Fingerprint detection method and device, program product and storage medium

Abstract

The embodiment of the invention provides a fingerprint detection method and device, a program product and a storage medium, and the method comprises the steps: obtaining the native domain information of a to-be-detected fingerprint, the native domain information being original data collected when the to-be-detected fingerprint is subjected to fingerprint verification; performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the to-be-detected fingerprint; performing classification operation on the fingerprint feature matrix through a trained medium classification network to obtain a fingerprint medium classification vector of the to-be-detected fingerprint, the fingerprint medium classification vector being used for representing the probability that the to-be-detected fingerprint belongs to a plurality of medium types in a vector form; and determining the medium type of the to-be-detected fingerprint based on the fingerprint medium classification vector. Through the method and the device, the problem that the type of the fingerprint medium cannot be effectively detected in related technologies is solved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of fingerprint detection, and more specifically, to a fingerprint detection method and device, a program product, and a storage medium. Background Art

[0002] In the current existing technology, the fingerprint pattern is mostly identified by explicit calculation and manual fitting of real-time physical characteristics such as acoustic impedance and surface distance through ultrasonic echo signals to achieve the identification of true and false fingerprints. However, this method can only identify the authenticity of fingerprints, but cannot detect the type of medium where the fingerprint is located, which is not conducive to further analysis of fingerprint identification.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present application provide a fingerprint detection method and device, a program product, and a storage medium to at least solve the problem in the related art that the type of fingerprint medium cannot be effectively detected.

[0005] According to an embodiment of the present application, a fingerprint detection method is provided, comprising: obtaining native domain information of a fingerprint to be detected, wherein the native domain information is original data collected when fingerprint verification is performed on the fingerprint to be detected; performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected; performing classification operations on the fingerprint feature matrix through a trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector; and determining the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0006] In an exemplary embodiment, before performing a classification operation on the fingerprint feature matrix through a trained medium classification network to obtain the fingerprint medium classification vector of the fingerprint to be detected, the method also includes: respectively collecting native domain information of simulated fingerprints simulated on multiple media to obtain multiple sets of simulated fingerprint native domain information, wherein the multiple sets of simulated fingerprint native domain information include: native domain information of multiple simulated fingerprints of the same medium obtained after multiple information collections on the same medium simulated fingerprint; performing fingerprint feature extraction on each of the simulated fingerprint native domain information included in the multiple sets of simulated fingerprint native domain information to obtain multiple sets of simulated fingerprint feature matrices; using the simulated fingerprint feature matrix set and the classification labels of the multiple media, training the medium classification network to obtain the medium classification network, wherein the classification labels are used to represent the media types of the multiple media.

[0007] In an exemplary embodiment, the simulated fingerprint feature matrix set and the classification labels of the multiple media are used to train the pre-trained media classification network to obtain the trained media classification network, including: inputting the simulated fingerprint feature matrix set into the media classification network to perform a training operation, wherein the training operation includes: inputting the simulated fingerprint feature matrix into a feature extraction layer included in the media classification network, outputting a feature vector of the simulated fingerprint feature matrix through the feature extraction layer, inputting the feature vector of the simulated fingerprint feature matrix into a classification layer included in the media classification network, calculating the probability that the feature vector of the simulated fingerprint feature matrix belongs to the classification labels of multiple media through the classification layer, and outputting a media classification vector; determining a loss value output by a loss function of the media classification network based on the media classification vector and the classification label of the media; terminating the training to obtain the trained media classification network when the loss value meets a preset training end condition; and adjusting the value of a parameter of the media classification network to reduce the loss value output by the loss function of the media classification network in the next training when the loss value does not meet the preset training end condition.

[0008] In an exemplary embodiment, after the simulated fingerprint feature matrix set is input into the medium classification network to perform a training operation, the method further includes: extracting the feature vector of each simulated fingerprint feature matrix included in the simulated fingerprint feature matrix set output by the feature extraction layer; mapping the feature vector of each simulated fingerprint feature matrix to a coordinate space of a preset dimension, and performing dimensionality reduction processing on the feature vector of each simulated fingerprint feature matrix in the coordinate space to obtain a feature vector of a target dimension; and visualizing the feature vector of the target dimension of each simulated fingerprint feature matrix to display it on a target interface.

[0009] In an exemplary embodiment, fingerprint feature extraction is performed on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected, including: constructing a native domain information set, wherein the native domain information set includes multiple native domain information of the fingerprint to be detected obtained in a current time period, or the native domain information set includes multiple historical native domain information obtained in a historical time period and the native domain information obtained in the current time period; converting the native domain information included in the native domain information set into digital signals through a signal converter to obtain multiple groups of digital signals; extracting digital features from each group of digital signals to obtain fingerprint features of each group of native domain information; generating a matrix including fingerprint features of each group of native domain information according to the signal acquisition channel of the target sensor that obtains the native domain information included in the native domain information set to obtain multiple groups of fingerprint feature matrices; combining the multiple groups of fingerprint feature matrices to obtain the fingerprint feature matrix to be detected.

