Fingerprint recognition network training methods, fingerprint recognition methods, devices and media

Abstract

This application provides a training method for a fingerprint recognition network, a fingerprint recognition method, an apparatus, and a medium. The training method for the fingerprint recognition network includes: determining fingerprint feature matching maps based on a first initial fingerprint image and a second initial fingerprint image; forming a training sample set from multiple fingerprint feature matching maps; and training an original neural network based on the training sample set to obtain a fingerprint recognition network. The technical solution provided by this application solves the problem of difficulty in distinguishing whether two different fingerprint images belong to the same finger, thus reducing the false acceptance and false recognition rates of fingerprint recognition.

Description

Technical Field

[0001] This article relates to image recognition technology, and more particularly to a training method for a fingerprint recognition network, a fingerprint recognition method, a device, and a medium. Background Technology

[0002] With the rapid development of computers and networks, people have placed higher demands on the accuracy, security, and practicality of identity authentication. Among the many biometric technologies, fingerprint recognition technology is the earliest developed and most widely used. Fingerprint recognition technology fully utilizes the universality, uniqueness, and permanent biometric characteristics of fingerprints, and has gradually replaced traditional identification methods based on marks and numbers. It is currently widely used in industries such as networking, banking, finance, healthcare, and security checks.

[0003] Most current fingerprint recognition algorithms compare the fingerprint to be identified with a template fingerprint, and determine whether the fingerprint to be identified is the template fingerprint based on the comparison result. During the comparison, due to the small area of ​​fingerprint images, their high degree of texture repetition, and the drastic changes in fingerprint image quality caused by the imaging environment, it becomes difficult to distinguish whether two fingerprint images belong to the same finger, affecting the false acceptance rate and false acceptance rate of the recognition algorithm. Summary of the Invention

[0004] This application provides a method for training a fingerprint recognition network, a fingerprint recognition method, an apparatus, and a medium, which can identify whether two fingerprint images belong to the same finger, thereby improving the accuracy of fingerprint recognition.

[0005] This application provides a method for training a fingerprint recognition network, the method including:

[0006] A fingerprint feature matching map is determined based on the first and second initial fingerprint images;

[0007] A training sample set is formed by combining multiple fingerprint feature matching maps;

[0008] The original neural network is trained based on the training sample set to obtain the fingerprint recognition network.

[0009] In an exemplary embodiment of this application, determining the fingerprint feature matching map based on the first initial fingerprint image and the second initial fingerprint image may include:

[0010] Obtain the first feature point of the first initial fingerprint image and the second feature point of the second initial fingerprint image;

[0011] The first feature point and the second feature point are matched to determine the matching feature point pair;

[0012] The transformation matrix is ​​determined based on the matched feature point pairs.

[0013] In an exemplary embodiment of this application, determining the fingerprint feature matching map based on the first initial fingerprint image and the second initial fingerprint image may include:

[0014] Using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix, a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system are obtained.

[0015] The fingerprint feature matching map is obtained based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map.

[0016] In an exemplary embodiment of this application, obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix may include:

[0017] The coordinate system is defined as the coordinate system in which one of the first and second initial fingerprint images is located, and the initial fingerprint image is the first matching fingerprint image in the matching coordinate system.

[0018] Obtain the first matching feature map of the first matching fingerprint image;

[0019] The transformation matrix is ​​used to transform another fingerprint image from the first initial fingerprint image and the second initial fingerprint image to the matching coordinate system, thereby obtaining the second matching fingerprint image in the matching coordinate system.

[0020] Obtain the first matching feature map of the second matching fingerprint image.

[0021] In an exemplary embodiment of this application, obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix may include:

[0022] Obtain the first initial feature map of the first initial fingerprint image and the second initial feature map of the second initial fingerprint image;

[0023] The coordinate system of one of the first initial fingerprint images and the second initial fingerprint image is used as the matching coordinate system. The initial fingerprint image is the first matching fingerprint image under the matching coordinate system, and the initial feature map corresponding to the first matching fingerprint image is the first matching feature map under the matching coordinate system.

