j-column represents the i-th minutiae point in the first group of minutiae points and the i-th minutiae point in the second group of minutiae points The similarity of j minutiae points.

This step is illustrated by taking only one matching pair (a, b) as an example. When multiple matching pairs are obtained, the matching certainty similarity matrix is calculated for each matching pair. Exemplarily, a specific calculation method of the matching deterministic similarity matrix corresponding to the matching pair (a, b) is:

Firstly, each minutiae F _i in the first group of minutiae points takes the ath minutiae as a reference to carry out the polar coordinate conversion operation, and converts the four-dimensional vector of the i-th minutiae point

Convert to 3D eigenvectors

Among them, (x _a ,y _a ) and

are the coordinates and angles of the ath minutiae point in the first group of minutiae points, respectively.

Similarly, each minutiae point G _j in the second group of minutiae points takes the bth minutiae point as a reference to carry out polar coordinate transformation operation, and converts the four-dimensional vector of the jth minutiae point

Convert to 3D eigenvectors

Among them, (x _b ,y _b ) and

Calculate eigenvectors

and

A three-dimensional vector v0=Fl _i -Gl _j is obtained, and the absolute value of each element of the vector v0 is obtained to obtain a new three-dimensional vector v=|v0|. The three elements of the three-dimensional vector v are d, α, β.

The matching certainty similarity matrix corresponding to the matching pair (a, b) is calculated by the following formula:

i=1, 2, ..., M, j = 1, 2, ..., N, the threshold {8, π/6, π/6} is the set limit box.

In the minutiae-based fingerprint matching algorithm of the related technology, when the matching minutiae pairs that can be matched after the rigidity changes are counted, in order to reduce the influence of the nonlinear deformation of the fingerprint image, it is generally considered that two minutiae points in the limited box are one matching pair. That is, when the related technology calculates ml(i,j), a fixed item 0.5 is directly added after 0.5*sl(i,j). This method ignores the distance difference and angle difference between the two detail points. The impact of matching ratings. And when there are two or more minutiae points to be matched that may match the minutiae to be matched in the bounding box, as shown in FIG. A better minutiae point to be matched cannot be selected, so that a better matching minutiae point pair cannot be obtained in the subsequent steps.

Different from related technologies, the method in this embodiment does not add a fixed item after 0.5*sl(i,j), but adds a dynamic value, namely (8-d)*cos(α) ² *cos (β) ² /16, the dynamic value changes according to the vertical dynamics of d, α, β, d represents the distance difference between two detail points, α, β represents the angle difference between two detail points, so that ml(i,j ) integrates the distance difference and angle difference information of two minutiae points, which can better measure the similarity between two minutiae points, ultimately improve the accuracy of fingerprint comparison, and reduce the evaluation indicators such as the equal error rate. And when there are two or more minutiae points to be matched that may match the minutiae points to be matched in the limited box, it is helpful to select a more optimal pair of matching minutiae points.

In this embodiment, the foregoing step S4: calculate the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each pair of minutiae points, specifically the following step S41b.

S41b: Select the matching detail point pairs of each matching deterministic similarity matrix and calculate the matching score of each matching deterministic similarity matrix; and calculate the total matching score according to all matching scores.

Still taking only one matching pair (a, b) as an example, the matching pair (a, b) corresponds to a matching deterministic similarity matrix, for example, the matching deterministic similarity of the matching pair (a, b) A specific calculation method of matching detail point pairs and matching scores of the degree matrix is:

First, for the matching deterministic similarity matrix, if the element ml(i,j) in the i-th row and j-th column is not the maximum value in the i-th row or not the maximum value in the j-th column, then the ml( i, j) are set to 0.

After such processing, those non-zero elements in the matching deterministic similarity matrix correspond to corresponding matching detail point pairs.

Then, the matching score s of the matching certainty similarity matrix is calculated by the following formula:

The higher the matching score s, the more likely the first fingerprint image and the second fingerprint image participating in the comparison are two images of the same fingerprint.

