A fingerprint search method and device

Abstract

The application provides a fingerprint retrieval method and device, the method comprising: constructing a contrast learning network; reading a plurality of first pictures, and performing picture reconstruction on each first picture to form a second picture respectively, wherein the first pictures contain first fingerprint information; inputting the plurality of first pictures and the second pictures into the contrast learning network to obtain first output data and second output data corresponding to each pair of first picture and second picture respectively; calculating the similarity of each corresponding first output data and second output data; updating the weight information of a first network in the contrast learning network based on the similarity; updating the weight information of a second network in the contrast learning network based on the updated weight information of the first network, thereby obtaining a fingerprint identification network storing first fingerprint information on all the first pictures; inputting second fingerprint information into the fingerprint identification network; and completing the retrieval of the second fingerprint information based on at least the fingerprint identification network.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a fingerprint retrieval method and apparatus. Background Technology

[0002] Biometrics refers to the use of unique biological and physiological characteristics of the human body to identify individuals. Unlike some external features, biometrics are difficult to forge, thus offering greater convenience and security, making them a rising star in the fields of identity authentication and security. Available biometric technologies include fingerprints, facial recognition, voiceprints, and iris scans. Among these, fingerprint recognition is the most widely used biometric technology. It possesses characteristics such as universality, uniqueness, persistence, ease of collection, and difficulty in deception, and is therefore widely applied in access control, e-commerce, and public security investigations. Fingerprint recognition methods mainly consist of two processes: feature extraction and comparison.

[0003] The most common feature extraction method internationally is minutiae feature extraction. Its specific steps include establishing a coordinate system for the fingerprint image, typically selecting the top left corner as the farthest point, and defining a square with the x-axis pointing right and the y-axis pointing downwards. Several minutiae are then selected from this square, and their pixel coordinates and ridge directions are used as feature data. The ridge direction is defined in 1° increments, with the positive x-axis at 0° (right), the negative y-axis at 90° (up), the negative x-axis at 180° (left), and the y-axis square at 270° (down). Fingerprint comparison involves quantitatively evaluating two fingerprint images using similarity metrics to assess the matching relationship between minutiae in both images. However, in practice, temporary changes in the sampling finger, such as peeling, air humidity, foreign objects on the skin, and the pressure applied during sampling, can cause blurred or distorted fingerprint ridges, making feature extraction impossible or resulting in false features. Therefore, designing efficient and accurate search algorithms is a key issue in large-scale fingerprint image retrieval. Summary of the Invention

[0004] This invention provides a fingerprint retrieval method and apparatus that can perform fingerprint identification and retrieval quickly and accurately.

[0005] To address the aforementioned technical problems, embodiments of the present invention provide a fingerprint retrieval method, comprising:

[0006] Construct a contrastive learning network;

[0007] Read multiple first images and reconstruct each first image to form a second image, wherein the first image contains first fingerprint information;

[0008] Multiple first images and second images are input into the contrast learning network to obtain first output data and second output data for each pair of first images and second images, respectively.

[0009] Calculate the similarity between each corresponding first output data and second output data;

[0010] The weight information of the first network in the compared learning network is updated based on the similarity.

[0011] The weight information of the second network in the contrastive learning network is updated based on the updated weight information of the first network, thereby obtaining a fingerprint recognition network that stores the first fingerprint information of all the first images.

[0012] Input the second fingerprint information into the fingerprint recognition network;

[0013] The retrieval of the second fingerprint information is completed at least based on the fingerprint recognition network.

[0014] As an optional embodiment, the construction of the contrastive learning network includes:

[0015] The contrastive learning network is formed by constructing a self-supervised learning network weighted by moving average.

[0016] As an optional embodiment, the step of reading multiple first images and reconstructing each first image to form a second image includes:

[0017] Each of the first images is divided into four equal parts by a cross-cutting method, and the positions of each part of the first image are swapped along the diagonal direction to form the second image.

[0018] As an optional embodiment, updating the weight information of the second network in the contrastive learning network based on the updated weight information of the first network includes:

[0019] The weight information of the second network is obtained by weighting and averaging the weight information of the first network based on multiple previous updates.