[0010] In an exemplary embodiment, a classification operation is performed on the fingerprint feature matrix through a trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, including: inputting the fingerprint feature matrix to be detected into a feature extraction layer included in the medium classification network, and extracting feature vectors of each group of fingerprint feature matrices included in the fingerprint feature matrix to be detected through the feature extraction layer; inputting the feature vectors of each group of fingerprint feature matrices outputted from the feature extraction layer into a classification layer included in the medium classification network, and calculating the probability that the feature vectors of each group of fingerprint feature matrices belong to classification labels of multiple media through the classification layer; and combining multiple classification vectors outputted from the classification layer to obtain the fingerprint medium classification vector.

[0011] According to another embodiment of the present application, a fingerprint detection device is provided, including: a first acquisition module, used to obtain native domain information of a fingerprint to be detected, wherein the native domain information is original data collected when fingerprint verification is performed on the fingerprint to be detected; a first extraction module, used to perform fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected; a first classification module, used to perform a classification operation on the fingerprint feature matrix through a trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector; a first determination module, used to determine the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0012] According to another embodiment of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0013] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when run.

[0014] According to another embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0015] In an embodiment of the present application, the native domain information of the fingerprint to be detected is obtained, and fingerprint feature extraction is performed on the native domain information to obtain a fingerprint feature matrix, and a classification operation is performed on the fingerprint feature matrix through a trained medium classification network to obtain a medium classification vector, and finally the medium type of the fingerprint to be detected is analyzed based on the medium classification vector. By using the trained medium classification network to perform classification operations on the fingerprint feature matrix, a more accurate and high-quality medium classification vector can be generated, thereby improving the accuracy and efficiency of identifying the medium type of the fingerprint to be detected, and effectively detecting the type of the fingerprint medium. The technical effect of accurately and quickly identifying the type of the fingerprint medium is achieved, thereby solving the problem that the type of the fingerprint medium cannot be effectively detected in the related art. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a schematic diagram of an application environment of a fingerprint detection method according to an embodiment of the present application;

[0017] Figure 2 is a flow chart of a fingerprint detection method according to an embodiment of the present application;

[0018] Figure 3 is a schematic diagram of the structure of a medium classification network according to an embodiment of the present application;

[0019] Figure 4 is a schematic diagram of a specific structure of a medium classification network according to an embodiment of the present application;

[0020] Figure 5 is a visual scatter plot according to an embodiment of the present application;

[0021] Figure 6 is a flowchart of constructing a CNN model according to an embodiment of the present application;

[0022] Figure 7 is a structural block diagram of a fingerprint detection device according to an embodiment of the present application;

[0023] Figure 8is a schematic structural diagram of a product for determining the type of fingerprint medium according to an embodiment of the present application;

[0024] Fig. 9 It is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0027] According to one aspect of an embodiment of the present application, a fingerprint detection method is provided. Optionally, in this embodiment, the fingerprint detection method can be applied to Figure 1 In the hardware environment composed of the server 101 and the terminal device 103 shown in FIG. Figure 1 As shown, the server 101 is connected to the terminal 103 via a network, and can be used to provide services for the terminal device or the application installed on the terminal device. The application can be a video application, an instant messaging application, a browser application, an educational application, a game application, etc. A database 105 may be set up on the server or independently of the server to provide data storage services for the server 101, for example, a game data storage server. The above-mentioned network may include, but is not limited to, a wired network and a wireless network, wherein the wired network includes, a local area network, a metropolitan area network and a wide area network; the wireless network includes, Bluetooth, WIFI and other networks that implement wireless communication; the terminal device 103 may be a terminal configured with an application, and may include, but is not limited to, at least one of the following: a mobile phone (such as an Android phone, an iOS phone, etc.), a laptop computer, a tablet computer, a PDA, a MID (Mobile Internet Devices), a PAD, a desktop computer, a smart TV, an intelligent voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a mixed reality (MR) terminal and other computer devices; the above-mentioned server may be a single server, a server cluster consisting of multiple servers, or a cloud server.

[0028] Combination Figure 1 As shown, the above fingerprint detection method can be implemented in the terminal device 103 through the following steps:

[0029] S1, obtaining native domain information of the fingerprint to be detected, wherein the native domain information is the original data collected when the fingerprint to be detected is verified;

[0030] S2, performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected;

[0031] S3, performing classification operation on the fingerprint feature matrix through the trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector;

[0032] S4, determining the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0033] Optionally, in this embodiment, the fingerprint detection method can also be implemented by a server, for example, Figure 1 It is implemented in the server 101 shown; or it is implemented by the terminal device and the server together.