Similarly, when multiple matching pairs are obtained, each matching pair is calculated to obtain a matching score.

When only one matching pair is fetched, only one matching score is calculated, and the total matching score is this one itself.

When multiple matching pairs are obtained, for example, 5 matching pairs, a matching score is calculated for each matching pair, and the total matching score is calculated according to the statistics of these matching scores, which can improve the fingerprint comparison results and achieve lower etc. Error rate. For example, the highest value among these matching scores may be taken as the total matching score, or the average value of the two highest values among some matching scores may be taken as the total matching score.

After the total matching score is calculated, the fingerprint comparison result can be obtained according to the total matching score.

In order to clearly illustrate the effect of this embodiment, the following test has been carried out: construct a fingerprint test set that is all 500DPI, if it is tested according to the fingerprint matching algorithm based on the minutiae of the prior art, its equal error rate is 5.5%, while this The equal error rate of the applied fingerprint comparison method is 3.7%, which is about 32.7% lower than the prior art, and the result has been greatly improved.

To sum up, the embodiment of the present application improves the calculation method of the matching deterministic similarity matrix, and combines the distance difference and angle difference information and the information related to the bounding box to calculate the matching deterministic similarity value, which can better measure the two The similarity between minutiae points and helps to select a better matching minutiae point pair. Compared with related technologies, the test results on the test set have been greatly improved, and the evaluation indicators such as the equal error rate have been reduced.

It should be understood that although the steps in the flow charts involved in the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flow charts involved in the above-mentioned embodiments may include multiple steps or stages, and these steps or stages are not necessarily executed at the same time, but may be performed at different times For execution, the execution order of these steps or stages is not necessarily performed sequentially, but may be executed in turn or alternately with other steps or at least a part of steps or stages in other steps.

Example 4:

An embodiment of the present application provides a fingerprint comparison device, as shown in Figure 14, the device includes:

The data preparation module 1 is used to obtain the first group of minutiae points of the first fingerprint image and the second group of minutiae points of the second fingerprint image.

The acquiring module 2 is configured to acquire matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points.

Intermediate value calculation module 3, for each matching detail point pair, according to the distance and angle difference between two detail points of the matching detail point pair, and the neighborhood detail point information to calculate the value of the matching detail point pair The median value of the similarity.

The matching score calculation module 4 is configured to calculate the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each matching pair of minutiae points.

Example 5:

Another embodiment of the present application provides a fingerprint comparison device, as shown in Figure 15, the device includes:

The data preparation module 1a is used to obtain the first group of minutiae points of the first fingerprint image and the second group of minutiae points of the second fingerprint image.

The data acquisition module 2a is configured to acquire matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points.

In this embodiment, the intermediate value acquisition module 3 may include a rigid transformation module 3a and a contribution value calculation module 4a.

The rigid transformation module 3a is configured to select a matching minutiae point pair as a reference minutiae point pair, and perform a rigid transformation on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide.

The contribution value calculation module 4a is configured to, for each matching minutiae point pair, calculate the contribution value of the matching minutiae point pair according to the matching minutiae point pair and the corresponding neighborhood minutiae point pair.

The contribution value is the median similarity value described in Embodiment 4.

The matching score calculation module 5a is configured to calculate the matching score between the first fingerprint image and the second fingerprint image according to the contribution values of all matching minutiae point pairs.

When calculating the contribution value of a matching minutiae point pair, this application not only calculates based on the two minutiae points of the matching minutiae point pair, but also calculates the neighborhood minutiae point pair within the neighborhood of the matching minutiae point pair, making full use of the matching minutiae point pair itself It can evaluate the similarity of two fingerprint images very well and improve the accuracy of fingerprint comparison.

As an improvement to the embodiment of the present application, the contribution numerical calculation module includes:

The first calculation unit is configured to calculate the first contribution value of the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair.

A rigid transformation unit, configured to perform rigidity on the second group of minutiae points of each neighborhood minutiae point pair within the set neighborhood range of the matching minutiae point pair so that the two minutiae points of the matching minutiae point pair coincide transform.