[0020] As an optional embodiment, the step of retrieving the second fingerprint information based at least on the fingerprint recognition network includes:

[0021] The second fingerprint information is used to generate a corresponding tag code based on the fingerprint recognition network;

[0022] The tag encoding is compared with the stored first fingerprint information;

[0023] Based on the comparison results, the candidate first fingerprint information with the highest similarity is determined.

[0024] As an optional embodiment, it also includes:

[0025] Minute feature extraction is performed on the candidate first fingerprint information to determine the first minute of the candidate first fingerprint information;

[0026] Minute feature extraction is performed on the second fingerprint information to determine the second minute feature of the second fingerprint information;

[0027] Arrange directional line segments along the directions of each of the first detail points and the second detail points, and extend them to the edge of each image.

[0028] The first region is determined based on the direction line segment corresponding to the first detail point;

[0029] The enclosed second region is determined based on the direction line segment corresponding to the second detail point;

[0030] The first region and the second region are compared, and based on the comparison result, it is determined whether the candidate first fingerprint information is the target first fingerprint information, wherein the target first fingerprint information and the second fingerprint information are the same fingerprint information.

[0031] As an optional embodiment, both the first region and the second region are multiple;

[0032] The step of comparing the first region and the second region, and determining whether the candidate first fingerprint information is the target first fingerprint information based on the comparison result, includes:

[0033] Compare the number of regions in the first region and the second region whose similarity meets the similarity threshold;

[0034] Based on the number of regions and the number threshold, it is determined whether the candidate first fingerprint information is the target first fingerprint information.

[0035] Another embodiment of the present invention also provides a fingerprint retrieval device, comprising:

[0036] Modules for building contrastive learning networks;

[0037] The reconstruction module is used to read multiple first images and reconstruct each first image to form a second image, wherein the first image contains first fingerprint information;

[0038] The first input module is used to input multiple first images and second images into the contrast learning network to obtain first output data and second output data corresponding to each pair of first images and second images.

[0039] The calculation module is used to calculate the similarity between each corresponding first output data and second output data;

[0040] The first update module is used to update the weight information of the first network in the compared learning network according to the similarity.

[0041] The second update module is used to update the weight information of the second network in the contrast learning network according to the updated weight information of the first network, thereby obtaining a fingerprint recognition network that stores the first fingerprint information on all the first images.

[0042] The second input module is used to input second fingerprint information into the fingerprint recognition network;

[0043] The retrieval module is used to retrieve the second fingerprint information based at least on the fingerprint recognition network.

[0044] As an optional embodiment, the construction of the contrastive learning network includes:

[0045] The contrastive learning network is formed by constructing a self-supervised learning network weighted by moving average.

[0046] As an optional embodiment, the step of reading multiple first images and reconstructing each first image to form a second image includes:

[0047] Each of the first images is divided into four equal parts by a cross-cutting method, and the positions of each part of the first image are swapped along the diagonal direction to form the second image.

[0048] Based on the disclosure of the above embodiments, it can be understood that the beneficial effects of the embodiments of the present invention include: constructing a contrastive learning network and reconstructing a first image to form a large amount of training data; training the contrastive learning network based on the training data to form a fingerprint recognition network that stores a large number of fingerprints and can be used to retrieve fingerprints; and then fingerprint retrieval can be completed by inputting the fingerprint to be retrieved into the fingerprint recognition network. The entire process is fast, efficient, and highly accurate; even if the fingerprint is damaged or lacks clarity, high-precision retrieval and matching can still be achieved.

[0049] Other features and advantages of the invention 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. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0050] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0051] The accompanying drawings are provided to further illustrate the present application and form part of the specification. They are used together with the embodiments of the present application to explain the application and do not constitute a limitation thereof. In the drawings:

[0052] Figure 1 This is a flowchart of the fingerprint retrieval method in an embodiment of the present invention.

[0053] Figure 2 This is a partial application flowchart of the fingerprint retrieval method in an embodiment of the present invention.

[0054] Figure 3 This is a structural comparison diagram of the first image and the second image in an embodiment of the present invention.

[0055] Figure 4 This is a region comparison diagram of the fingerprint retrieval method in an embodiment of the present invention.

[0056] Figure 5 This is a structural block diagram of the fingerprint retrieval device in an embodiment of the present invention. Detailed Implementation

[0057] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but these are not intended to limit the scope of the invention.