[0024] The transformation matrix is ​​used to transform another fingerprint image from the first initial fingerprint image and the second initial fingerprint image, as well as the initial feature map corresponding to the other fingerprint image, into the matching coordinate system to obtain the second matching fingerprint image and the second matching feature map in the matching coordinate system.

[0025] In an exemplary embodiment of this application, obtaining the first fingerprint image, the second fingerprint image, the first feature map, the second feature map, and the matching coding map in the matching coordinate system using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix may include:

[0026] The matching feature point pairs are transformed to the matching coordinate system using the transformation matrix to obtain fingerprint matching point pairs in the matching coordinate system.

[0027] Determine the matching position corresponding to each fingerprint matching point pair;

[0028] Mark the Gaussian kernel at all the matching positions to obtain the matching encoding map in the matching coordinate system.

[0029] In an exemplary embodiment of this application, obtaining the fingerprint feature matching map based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map may include:

[0030] The fingerprint feature matching map is obtained by overlaying the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map.

[0031] This application also provides a fingerprint recognition method, which may include:

[0032] Obtain fingerprint feature matching maps of two fingerprint images to be identified;

[0033] The fingerprint feature matching map is input into the fingerprint recognition network obtained by the training method according to any one of claims 1 to 7 to obtain the recognition result;

[0034] Based on the recognition results, it is determined whether the two fingerprint images to be recognized belong to the same finger.

[0035] In an exemplary embodiment of this application, obtaining the fingerprint feature matching map of two fingerprint images to be identified may include:

[0036] Obtain feature points from the two fingerprint images to be identified;

[0037] The feature points in the two fingerprint images to be identified are matched to determine the matching feature point pairs;

[0038] The transformation matrix is ​​determined based on the matched feature point pairs.

[0039] In an exemplary embodiment of this application, obtaining the fingerprint feature matching map of two fingerprint images to be identified may include:

[0040] By acquiring two fingerprint images to be identified and the transformation matrix, a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system are obtained.

[0041] The fingerprint feature matching map is obtained based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map.

[0042] This application also provides a computer-readable storage medium that can store one or more programs that can be executed by one or more processors to implement the training method of the fingerprint recognition network described in any of the above claims, or to implement the fingerprint recognition method described in any of the above claims.

[0043] This application also provides a training apparatus for a fingerprint recognition network, which may include a memory and a processor. The memory stores a program, which, when read and executed by the processor, implements the fingerprint recognition network training method described in any of the above claims.

[0044] This application also provides a fingerprint recognition device, which may include a memory and a processor. The memory stores a program, which, when read and executed by the processor, implements the fingerprint recognition method described in any of the above claims.

[0045] Compared with related technologies, the embodiments of this application may include: determining fingerprint feature matching maps based on a first initial fingerprint image and a second initial fingerprint image; forming a training sample set from multiple fingerprint feature matching maps; and training the original neural network based on the training sample set to obtain a fingerprint recognition network. The fingerprint recognition network trained using this embodiment fully utilizes the feature information of each fingerprint image in two different fingerprint images, enabling effective identification and differentiation between two different fingerprint images. This solves the problem of difficulty in distinguishing whether two different fingerprint images belong to the same finger, reducing the false acceptance and false recognition rates of fingerprint recognition.

[0046] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. Other advantages of this application can be realized and obtained by means of the solutions described in the description and the accompanying drawings. Attached Figure Description

[0047] The accompanying drawings are used to provide an understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0048] Figure 1 A flowchart illustrating the training method for a fingerprint recognition network provided in this application embodiment;

[0049] Figure 2 A flowchart illustrating a method for determining a fingerprint feature matching map based on a first initial fingerprint image and a second initial fingerprint image, provided in an embodiment of this application;

[0050] Figure 3 A flowchart of a method for determining a fingerprint feature matching map based on a first initial fingerprint image and a second initial fingerprint image, provided in another embodiment of this application;

[0051] Figure 4 A flowchart illustrating the method for obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, and a second matching feature map in a matching coordinate system, as provided in the embodiments of this application.