The second calculation unit is used to calculate the first-th value of each neighborhood detail point pair according to the distance and angle difference between the two detail points of the neighborhood detail point pair and the corresponding interior angle value for each neighborhood detail point pair. 2. Contribution value.

The third calculation unit is configured to calculate the contribution value of the matching minutiae point pair according to the first contribution value of the matching minutiae point pair and the second contribution values of all neighborhood minutiae point pairs.

Further, the first contribution value of the matching detail point pair can be calculated by the following formula;

sd＝Z(d _i ,μ ₁ ,τ ₁ )*Z(Δθ _i ,μ ₂ ,τ ₂ )

Wherein, pairscpre _i is the contribution value of the ith matching minutiae point pair, np and nq are respectively the number of the first group of minutiae points and the number of the second group of minutiae points in the set neighborhood of the matching minutiae point ;

.

The application can calculate the matching score between the first fingerprint image and the second fingerprint image through the following formula:

The aforementioned rigid transformation modules include:

The center of gravity calculation unit is used to calculate the center of gravity of the first group of minutiae points of all matching minutiae point pairs.

The reference minutiae point pair selection unit is used to find out the first group of minutiae points closest to the center of gravity from the first group of minutiae points of all matching minutiae point pairs, and the matching minutiae to which the found first group of minutiae points belong point pair as the reference detail point pair.

The fourth calculation unit is used to calculate the angle difference Δθ and the coordinate difference Δx, Δy of the two minutiae points of the reference minutiae point pair.

Δθ=θ2-θ1

Among them, (x2, y2, θ2) are the coordinates and direction angles of the first group of minutiae points of the reference minutiae point pair, and (x1, y1, θ1) are the coordinates of the second group of minutiae points of the reference minutiae point pair and orientation angle.

The transformation unit is configured to calculate the coordinates and direction angles of the second group of minutiae after rigid transformation according to the angle difference Δθ and the coordinate difference Δx, Δy between the two minutiae of the reference minutiae point pair.

θ'=θ-Δθ

The device of the present application may also include:

A comparison module, configured to obtain a fingerprint comparison result according to the matching score between the first fingerprint image and the second fingerprint image.

The implementation principle and technical effect of the device provided by the embodiment of the present application are the same as those of the aforementioned method embodiment 2. For a brief description, for the parts not mentioned in the device embodiment, please refer to the corresponding content in the aforementioned method embodiment 2 . Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the devices and units described above can refer to the corresponding process in the above method embodiment 2, which will not be repeated here.

Embodiment 6:

Another embodiment of the present application provides a fingerprint comparison device, as shown in Figure 16, the device includes:

The data preparation module 1b is used to acquire the first group of minutiae points of the first fingerprint image and the second group of minutiae points of the second fingerprint image.

In this embodiment, the data acquisition module 2 in Embodiment 4 may include a 2-nearest neighbor structure representation calculation module 2b, a local similarity matrix calculation module 3b, and a matching pair determination module 4b.

The 2-nearest neighbor structure representation calculation module 2b is configured to calculate the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points.

The local similarity matrix calculation module 3b is used to calculate the first group of minutiae points according to the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points and the local similarity matrix of the second set of minutiae points.

The matching pair determination module 4b is configured to select several elements from the local similarity matrix in descending order, and obtain two minutiae points corresponding to each selected element as a matching pair.

In this embodiment, the intermediate value calculation module 3 in Embodiment 4 may include a matching deterministic similarity matrix calculation module 5b.

Matching certainty similarity matrix calculation module 5b, for each matching pair, calculate the matching certainty of the matching pair according to the distance difference and angle difference between the local similarity matrix and the two minutiae points of the matching pair similarity matrix.

The matching deterministic similarity matrix is the median similarity value described in Embodiment 4.

The matching score calculation module 6b is used to select the matching detail point pairs of each matching deterministic similarity matrix and calculate the matching score of each matching deterministic similarity matrix; and calculate the total matching score according to all matching scores.