[0058] It should be understood that various modifications can be made to the embodiments disclosed herein. Therefore, the following description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of this disclosure will be apparent to those skilled in the art.

[0059] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present disclosure and, together with the general description of the disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the disclosure.

[0060] These and other features of the invention will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0061] It should also be understood that although the invention has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of the invention, which have the features described in the claims and are therefore all within the scope of protection defined herein.

[0062] The above and other aspects, features and advantages of this disclosure will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0063] Specific embodiments of the present disclosure are described thereafter with reference to the accompanying drawings; however, it should be understood that the disclosed embodiments are merely examples of the present disclosure and can be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the present disclosure. Therefore, the specific structural and functional details disclosed herein are not intended to be limiting, but merely to serve as the basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in a variety of substantially any suitable detailed structures.

[0064] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in still another embodiment,” all of which may refer to one or more of the same or different embodiments according to this disclosure.

[0065] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0066] like Figure 1 As shown, an embodiment of the present invention provides a fingerprint retrieval method, including:

[0067] Construct a contrastive learning network;

[0068] Read multiple first images and reconstruct each first image to form a second image, wherein the first image contains first fingerprint information;

[0069] Multiple first and second images are input into the contrastive learning network to obtain the first and second output data for each pair of first and second images, respectively.

[0070] Calculate the similarity between each corresponding first output data and second output data;

[0071] The weight information of the first network in the compared learning network is updated based on similarity;

[0072] The weight information of the second network in the comparative learning network is updated based on the updated weight information of the first network, thereby obtaining a fingerprint recognition network that stores the first fingerprint information of all the first images.

[0073] Input the second fingerprint information into the fingerprint recognition network;

[0074] The retrieval of the second fingerprint information is completed at least based on the fingerprint recognition network.

[0075] Based on the disclosure of the above embodiments, it can be understood that the beneficial effects of this embodiment include: constructing a contrastive learning network and reconstructing the first image to form a large amount of training data; training the contrastive learning network based on the training data to form a fingerprint recognition network that stores a large number of fingerprints and can be used to retrieve fingerprints; and then fingerprint retrieval can be completed by inputting the fingerprint to be retrieved into the fingerprint recognition network. The entire process is fast, efficient, and highly accurate; even if the fingerprint is damaged or lacks clarity, high-precision retrieval and matching can still be achieved.

[0076] Specifically, the construction of a contrastive learning network includes:

[0077] A self-supervised learning network weighted by moving averages is constructed to form a contrastive learning network.

[0078] For example, such as Figure 2 As shown, this embodiment constructs a MaWe moving average weighted self-supervised learning network, whose backbone network uses the ResNet50 model as the decoder, and the main training network is used to identify fingerprint features.

[0079] Furthermore, when reading multiple first images and reconstructing each first image to form a second image, the process includes:

[0080] Each first image is divided into four equal parts by a cross-cutting method, and the positions of each first image are swapped along the diagonal to form a second image.

[0081] Combination Figure 3 As shown, assuming an image x (the first image) is read, the image is processed as follows... Figure 3 The image is divided into four equal parts as shown, and then the positions of each part are swapped along the diagonal to form the second image. Figure 3 The operations described are for further random data augmentation of the two images, thereby improving the fingerprint recognition accuracy of subsequent models.

[0082] Furthermore, when updating the weight information of the second network in the contrastive learning network based on the updated weight information of the first network, the process includes:

[0083] The weight information of the second network is obtained by weighting and averaging the weight information of the first network based on multiple previous updates.

[0084] Specifically, the arrangement of the aforementioned multiple first and second images constitutes the training data. During training, each of these images is fed into a contrastive learning network with ResNet50 as the backbone, thereby obtaining the first and second output data for each pair of first and second images, denoted as z1 and z2 respectively. Then, the similarity between the two is calculated. The gradient is then backpropagated to the network based on the calculated similarity result, updating the weight information w1 of the first network in the contrastive learning network, but not directly updating the weight information w2 of the second network. When calculating the weights of the second network, the weight w2 of the second network is taken as the weighted average of the weights w1 of the first network over the past four batches (it can also be more or fewer batches, the specifics are not unique), that is, the weight information of the most recent four batches, expressed as:

[0085] W2=0.1*w14+0.2*w13+0.3*w12+0.4*w11

[0086] W11 is the weight value of the first network after the last batch update. It needs to be given a higher weight so that the second comparison network pays more attention to the weight of the first network that was recently updated. The multiple weighted averages make the second network update more smoothly and will not cause the collapse phenomenon caused by the second network updating synchronously with the first network.