[0052] Figure 5 A flowchart of a method for obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, and a second matching feature map in a matching coordinate system, provided in another embodiment of this application;

[0053] Figure 6 A flowchart illustrating a method for obtaining a matching encoding map in a matching coordinate system, provided in an embodiment of this application;

[0054] Figure 7A flowchart illustrating the method for obtaining a fingerprint feature matching map provided in this application embodiment;

[0055] Figure 8 A schematic diagram illustrating the overlay generation of fingerprint feature matching maps provided in an embodiment of this application;

[0056] Figure 9 A flowchart of the fingerprint recognition method provided in the embodiments of this application;

[0057] Figure 10 A flowchart illustrating a method for obtaining fingerprint feature matching maps of two fingerprint images to be identified, provided in an embodiment of this application;

[0058] Figure 11 A flowchart illustrating a method for obtaining fingerprint feature matching maps of two fingerprint images to be identified, provided in another embodiment of this application;

[0059] Figure 12 This is a structural diagram of a training device for a fingerprint recognition network provided in an embodiment of this application. Detailed Implementation

[0060] This application describes several embodiments, but these descriptions are exemplary and not restrictive, and it will be apparent to those skilled in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with, or may replace, any feature or element of any other embodiment.

[0061] This application includes and contemplates combinations of features and elements known to those skilled in the art. The embodiments, features, and elements disclosed in this application may also be combined with any conventional features or elements to form a unique inventive scheme as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive schemes to form another unique inventive scheme as defined by the claims. Therefore, it should be understood that any feature shown and / or discussed in this application may be implemented individually or in any suitable combination. Therefore, the embodiments are not limited except by the limitations imposed by the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

[0062] Furthermore, in describing representative embodiments, the specification may have presented methods and / or processes as a specific sequence of steps. However, the method or process should not be limited to the specific order of steps described herein, to the extent that it does not depend on such a specific order. As will be understood by those skilled in the art, other sequences of steps are also possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation of the claims. Moreover, the claims concerning the method and / or process should not be limited to the steps performed in the written order, and those skilled in the art will readily understand that these orders can be varied and still remain within the spirit and scope of the embodiments of this application.

[0063] This application provides a method for training a fingerprint recognition network, such as... Figure 1 As shown, the method includes:

[0064] Step S101: Determine the fingerprint feature matching map based on the first initial fingerprint image and the second initial fingerprint image;

[0065] In an exemplary embodiment of this application, the first initial fingerprint image and the second initial fingerprint image may belong to the same finger or different fingers; the first initial fingerprint image and the second initial fingerprint image belonging to the same finger may be acquired from different angles;

[0066] In an exemplary embodiment of this application, the fingerprint feature matching map includes feature information of both a first initial fingerprint image and a second initial fingerprint image.

[0067] Step S102: Combine multiple fingerprint feature matching maps into a training dataset;

[0068] If the first initial fingerprint image and the second initial fingerprint image belong to the same finger, the obtained fingerprint feature matching map is a positive sample in the sample training set; if the first initial fingerprint image and the second initial fingerprint image do not belong to the same finger, the obtained fingerprint feature matching map is a negative sample in the sample training set.

[0069] In an exemplary embodiment of this application, the training sample set includes both positive and negative samples;

[0070] Step S103: Train the original neural network based on the training sample set to obtain the fingerprint recognition network;

[0071] In an exemplary embodiment of this application, the training objective of the fingerprint neural network is: if a positive sample is input, the output indicates that the first initial fingerprint image and the second initial fingerprint image belong to the same finger; if a negative sample is output, the output indicates that the first initial fingerprint image and the second initial fingerprint image do not belong to the same finger.

[0072] In one exemplary embodiment, the original neural network can be a binary classification neural network.

[0073] The fingerprint recognition network trained by the method described in the embodiments of this application can effectively identify and distinguish between two different fingerprint images by making full use of the feature information of each fingerprint image in two different fingerprint images.

[0074] This application provides a method for determining a fingerprint feature matching map based on a first initial fingerprint image and a second initial fingerprint image, such as... Figure 2 As shown, the method includes:

[0075] Step S201: Obtain the first feature point of the first initial fingerprint image and the second feature point of the second initial fingerprint image;

[0076] The methods for obtaining the first feature point of the first initial fingerprint image and the second feature point of the second initial fingerprint image include, but are not limited to, the SIFT (Scale-invariant feature transform) method or a method based on a deep neural network.