Among them, the matching certainty similarity matrix of each matching pair is calculated by the following formula:

Among them, the matching pair includes the a-th minutiae point in the first group of minutiae points and the b-th minutiae point in the second group of minutiae points, ml(i, j) is the i-th row and j-th column of the matching certainty similarity matrix elements of i=1,2,...,M, j=1,2,...,N, sl(i,j) is the element of row i, column j of the local similarity matrix, d,α,β are three-dimensional the absolute value of the three elements of vector v0,

The embodiment of the present application improves the calculation method of the matching deterministic similarity matrix, and combines the distance difference and angle difference information and the information related to the bounding box to calculate the matching deterministic similarity value, which can better measure the relationship between two detail points similarity, and help to select better matching detail point pairs. Compared with the existing technology, the test results on the test set have been greatly improved, and the evaluation indicators such as the equal error rate have been reduced.

In one of the examples, the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae is calculated by the following formula:

t _l , (x _l ,y _l ) and

dφ(t ₁ ,t ₂ ) is the angle difference function, which is defined as:

.

Correspondingly, the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points is calculated by the following formula:

t _h , (x _h ,y _h ) and

Exemplarily, the local similarity matrix is calculated by the following formula:

w _d ＝1,w _θ ＝0.3*180/π,

w _n =3, w _t =3, V=|V0|, V0=Fl _i -Gl _j .

Calculate the eigenvectors by the following formula

and

Among them, (x _a ,y _a ) and

As an improvement of the embodiment of the present application, the matching score calculation module includes:

The matching detail point pair determination unit is used for each matching deterministic similarity matrix, if the element ml(i, j) in the i-th row and j-th column is not the maximum value in its i-th row or not its j-th column The maximum value in , set ml(i,j) to 0.

The matching score calculation unit is used to calculate the matching score s of each matching certainty similarity matrix by the following formula:

.

The fingerprint comparison device of the present embodiment may also include:

The comparison result determination module is used to obtain the fingerprint comparison result according to the total matching score.

The implementation principle and technical effects of the device provided in this embodiment are the same as those of the aforementioned method embodiment 3. For a brief description, for the part not mentioned in the device embodiment, reference may be made to the corresponding content in the aforementioned method embodiment 3. Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the devices and units described above can refer to the corresponding process in the above method embodiment 3, which will not be repeated here.

Each module in the above-mentioned device can be fully or partially realized by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can invoke and execute the corresponding operations of the above-mentioned modules.

Embodiment 7:

The methods described in the above-mentioned embodiments 1-3 provided by this application can implement business logic through computer programs and record them on storage media, and the storage media can be read and executed by computers to realize the methods described in embodiments 1-3 of this specification. Describe the effects of the program. Therefore, the present application also provides a computer-readable storage medium for fingerprint comparison, including a memory for storing processor-executable instructions. When the instructions are executed by the processor, the steps of the fingerprint comparison method of Embodiments 1-3 are realized. .

In some embodiments, when the present application calculates the contribution value of a matching minutiae pair, it is not only calculated based on the two minutiae points of the matching minutiae point pair, but also calculated based on the neighborhood minutiae point pair within the neighborhood of the matching minutiae point pair, fully Utilizing the useful information contained in the matching minutiae point itself and within a certain range around it, the similarity of two fingerprint images is well evaluated, and the accuracy of fingerprint comparison is improved.

In some embodiments, the present application improves the calculation method of the matching deterministic similarity matrix, and combines the distance difference and angle difference information and the information related to the bounding box to calculate the matching deterministic similarity value, which can better measure the two The similarity between minutiae points helps to select better matching minutiae pairs. Compared with related technologies, the test results on the test set have been greatly improved, and the evaluation indicators such as the equal error rate have been reduced.

The storage medium may include a physical device for storing information, and information is usually digitized and then stored using an electrical, magnetic, or optical medium. Described storage medium can include: the device that utilizes electric energy mode to store information such as, various memory, as RAM, ROM etc.; USB stick; a device that stores information optically, such as a CD or DVD. Of course, there are other readable storage media, such as quantum memory, graphene memory and so on.