[0087] In this embodiment, the contrastive learning network converges after 200 epochs of training, thus forming the fingerprint recognition network. This fingerprint recognition network stores information from all first fingerprints, forming a fingerprint database for subsequent fingerprint retrieval.

[0088] Furthermore, the retrieval of the second fingerprint information is completed at least based on the fingerprint recognition network, including:

[0089] The second fingerprint information is used to generate a corresponding tag code based on the fingerprint recognition network;

[0090] The tag encoding is compared with the stored first fingerprint information;

[0091] Based on the comparison results, the candidate first fingerprint information with the highest similarity is determined.

[0092] For example, in the fingerprint retrieval stage, the fingerprint to be queried (the second fingerprint) is input into the network model. After the second fingerprint is processed by the moving average weighted unsupervised learning framework MaWe to generate the corresponding label encoding, it is compared with the image (the first image) in its database in the network, thereby outputting the candidate first fingerprint with the highest similarity (which can be output in image form). The output of the entire model structure in this embodiment can also be a multi-class encoding, where the classification encoding is obtained by manually performing fingerprint matching, classification, and encoding on existing fingerprint datasets. During the training stage, the fingerprint image and the corresponding fingerprint classification and classification encoding can be combined for training, thereby enabling the model to directly output a multi-class encoding that matches the second fingerprint.

[0093] In this embodiment, the fingerprint recognition network can actually store a similarity matrix and perform retrieval based on the matrix to select an image (the first image) that is similar to the query image (the image containing the second fingerprint). Then, in order to further improve the retrieval accuracy, this embodiment also needs to enter the next stage of precise matching, and finally determine whether the fingerprint belongs to the database, that is, whether the candidate first fingerprint found is the same as the second fingerprint.

[0094] Specifically, based on the above, the method in this embodiment further includes:

[0095] Minute feature extraction is performed on the candidate first fingerprint information to determine the first minute of the candidate first fingerprint information;

[0096] Minute feature extraction is performed on the second fingerprint information to determine the second minute feature of the second fingerprint information;

[0097] Configure directional line segments along the directions of each first detail point and second detail point, and extend them to the edge of the image where they are located;

[0098] The first region is determined based on the direction line segment corresponding to the first detail point;

[0099] The enclosed second region is determined based on the direction line segment corresponding to the second detail point;

[0100] The first region and the second region are compared, and based on the comparison result, it is determined whether the candidate first fingerprint information is the target first fingerprint information. The target first fingerprint information and the second fingerprint information are the same fingerprint information.

[0101] Both the first and second regions are multiple;

[0102] Compare the first region and the second region, and determine whether the candidate first fingerprint information is the target first fingerprint information based on the comparison result, including:

[0103] Compare the number of regions in the first region and the second region whose similarity meets the similarity threshold;

[0104] The candidate first fingerprint information is determined as the target first fingerprint information based on the number of regions and the number threshold.

[0105] For example, extending the directional line segments in each detail point to the edge of the image will cause these lines to intersect, thus forming multiple different regions, namely multiple first regions and second regions. Then, by selecting the enclosed region in the query fingerprint and comparing it with the enclosed regions in the retrieved similar fingerprints, if the number of regions with the same shape and structure meets a threshold, it indicates that the two fingerprints are the same; otherwise, they are different fingerprints. Preferably, in this embodiment, the enclosed region is a triangle, but it can also be other shapes. When the number of identical triangular regions reaches 5 pairs, the query and the queried fingerprints are considered to be the same.

[0106] Specifically, in combination Figure 4 As shown, (a) is a visualization of the minutiae of the query fingerprint (second fingerprint), where the lines are extensions of the three minutiae, and the three line segments intersect to form a triangular region. (b) is the query image (candidate first fingerprint), similar to (a), where the extensions of the three minutiae also form a triangular region. By comparing the triangular regions in the two images, if five pairs of identical triangles are found, the two fingerprints are considered to be identical.