[0077] In one exemplary embodiment, the number of first feature points in the first initial fingerprint image is the same as the number of second feature points in the second initial fingerprint image;

[0078] In one exemplary embodiment, the number of first feature points in the first initial fingerprint image is different from the number of second feature points in the second initial fingerprint image;

[0079] Step S202: Match the first feature point and the second feature point to determine the matching feature point pair;

[0080] In an exemplary embodiment of this application, matching a first feature point and a second feature point to determine a matching feature point pair includes:

[0081] Step S2021: Obtain the first feature descriptor of the first feature point and the second feature descriptor of the second feature point;

[0082] The methods for obtaining the first feature descriptor of the first feature point and the second feature descriptor of the second feature point include, but are not limited to, the SIFT (Scale-invariant feature transform) method or a method based on a deep neural network.

[0083] Step S2022: Match the first feature point and the second feature point according to the first feature descriptor and the second feature descriptor to obtain a matching feature point pair;

[0084] In an exemplary embodiment, a first feature point and a second feature point can be matched based on the distance and direction between the first feature descriptor and the second feature descriptor, wherein the matching method includes, but is not limited to, brute-force matching, k-nearest neighbor matching, and RANSAC matching;

[0085] In an exemplary embodiment, the number of first feature points in the first initial fingerprint image is the same as the number of second feature points in the second initial fingerprint image, and the first feature points and the second feature points correspond one-to-one.

[0086] In one exemplary embodiment, the number of first feature points in the first initial fingerprint image is different from the number of second feature points in the second initial fingerprint image, but there are first feature points and second feature points of the same number that correspond one-to-one.

[0087] Step S203: Determine the transformation matrix based on the matching feature point pairs;

[0088] In an exemplary embodiment, the transformation matrix pair between matching feature point pairs can be calculated using the least squares algorithm, and the transformation matrix can be determined based on the transformation matrix pair.

[0089] Transformation matrices can include one or more of the following: rigid transformation matrix, affine transformation matrix, similarity transformation matrix, and perspective transformation matrix.

[0090] This application provides a method for determining a fingerprint feature matching map based on a first initial fingerprint image and a second initial fingerprint image, such as... Figure 3 As shown, the method includes:

[0091] Step S301: Using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix, obtain the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map in the matching coordinate system;

[0092] In an exemplary embodiment of this application, the matching coordinate system is either the coordinate system where the first initial fingerprint image is located or the coordinate system where the second initial fingerprint image is located;

[0093] In an exemplary embodiment of this application, the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map are placed in the same coordinate system to facilitate subsequent alignment and overlay operations.

[0094] Step S302: Obtain a fingerprint feature matching map based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map;

[0095] The fingerprint feature matching map, obtained from the first matched fingerprint image, the second matched fingerprint image, the first matched feature map, the second matched feature map, and the matching coding map, contains various feature information of the first initial fingerprint image and the second initial fingerprint image. The fingerprint recognition network trained by the fingerprint feature matching map makes full use of various feature information and can effectively identify and distinguish the first initial fingerprint image and the second initial fingerprint image.

[0096] This application provides a method for obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, and a second matching feature map in a matching coordinate system, such as... Figure 4 As shown, the method includes:

[0097] Step S401: Take the coordinate system of one of the first and second initial fingerprint images as the matching coordinate system, and the fingerprint image is the first matching fingerprint image under the matching coordinate system;

[0098] In an exemplary embodiment, the coordinate system where the first initial fingerprint image is located is used as the matching coordinate system. In this case, the first initial fingerprint image is the first matching fingerprint image under the matching coordinate system.

[0099] In an exemplary embodiment, the coordinate system where the second initial fingerprint image is located is used as the matching coordinate system. In this case, the second initial fingerprint image is the first matching fingerprint image under the matching coordinate system.

[0100] Step S402: Obtain the first matching feature map of the first matching fingerprint image;

[0101] In an exemplary embodiment of this application, the first matching feature map may be one or more of the following: fingerprint minutiae probability map, fingerprint quality map, fingerprint orientation field, etc., of the first matching fingerprint image.