The above-mentioned storage medium may also include other implementations according to the descriptions of method embodiments 1-3. The implementation principle and technical effects of this embodiment are the same as those of the aforementioned method embodiments 1-3. For details, refer to related methods for implementation. The descriptions of Examples 1-3 are not repeated here.

Embodiment 8:

The present application also provides a device for fingerprint comparison, which may be a separate computer, or may include the actual operation of using one or more of the methods described in this specification or one or more embodiments of the device device etc. The device for fingerprint comparison may include at least one processor and a memory storing computer-executable instructions. When the processor executes the instructions, the steps of the fingerprint comparison method described in any one or more of the above-mentioned embodiments 1-3 are implemented. .

In some embodiments of the present application, when calculating the contribution value of a matching minutiae pair, it is not only calculated based on the two minutiae points of the matching minutiae point pair, but also calculated based on the neighborhood minutiae point pair within the neighborhood of the matching minutiae point pair, making full use of The useful information contained in the matching minutiae within a certain range of itself and its surroundings can be used to evaluate the similarity of two fingerprint images and improve the accuracy of fingerprint comparison.

In some embodiments of the present application, the calculation method of the matching deterministic similarity matrix is improved, and the distance difference and angle difference information and the information related to the limited box are fused to calculate the matching deterministic similarity value, which can better measure the two details The similarity between points, and helps to select a better pair of matching minutiae points. Compared with the existing technology, the test results on the test set have been greatly improved, and the evaluation indicators such as the equal error rate have been reduced.

The above-mentioned equipment can also include other implementations according to the descriptions of method embodiments 1-3. The implementation principle and technical effects of this embodiment are the same as those of the aforementioned method embodiments 1-3. For details, please refer to related method embodiments The descriptions of 1-3 will not be repeated here.

Finally, it should be noted that: the above-described embodiments are only specific implementations of the application, used to illustrate the technical solutions of the application, rather than limiting it, and the scope of protection of the application is not limited thereto, although referring to the aforementioned The embodiment has described this application in detail, and those of ordinary skill in the art should understand that any person familiar with this technical field can still modify the technical solutions described in the foregoing embodiments within the technical scope disclosed in this application Changes can be easily imagined, or equivalent replacements can be made to some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. All should be covered within the scope of protection of this application. Therefore, the protection scope of the present application should be determined by the protection scope of the claims.

Claims

A fingerprint comparison method, comprising:

Obtain the first set of minutiae points of the first fingerprint image and the second set of minutiae points of the second fingerprint image;

Obtain the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points;

For each matching minutiae point pair, calculate the median similarity value of the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair and the neighborhood minutiae point information; and

Calculate the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each matching pair of minutiae points.
The fingerprint comparison method according to claim 1, wherein, for each matching minutiae point pair, according to the distance and angle difference between two minutiae points of the matching minutiae point pair, and the neighborhood minutiae point information calculating the median similarity value of the pair of matching minutiae points; and calculating the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each minutiae point matching pair, including :

Selecting a matching minutiae point pair as a reference minutiae point pair, and performing a rigid transformation on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide;

For each matching minutiae point pair, calculate the contribution value of the matching minutiae point pair according to the matching minutiae point pair and the corresponding neighborhood minutiae point pair, wherein the contribution value is the median value of the similarity;

Wherein, the neighborhood minutiae point pair sets all matching minutiae point pairs within the neighborhood range for the matching minutiae point pair;

The matching score between the first fingerprint image and the second fingerprint image is calculated according to the contribution values of all matching minutiae point pairs.
The fingerprint comparison method according to claim 2, wherein, for each matching minutiae point pair, the contribution value of the matching minutiae point pair is calculated according to the matching minutiae point pair and the corresponding neighborhood minutiae point pair, include:

calculating the first contribution value of the matching minutiae point pair according to the distance and angle difference between two minutiae points of the matching minutiae point pair;

performing a rigid transformation on the second group of minutiae points of each neighborhood minutiae point pair within the set neighborhood range of the matching minutiae point pair so that the two minutiae points of the matching minutiae point pair coincide;