[0107] like Figure 5 As shown, another embodiment of the present invention also provides a fingerprint retrieval device, including:

[0108] Modules for building contrastive learning networks;

[0109] The reconstruction module is used to read multiple first images and reconstruct each first image to form a second image, wherein the first image contains first fingerprint information;

[0110] The first input module is used to input multiple first images and second images into the contrast learning network to obtain first output data and second output data corresponding to each pair of first images and second images.

[0111] The calculation module is used to calculate the similarity between each corresponding first output data and second output data;

[0112] The first update module is used to update the weight information of the first network in the compared learning network according to the similarity.

[0113] The second update module is used to update the weight information of the second network in the contrast learning network according to the updated weight information of the first network, thereby obtaining a fingerprint recognition network that stores the first fingerprint information on all the first images.

[0114] The second input module is used to input second fingerprint information into the fingerprint recognition network;

[0115] The retrieval module is used to retrieve the second fingerprint information based at least on the fingerprint recognition network.

[0116] As an optional embodiment, the construction of the contrastive learning network includes:

[0117] The contrastive learning network is formed by constructing a self-supervised learning network weighted by moving average.

[0118] As an optional embodiment, the step of reading multiple first images and reconstructing each first image to form a second image includes:

[0119] Each of the first images is divided into four equal parts by a cross-cutting method, and the positions of each part of the first image are swapped along the diagonal direction to form the second image.

[0120] As an optional embodiment, updating the weight information of the second network in the contrastive learning network based on the updated weight information of the first network includes:

[0121] The weight information of the second network is obtained by weighting and averaging the weight information of the first network based on multiple previous updates.

[0122] As an optional embodiment, the step of retrieving the second fingerprint information based at least on the fingerprint recognition network includes:

[0123] The second fingerprint information is used to generate a corresponding tag code based on the fingerprint recognition network;

[0124] The tag encoding is compared with the stored first fingerprint information;

[0125] Based on the comparison results, the candidate first fingerprint information with the highest similarity is determined.

[0126] As an optional embodiment, it also includes:

[0127] Minute feature extraction is performed on the candidate first fingerprint information to determine the first minute of the candidate first fingerprint information;

[0128] Minute feature extraction is performed on the second fingerprint information to determine the second minute feature of the second fingerprint information;

[0129] Arrange directional line segments along the directions of each of the first detail points and the second detail points, and extend them to the edge of each image.

[0130] The first region is determined based on the direction line segment corresponding to the first detail point;

[0131] The enclosed second region is determined based on the direction line segment corresponding to the second detail point;

[0132] The first region and the second region are compared, and based on the comparison result, it is determined whether the candidate first fingerprint information is the target first fingerprint information, wherein the target first fingerprint information and the second fingerprint information are the same fingerprint information.

[0133] As an optional embodiment, both the first region and the second region are multiple;

[0134] The step of comparing the first region and the second region, and determining whether the candidate first fingerprint information is the target first fingerprint information based on the comparison result, includes:

[0135] Compare the number of regions in the first region and the second region whose similarity meets the similarity threshold;

[0136] Based on the number of regions and the number threshold, it is determined whether the candidate first fingerprint information is the target first fingerprint information.

[0137] Another embodiment of the present invention also provides an electronic device, comprising:

[0138] One or more processors;

[0139] Memory, configured to store one or more programs;

[0140] When the one or more programs are executed by the one or more processors, the one or more processors shall implement the methods described above.

[0141] An embodiment of the present invention also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the method described above. It should be understood that the various solutions in this embodiment have the corresponding technical effects in the above method embodiments, and will not be repeated here.

[0142] This invention also provides a computer program product tangibly stored on a computer-readable medium and comprising computer-readable instructions that, when executed, cause at least one processor to perform methods such as those described in the embodiments above.

[0143] It should be understood that the various solutions in this embodiment have the same technical effects as those in the above method embodiments, and will not be repeated here.

[0144] It should be noted that the computer storage medium of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access storage media (RAM), read-only storage media (ROM), erasable programmable read-only storage media (EPROM or flash memory), optical fibers, portable compact disk read-only storage media (CD-ROM), optical storage media, magnetic storage media, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program configured for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, antenna, optical fiber, RF, etc., or any suitable combination thereof.