[0102] Step S403: Using a transformation matrix, transform another fingerprint image from the first initial fingerprint image and the second initial fingerprint image to the matching coordinate system to obtain the second matching fingerprint image in the matching coordinate system.

[0103] In an exemplary embodiment, the coordinate system where the first initial fingerprint image is located is taken as the matching coordinate system. At this time, the first initial fingerprint image is the first matching fingerprint image under the matching coordinate system. By transforming the second initial fingerprint image to the matching coordinate system, the second matching fingerprint image under the matching coordinate system is obtained.

[0104] In an exemplary embodiment, the coordinate system where the second initial fingerprint image is located is used as the matching coordinate system. At this time, the second initial fingerprint image is the first matching fingerprint image under the matching coordinate system. By transforming the first initial fingerprint image to the matching coordinate system, the second matching fingerprint image under the matching coordinate system is obtained.

[0105] During the process of transforming the first initial fingerprint image or the second initial fingerprint image to the matching coordinate system by using a transformation matrix, there are unfilled areas in the transformed first initial fingerprint image or the second initial fingerprint image. The unfilled areas are set to zero to obtain the second matching fingerprint image, so that the first matching fingerprint image and the second matching fingerprint image can be aligned.

[0106] Step S404: Obtain the first matching feature map of the second matching fingerprint image;

[0107] In an exemplary embodiment of this application, the second matching feature map corresponds to the same features as the first matching feature map, which can be one or more of the fingerprint minutiae probability map, fingerprint quality map, fingerprint orientation field, etc. of the second matching fingerprint image; for example, if the first matching feature map is the fingerprint minutiae probability map of the first matching fingerprint image, then the second matching feature map is the fingerprint minutiae probability map of the second matching fingerprint image.

[0108] This application provides a method for obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, and a second matching feature map in a matching coordinate system, such as... Figure 5 As shown, the method includes:

[0109] Step S501: Obtain the first initial feature map of the first initial fingerprint image and the second initial feature map of the second initial fingerprint image;

[0110] In an exemplary embodiment of this application, the first matching feature map may be one or more of the following: fingerprint minutiae probability map, fingerprint quality map, fingerprint orientation field, etc., of the first matching fingerprint image; the second matching feature map may be one or more of the following: fingerprint minutiae probability map, fingerprint quality map, fingerprint orientation field, etc., of the second matching fingerprint image; for example, if the first matching feature map is the fingerprint minutiae probability map of the first matching fingerprint image, then the second matching feature map is the fingerprint minutiae probability map of the second matching fingerprint image.

[0111] Step S502: Using the coordinate system of one of the first initial fingerprint images and the second initial fingerprint image as the matching coordinate system, the initial fingerprint image is the first matching fingerprint image under the matching coordinate system, and the initial feature map corresponding to the first matching fingerprint image is the first matching feature map under the matching coordinate system.

[0112] In an exemplary embodiment, the coordinate system where the first initial fingerprint image is located is used as the matching coordinate system. In this case, the first initial fingerprint image is the first matching fingerprint image under the matching coordinate system, and the first initial feature map is the first matching feature map under the matching coordinate system.

[0113] In an exemplary embodiment, the coordinate system where the second initial fingerprint image is located is used as the matching coordinate system. In this case, the second initial fingerprint image is the first matching fingerprint image under the matching coordinate system, and the second initial feature map is the first matching feature map under the matching coordinate system.

[0114] Step S503: By transforming the first initial fingerprint image and the second initial fingerprint image, another fingerprint image and the initial feature map corresponding to the other fingerprint image are transformed to the matching coordinate system to obtain the second matching fingerprint image and the second matching feature map in the matching coordinate system.

[0115] In an exemplary embodiment, the coordinate system where the first initial fingerprint image is located is used as the matching coordinate system. At this time, the first initial fingerprint image is the first matching fingerprint image under the matching coordinate system. By transforming the matrix, the second initial fingerprint image and the second initial feature map are transformed to the matching coordinate system to obtain the second matching fingerprint image and the second matching feature map under the matching coordinate system.

[0116] In an exemplary embodiment, the coordinate system where the second initial fingerprint image is located is used as the matching coordinate system. At this time, the second initial fingerprint image is the first matching fingerprint image under the matching coordinate system. By transforming the first initial fingerprint image and the first initial feature map to the matching coordinate system, the second matching fingerprint image and the second matching feature map under the matching coordinate system are obtained.