For each neighborhood minutiae point pair, calculate the second contribution value of each neighborhood minutiae point pair according to the distance and angle difference between the two minutiae points of the neighborhood minutiae point pair, and the corresponding interior angle value;

Wherein, the inner angle value is an angle value of an inner angle formed by two minutiae points of the neighborhood minutiae point pair and the matching minutiae point pair;

The contribution value of the matching minutiae point pair is calculated according to the first contribution value of the matching minutiae point pair and the second contribution values of all neighborhood minutiae point pairs.
The fingerprint comparison method according to claim 3, wherein the first contribution value of the matching minutiae point pair is calculated by the following formula;

sd＝Z(d i ,μ 1 ,τ 1 )*Z(Δθ i ,μ 2 ,τ 2 )

Wherein, sd is the first contribution value of the pair of matching minutiae points, Z(,,) is a sigmoid function, d i and Δθ i are respectively the distance and angle difference between two minutiae points of i matching minutiae point pairs, i =1,2,...,N, N is the number of matching minutiae point pairs, μ 1 , τ 1 , μ 2 , τ 2 are the set parameters;

The second contribution value of each neighborhood detail point pair is calculated by the following formula;

Among them, sn j is the second contribution value of the jth neighborhood detail point pair, j=1, 2,..., M, M is the neighborhood detail point pair within the neighborhood range of the matching detail point pair number, sn j = Z(d j ,μ 1 ,τ 1 )*Z(Δθ j ,μ 2 ,τ 2 )*Z(α j ,μ 3 ,τ 3 ), d j and Δθ j are respectively The distance and angle difference between the two minutiae points of the j neighborhood minutiae point pair, α j is the angle value of the interior angle formed by the two minutiae points of the jth neighborhood minutiae point pair and the matching minutiae point pair, μ 3 , τ 3 is the set parameter;

Calculate the contribution value of the matching detail point pair by the following formula;

Among them, pairscore i is the contribution value of the i-th matching minutiae point pair, np and nq are respectively the number of the first group of minutiae points and the number of the second group of minutiae points within the set neighborhood of the matching minutiae point ;
The fingerprint comparison method according to claim 4, wherein the matching score between the first fingerprint image and the second fingerprint image is calculated by the following formula:

Wherein, s is the matching score between the first fingerprint image and the second fingerprint image, N1 and N2 are the numbers of the first group of minutiae points and the number of the second group of minutiae points respectively.
The fingerprint comparison method according to claim 2, wherein said selecting a matching minutiae point pair as a reference minutiae point pair comprises:

Calculate the center of gravity of the first set of minutiae points of all matching minutiae point pairs;

Find a first group of minutiae points closest to the center of gravity from the first group of minutiae points of all matching minutiae point pairs, and use the matching minutiae point pair to which the found first group of minutiae points belong as the reference minutiae point pair .
The method for comparing fingerprints according to claim 2, wherein said all minutiae points in the second group of minutiae points are subjected to a rigid transformation such that two minutiae points of said reference minutiae point pair coincide, comprising:

Calculate the angle difference Δθ and the coordinate difference Δx, Δy of the two minutiae points of the reference minutiae point pair;

Δθ=θ2-θ1

Among them, (x2, y2, θ2) are the coordinates and direction angles of the first group of minutiae points of the reference minutiae point pair, and (x1, y1, θ1) are the coordinates of the second group of minutiae points of the reference minutiae point pair and orientation angle;

Calculate the coordinates and orientation angles of the second group of minutiae points after the rigid transformation according to the angle difference Δθ and the coordinate difference Δx, Δy of the two details of the reference minutiae point pair;

θ'=θ-Δθ

Among them, (x, y, θ) are the coordinates and orientation angles of the second group of minutiae points before rigid transformation, and (x′, y′, θ′) are the coordinates and orientation angles of the second group of minutiae points after rigid transformation.
The fingerprint comparison method according to any one of claims 2-7, wherein said method further comprises:

A fingerprint comparison result is obtained according to the matching score between the first fingerprint image and the second fingerprint image.
The method for comparing fingerprints according to claim 1, wherein said obtaining the matching minutiae pairs of the first group of minutiae points and the second group of minutiae points comprises:

calculating a 2-nearest neighbor structure representation of each minutiae in the first set of minutiae points and a 2-nearest neighbor structure representation of each minutiae point in the second set of minutiae points;

Calculate the first group of minutiae points and the second group of minutiae points according to the 2-nearest neighbor structure representation of each minutiae point in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points The local similarity matrix of ;

Select several elements from the local similarity matrix in descending order, and obtain two minutiae points corresponding to each selected element as a matching pair.
The fingerprint comparison method according to claim 9, wherein, for each matching minutiae point pair, according to the distance and angle difference between two minutiae points of the matching minutiae point pair, and the neighborhood minutiae point information calculating the median similarity value of the pair of matching minutiae points; and calculating the matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each minutiae point matching pair, including :

For each matching pair, calculate the matching certainty similarity matrix of the matching pair according to the local similarity matrix and the distance difference and angle difference between two minutiae points of the matching pair, wherein the matching determination The sexual similarity matrix is the median value of the similarity; and

Select the matching detail point pairs of each matching deterministic similarity matrix and calculate the matching score of each matching deterministic similarity matrix; and calculate the total matching score according to all matching scores.
The method for comparing fingerprints according to claim 10, wherein, the matching certainty similarity matrix of each matching pair is calculated by the following formula:

Wherein, the matching pair includes the a-th minutiae point in the first group of minutiae points and the b-th minutiae point in the second group of minutiae points, ml(i,j) is the i-th minutiae of the matching certainty similarity matrix The element in the jth column of the row, i=1, 2,..., M, j=1, 2,..., N, sl(i, j) is the element in the jth row of the local similarity matrix, d , α, β are the absolute values of the three elements of the three-dimensional vector v0,
is the eigenvector obtained by performing the polar coordinate operation on the i-th minutiae point in the first group of minutiae points with reference to the a-th minutiae point in the first group of minutiae points,
is a feature vector obtained by performing polar coordinate operation on the jth minutiae point in the second group of minutiae points with reference to the bth minutiae point in the second group of minutiae points.
The fingerprint comparison method according to claim 11, wherein the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae is calculated by the following formula:

Among them, Fl i is the 2-nearest neighbor structure representation of the i-th minutiae F i in the first group of minutiae points, p and q are the two minutiae points F p and F q closest to the i-th minutiae F i respectively serial number,
t l , (x l ,y l ) and
Respectively, the type, coordinates and angle of the lth detail point F i , l takes i, p, q;

n ip is the number of ridge lines on the line between the i-th detail point and the p-th detail point in the first group of detail points, and n iq is the i-th detail point and the q-th detail point in the first group of detail points The number of ridges on the connection;

dφ(t 1 ,t 2 ) is the angle difference function, which is defined as:

Calculate the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points by the following formula:

Among them, Gl j is the 2-nearest neighbor structure representation of the jth minutiae G j in the second group of minutiae points, m and n are the two minutiae points G m and G n closest to the jth minutiae G j respectively serial number,
t h , (x h ,y h ) and
Respectively, the type, coordinates and angle of the hth detail point G h , where h takes j, m, n;

n jm is the number of ridge lines on the line between the jth minutiae point and the mth minutiae point in the second group of minutiae points, and n jn is the jth minutiae point and the nth minutiae point in the second group of minutiae points The number of ridges on the link.
The method for comparing fingerprints according to claim 12, wherein the local similarity matrix is calculated by the following formula:

Among them, A is the dot product of the weighting coefficient vector W and the vector V;

w d ＝1,w θ ＝0.3*180/π,
w n =3, w t =3, V=|V0|, V0=Fl i -Gl j .
The fingerprint comparison method according to claim 13, wherein, the feature vector is calculated by the following formula
and