[0145] It should be understood that although this application is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0146] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the present invention. The scope of protection of the present invention is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to the present invention within its spirit and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of the present invention.

Claims

1. A fingerprint retrieval method, characterized in that, include: Constructing a contrastive learning network includes constructing a moving average weighted self-supervised learning network to form the contrastive learning network; Read multiple first images and reconstruct each first image to form a second image, wherein the first image contains first fingerprint information; Multiple first images and second images are input into the contrast learning network to obtain first output data and second output data for each pair of first images and second images, respectively. Calculate the similarity between each corresponding first output data and second output data; The weight information of the first network in the compared learning network is updated based on the similarity. The weight information of the second network in the contrastive learning network is updated based on the updated weight information of the first network, thereby obtaining a fingerprint recognition network that stores the first fingerprint information of all the first images. Input the second fingerprint information into the fingerprint recognition network; The retrieval of the second fingerprint information is completed at least based on the fingerprint recognition network; The step of reading multiple first images and reconstructing each first image to form a second image includes: Each of the first images is divided into four equal parts by a cross-cutting method, and the positions of each part of the first image are swapped along the diagonal direction to form the second image. The construction of the contrastive learning network includes: The contrastive learning network is formed by constructing a self-supervised learning network weighted by moving average.

2. The method according to claim 1, characterized in that, Updating the weight information of the second network in the contrastive learning network based on the updated weight information of the first network includes: The weight information of the second network is obtained by weighting and averaging the weight information of the first network based on multiple previous updates.

3. The method according to claim 1, characterized in that, The retrieval of the second fingerprint information based at least on the fingerprint recognition network includes: The second fingerprint information is used to generate a corresponding tag code based on the fingerprint recognition network; The tag encoding is compared with the stored first fingerprint information; Based on the comparison results, the candidate first fingerprint information with the highest similarity is determined.

4. The method according to claim 3, characterized in that, Also includes: Minute feature extraction is performed on the candidate first fingerprint information to determine the first minute of the candidate first fingerprint information; Minute feature extraction is performed on the second fingerprint information to determine the second minute feature of the second fingerprint information; Arrange directional line segments along the directions of each of the first detail points and the second detail points, and extend them to the edge of each image. The first region is determined based on the direction line segment corresponding to the first detail point; The enclosed second region is determined based on the direction line segment corresponding to the second detail point; The first region and the second region are compared, and based on the comparison result, it is determined whether the candidate first fingerprint information is the target first fingerprint information, wherein the target first fingerprint information and the second fingerprint information are the same fingerprint information.

5. The method according to claim 4, characterized in that, Both the first region and the second region are multiple; The step of comparing the first region and the second region, and determining whether the candidate first fingerprint information is the target first fingerprint information based on the comparison result, includes: Compare the number of regions in the first region and the second region whose similarity meets the similarity threshold; Based on the number of regions and the number threshold, it is determined whether the candidate first fingerprint information is the target first fingerprint information.

6. A fingerprint retrieval device, characterized in that, include: Modules for building contrastive learning networks; The reconstruction module is used to read multiple first images and reconstruct each first image to form a second image, wherein the first image contains first fingerprint information; The first input module is used to input multiple first images and second images into the contrast learning network to obtain first output data and second output data corresponding to each pair of first images and second images. The calculation module is used to calculate the similarity between each corresponding first output data and second output data; The first update module is used to update the weight information of the first network in the compared learning network according to the similarity. The second update module is used to update the weight information of the second network in the contrast learning network according to the updated weight information of the first network, thereby obtaining a fingerprint recognition network that stores the first fingerprint information on all the first images. The second input module is used to input second fingerprint information into the fingerprint recognition network; The retrieval module is used to retrieve the second fingerprint information based at least on the fingerprint recognition network; The step of reading multiple first images and reconstructing each first image to form a second image includes: Each of the first images is divided into four equal parts by a cross-cutting method, and the positions of each part of the first image are swapped along the diagonal direction to form the second image. The construction of the contrastive learning network includes: The contrastive learning network is formed by constructing a self-supervised learning network weighted by moving average.

Images