[0117] During the process of transforming the first initial fingerprint image and the first initial feature map, or the second initial fingerprint image and the second initial feature map, to the matching coordinate system using a transformation matrix, there are unfilled areas in the transformed first initial fingerprint image and the first initial feature map, or the second initial fingerprint image and the second initial feature map. These unfilled areas are set to zero to obtain the second matching fingerprint image and the second matching feature map, so that the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, and the second matching feature map can be aligned.

[0118] This application provides a method for obtaining a matching encoding map in a matching coordinate system, such as... Figure 6 As shown, the method includes:

[0119] Step S601: Transform the matching feature point pairs to the matching coordinate system using a transformation matrix to obtain fingerprint matching point pairs in the matching coordinate system;

[0120] In an exemplary embodiment, the coordinate system where the first initial fingerprint image is located is used as the matching coordinate system. At this time, the first feature point in the matching feature point is in the matching coordinate system. The second feature point in the matching feature point pair is transformed to the matching coordinate system by the transformation matrix to obtain the fingerprint matching point pair in the matching coordinate system.

[0121] In an exemplary embodiment, the coordinate system where the second initial fingerprint image is located is used as the matching coordinate system. At this time, the second feature point in the matching feature point is in the matching coordinate system. The first feature point in the matching feature point pair is transformed to the matching coordinate system by the transformation matrix to obtain the fingerprint matching point pair in the matching coordinate system.

[0122] Step S602: Determine the matching position corresponding to each fingerprint matching point pair;

[0123] In an exemplary embodiment of this application, the matching position is the average coordinate position of the fingerprint matching point pair;

[0124] Step S603: Mark the Gaussian kernel at all matching positions to obtain the matching encoding map in the matching coordinate system;

[0125] Preferably, the Gaussian kernel can be 3-5 pixels.

[0126] This application provides a method for obtaining a fingerprint feature matching map, such as... Figure 7 As shown, the method includes:

[0127] Step S701: Overlay the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map to obtain a fingerprint feature matching map;

[0128] The stacking order of the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map is not limited;

[0129] In an exemplary embodiment of this application, the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map are aligned and superimposed, with the superposition direction being the depth direction; the number of channels in the fingerprint feature matching map is equal to the total number of images participating in the superposition.

[0130] In an exemplary embodiment, the superposition of the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map is as follows: Figure 8 As shown, Figure 8 The left side, from bottom to top, consists of the first matching fingerprint image 8g, the second matching fingerprint image 8f, the first matching feature map, and the second matching feature map. Figure 8 e-8b, Matching Encoding Figure 8a, where the first matching feature Figure 8 e and second matching features Figure 8 d represents the fingerprint quality images of the first matching fingerprint image 8g and the second matching fingerprint image 8f, respectively, and the first matching feature Figure 8 c and second matching features Figure 8 b represents the fingerprint detail maps of the first matched fingerprint image 8g and the second matched fingerprint image 8f, respectively; the first matched fingerprint image 8g, the second matched fingerprint image 8f, the first matched feature map, and the second matched feature map are compared. Figure 8 e-8b, Matching Encoding Figure 8 Align and overlap 'a', such as Figure 8 As shown on the right; the fingerprint feature matching map obtained by superposition has 7 channels;

[0131] In the technical solution provided in the embodiments of this application, the fingerprint feature matching map includes all feature information of the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map. Using the fingerprint feature matching map as a sample to train the neural network can effectively distinguish between true and false fingerprint matches. Even when the number of matching feature point pairs is insufficient, other fingerprint feature information can still ensure that the fingerprint recognition result is true and valid.

[0132] This application also provides a fingerprint recognition method, such as... Figure 9 As shown, the method includes:

[0133] Step S801: Obtain fingerprint feature matching images of two fingerprint images to be identified;

[0134] Step S802: Input the fingerprint feature matching map into the fingerprint recognition network obtained by the training method according to any of the above embodiments to obtain the recognition result;

[0135] Step S803: Determine whether the two fingerprint images to be identified belong to the same finger based on the recognition results;

[0136] According to the method described in the embodiments of this application, the fingerprint recognition network trained using fingerprint feature matching maps as samples makes full use of the feature information of each fingerprint image in two different fingerprint images, and can effectively identify and distinguish two different fingerprint images, solving the problem of not being able to distinguish whether two different fingerprint images belong to the same finger, and reducing the rejection rate and false recognition rate of fingerprint recognition.