Among them, (x a ,y a ) and
are respectively the coordinates and angles of the ath minutiae point in the first group of minutiae points, (x b , y b ) and
are the coordinates and angles of the bth minutiae point in the second group of minutiae points, respectively.
The fingerprint comparison method according to any one of claims 9-14, wherein said selecting matching detail point pairs of each matching deterministic similarity matrix and calculating the matching score of each matching deterministic similarity matrix comprises:

For each matching deterministic similarity matrix, if the element ml(i,j) in the i-th row and j-th column is not the maximum value in its i-th row or not the maximum value in its j-th column, then ml( i, j) is set to 0;

The matching score s of each matching deterministic similarity matrix is calculated by the following formula:
The fingerprint comparison method according to claim 2, wherein said method further comprises:

After the total matching score is calculated according to all the matching scores, the fingerprint comparison result is obtained according to the total matching score.
A fingerprint comparison device, said device comprising:

The data preparation module is used to obtain the first group of minutiae points of the first fingerprint image and the second group of minutiae points of the second fingerprint image;

An acquisition module, configured to acquire the matching minutiae point pairs of the first group of minutiae points and the second group of minutiae points;

An intermediate value calculation module, for each matching detail point pair, according to the distance and angle difference between two detail points of the matching detail point pair, and the neighborhood detail point information to calculate the similarity of the matching detail point pair degree intermediate value;

A matching score calculation module, configured to calculate a matching score between the first fingerprint image and the second fingerprint image according to the median similarity value of each matching pair of minutiae points.
The fingerprint comparison device according to claim 17, wherein the intermediate value calculation module includes a rigid transformation module and a contribution value calculation module:

The rigid transformation module is configured to select a matching minutiae point pair as a reference minutiae point pair, and perform a rigid transformation on all minutiae points in the second group of minutiae points so that the two minutiae points of the reference minutiae point pair coincide ;

The contribution value calculation module is used to calculate the contribution value of the matching detail point pair according to the matching detail point pair and the corresponding neighborhood detail point pair for each matching detail point pair;

Wherein, the neighborhood minutiae point pair sets all matching minutiae point pairs within the neighborhood range for the matching minutiae point pair;

The matching score calculation module calculates the matching score between the first fingerprint image and the second fingerprint image according to the contribution values of all matching minutiae point pairs.
The fingerprint comparison device according to claim 17, wherein the acquisition module includes a 2-nearest neighbor structure representation calculation module, a local similarity matrix calculation module, and a matching pair determination module, and the intermediate value calculation module includes a matching deterministic similarity Degree matrix calculation module:

The 2-nearest neighbor structure representation calculation module is used to calculate the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points ;

The local similarity matrix calculation module is used to calculate according to the 2-nearest neighbor structure representation of each minutiae in the first group of minutiae points and the 2-nearest neighbor structure representation of each minutiae point in the second group of minutiae points The local similarity matrix of the first group of minutiae points and the second group of minutiae points;

The matching pair determination module is configured to select several elements from the local similarity matrix in descending order, and obtain two minutiae points corresponding to each selected element as a matching minutiae point pair;

The matching deterministic similarity matrix calculation module, for each matching pair, calculates the matching determination of the matching pair according to the distance difference and angle difference between the local similarity matrix and the two minutiae points of the matching pair sex similarity matrix;

The matching score calculation module is used to select the matching detail point pairs of each matching deterministic similarity matrix and calculate the matching score of each matching deterministic similarity matrix; and calculate the total matching score according to all matching scores.
A computer-readable storage medium for fingerprint comparison, characterized in that it includes a memory for storing processor-executable instructions, and when the instructions are executed by the processor, it implements any of claims 1-16. Describe the steps of the fingerprint comparison method.
A device for fingerprint comparison, characterized in that it includes at least one processor and a memory storing computer-executable instructions, and when the processor executes the instructions, the fingerprint described in any one of claims 1-16 is realized. Steps of the comparison method.