[0137] This application provides a method for obtaining fingerprint feature matching maps of two fingerprint images to be identified, such as... Figure 10 As shown, the method includes:

[0138] Step S901: Obtain feature points from two fingerprint images to be identified;

[0139] Methods for obtaining feature points from two fingerprint images to be identified include, but are not limited to, SIFT (Scale-invariant feature transform) or methods based on deep neural networks;

[0140] Step S902: Match the feature points in the two fingerprint images to be identified to determine the matching feature point pairs;

[0141] In an exemplary embodiment of this application, matching feature points in two fingerprint images to be identified to determine a matching feature point pair includes:

[0142] Step S9021: Obtain feature descriptors of feature points in two fingerprint images to be identified;

[0143] Methods for obtaining feature descriptors of feature points in two fingerprint images to be identified include, but are not limited to, SIFT (Scale-invariant feature transform) or methods based on deep neural networks;

[0144] Step S9022: Based on the feature descriptors of the feature points in the two fingerprint images to be identified, match the feature points in the two fingerprint images to be identified to obtain a matching feature point pair;

[0145] In an exemplary embodiment, feature points in two fingerprint images to be identified can be matched based on the distance and direction between feature descriptors of feature points in the two fingerprint images to be identified. The matching method includes, but is not limited to, brute-force matching, k-nearest neighbor matching, and RANSAC matching.

[0146] Step S903: Determine the transformation matrix based on the matching feature point pairs;

[0147] In an exemplary embodiment, the transformation matrix pair between matching feature point pairs can be calculated using the least squares algorithm, and the transformation matrix can be determined based on the transformation matrix pair.

[0148] Transformation matrices can include one or more of the following: rigid transformation matrix, affine transformation matrix, similarity transformation matrix, and perspective transformation matrix.

[0149] This application provides a method for obtaining fingerprint feature matching maps of two fingerprint images to be identified, such as... Figure 11 As shown, the method includes:

[0150] Step S1001: By acquiring two fingerprint images to be identified and the transformation matrix, a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system are obtained.

[0151] In an exemplary embodiment of this application, the matching coordinate system is the coordinate system of either of the two fingerprint images to be identified;

[0152] In an exemplary embodiment of this application, the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map are placed in the same coordinate system to facilitate subsequent alignment and overlay operations.

[0153] Step S1002: Obtain a fingerprint feature matching map based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map.

[0154] The fingerprint feature matching map, obtained from the first matched fingerprint image, the second matched fingerprint image, the first matched feature map, the second matched feature map, and the matching coding map, contains various feature information of the two fingerprint images to be identified. The fingerprint recognition network trained by the fingerprint feature matching map makes full use of various feature information and can effectively identify and distinguish the two fingerprint images to be identified.

[0155] This application also provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the fingerprint recognition network training method as described in any of the preceding embodiments.

[0156] This application also provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the fingerprint recognition method as described in any of the preceding embodiments.

[0157] This application also provides a training device for a fingerprint recognition network, such as... Figure 10 As shown, it includes a memory 11 and a processor 12. The memory 11 stores a program, which, when read and executed by the processor 12, implements the training method of the fingerprint recognition network as described in any of the previous embodiments.

[0158] This application also provides a fingerprint recognition device, including a memory and a processor. The memory stores a program, which, when read and executed by the processor, implements the fingerprint recognition method as described in any of the preceding embodiments.

[0159] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A method for training a fingerprint recognition network, the method comprising: A fingerprint feature matching map is determined based on the first and second initial fingerprint images; A training sample set is formed by combining multiple fingerprint feature matching maps; The original neural network is trained based on the training sample set to obtain a fingerprint recognition network; The step of determining the fingerprint feature matching map based on the first initial fingerprint image and the second initial fingerprint image includes: Obtain the first feature point of the first initial fingerprint image and the second feature point of the second initial fingerprint image; The first feature point and the second feature point are matched to determine the matching feature point pair; Determine the transformation matrix based on the matched feature point pairs; The step of determining the fingerprint feature matching map based on the first initial fingerprint image and the second initial fingerprint image includes: Using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix, a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system are obtained. The fingerprint feature matching map is obtained based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map; The step of obtaining the first fingerprint image, the second fingerprint image, the first feature map, the second feature map, and the matching encoding map in the matching coordinate system using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix includes: The matching feature point pairs are transformed to the matching coordinate system using the transformation matrix to obtain fingerprint matching point pairs in the matching coordinate system. Determine the matching position corresponding to each fingerprint matching point pair; Mark the Gaussian kernel at all the matching positions to obtain the matching encoding map in the matching coordinate system.

2. The training method according to claim 1, characterized in that, The step of obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix includes: The coordinate system is defined as the coordinate system in which one of the first and second initial fingerprint images is located, and the initial fingerprint image is the first matching fingerprint image in the matching coordinate system. Obtain the first matching feature map of the first matching fingerprint image; The transformation matrix is ​​used to transform another fingerprint image from the first initial fingerprint image and the second initial fingerprint image to the matching coordinate system, thereby obtaining the second matching fingerprint image in the matching coordinate system. Obtain the first matching feature map of the second matching fingerprint image.

3. The training method according to claim 1, characterized in that, The step of obtaining a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system using the first initial fingerprint image, the second initial fingerprint image, and the transformation matrix includes: Obtain the first initial feature map of the first initial fingerprint image and the second initial feature map of the second initial fingerprint image; The coordinate system of one of the first initial fingerprint images and the second initial fingerprint image is used as the matching coordinate system. The initial fingerprint image is the first matching fingerprint image under the matching coordinate system, and the initial feature map corresponding to the first matching fingerprint image is the first matching feature map under the matching coordinate system. The transformation matrix is ​​used to transform another fingerprint image from the first initial fingerprint image and the second initial fingerprint image, as well as the initial feature map corresponding to the other fingerprint image, into the matching coordinate system to obtain the second matching fingerprint image and the second matching feature map in the matching coordinate system.

4. The training method according to claim 1, characterized in that, The step of obtaining the fingerprint feature matching map based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map includes: The fingerprint feature matching map is obtained by overlaying the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map.

5. A fingerprint recognition method, characterized in that, The method includes: Obtain fingerprint feature matching maps of two fingerprint images to be identified; The fingerprint feature matching map is input into the fingerprint recognition network obtained by the training method according to any one of claims 1 to 4 to obtain the recognition result; Based on the recognition results, it is determined whether the two fingerprint images to be recognized belong to the same finger.

6. The fingerprint recognition method according to claim 5, characterized in that, The step of obtaining fingerprint feature matching maps of two fingerprint images to be identified includes: Obtain feature points from the two fingerprint images to be identified; The feature points in the two fingerprint images to be identified are matched to determine the matching feature point pairs; The transformation matrix is ​​determined based on the matched feature point pairs.

7. The fingerprint recognition method according to claim 6, characterized in that, The step of obtaining fingerprint feature matching maps of two fingerprint images to be identified includes: By acquiring two fingerprint images to be identified and the transformation matrix, a first matching fingerprint image, a second matching fingerprint image, a first matching feature map, a second matching feature map, and a matching encoding map in the matching coordinate system are obtained. The fingerprint feature matching map is obtained based on the first matching fingerprint image, the second matching fingerprint image, the first matching feature map, the second matching feature map, and the matching encoding map.

8. A computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the training method of the fingerprint recognition network as described in any one of claims 1 to 4, or to implement the fingerprint recognition method as described in any one of claims 5 to 7.

9. A training apparatus for a fingerprint recognition network, comprising a memory and a processor, wherein the memory stores a program, and the program, when read and executed by the processor, implements the training method for a fingerprint recognition network as described in any one of claims 1 to 4.

10. A fingerprint recognition device, comprising a memory and a processor, wherein the memory stores a program that, when read and executed by the processor, implements the fingerprint recognition method as described in any one of claims 5 to 7.

Images