[0034] It should be noted that the above-mentioned server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected by wired or wireless communication, and this application does not limit this.

[0035] The above is only an example and is not specifically limited in this embodiment.

[0036] In this embodiment, a fingerprint detection method is provided. Figure 2 is a flow chart of a fingerprint detection method according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:

[0037] Step S202, obtaining native domain information of the fingerprint to be detected, wherein the native domain information is the original data collected when the fingerprint to be detected is verified;

[0038] Optionally, the fingerprint to be detected in this embodiment may be a fingerprint simulated on the target medium. This embodiment may obtain the native domain information of the fingerprint to be detected simulated on the target medium by at least one of the following methods: Optical scanning method: by using an optical scanner to scan the fingerprint to be detected simulated on the target medium, the fingerprint image of the fingerprint to be detected is converted into a digital signal. Capacitive scanning method: the fingerprint information is obtained by detecting the capacitance change between the fingerprint to be detected simulated on the target medium and the sensor. Piezoelectric scanning method: the fingerprint information is obtained by detecting the pressure change applied by the target medium on the sensor. Ultrasonic scanning method: the fingerprint information of the fingerprint to be detected simulated on the target medium is obtained by emitting and receiving ultrasonic waves. Thermal imaging scanning method: the fingerprint information is obtained by detecting the temperature change on the surface of the target medium. Radio frequency scanning method: the fingerprint information of the fingerprint to be detected simulated on the target medium is obtained by emitting radio frequency signals and receiving the reflected signals. Multi-spectral scanning method: the fingerprint information of the fingerprint to be detected simulated on the target medium is obtained by using light of different wavelengths.

[0039] Optionally, the material of the target medium in this embodiment includes but is not limited to gel, polyurethane (PU), latex, silicone, acrylic, and resin. Simulating the fingerprint to be detected on the target medium means creating a replica with similar features to the original fingerprint on the target medium. For example, under the same environment, different media processes such as gel, PU, ​​latex, silicone, acrylic, and resin are used to manufacture a forged fingerprint corresponding to a real finger.

[0040] Optionally, in the process of collecting fingerprints to be detected, raw data refers to the fingerprint information obtained by the fingerprint collection device without any processing. The raw data includes but is not limited to the following information: fingerprint pattern, fingerprint direction, fingerprint size, fingerprint image quality, collection time, collection device information, and other biometric information (for example, finger thickness, temperature, etc.). For example, using an optical fingerprint collector, an image of the fingerprint is obtained by scanning the fingerprint set on silicone. These image data are native domain information. They capture features such as fingerprint texture, ridges, and sweat pores. Capacitive sensors collect fingerprints by detecting the change in capacitance between the silicone finger and the sensor surface. These capacitance change data reflect the distribution of the ridges and valleys of the fingerprint and constitute the native domain information.

[0041] Step S204, performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected;

[0042] Optionally, this embodiment can perform fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix in the following manner: perform signal conversion operation on the native domain information through a signal converter to obtain a group of digital signals of the native domain information; extract digital features from a group of digital signals to obtain a group of digital features of the native domain information; determine the digital matrix of the native domain information as a group of digital features including the native domain information.

[0043] In this embodiment, the signal converter is a device that converts different types of signals into another type of signal, for example, conversion between analog signals and digital signals, conversion of signals of different frequencies or encoding methods, etc. The signal conversion of native domain information can be performed in the following manner: Determine the type of native domain information, whether it is an analog signal, a digital signal, or other forms of information. Select a suitable signal converter according to the type of native domain information. For example, if an analog signal is converted to a digital signal, an analog-to-digital converter is required to configure the relevant parameters of the converter, such as sampling rate, resolution, filtering, etc., according to the characteristics and requirements of the analog signal. Input the native domain information into the signal converter and perform the signal conversion operation. After the conversion is completed, the signal converter will output a set of digital signals, which are digital representations of the native domain information.

[0044] For example, a capacitive sensor is used to obtain an image of the fingerprint to be detected. This image contains detailed information such as the ridges and valleys of the fingerprint. A signal converter is used to convert the fingerprint image, including mathematical methods such as Fourier transform and wavelet transform, to convert the image information into a set of digital signals. These digital signals contain the characteristic information of the fingerprint to be detected. Digital features are extracted from the converted digital signals. These features can be the ridge direction, ridge frequency, ridge curvature, etc. of the fingerprint. The extracted digital features are arranged in a certain order to form a digital matrix. This matrix contains all the key information of the fingerprint to be detected. In the above way, signal conversion, digital feature extraction and digital matrix construction of the fingerprint to be detected can be realized, providing a key information basis for the identification of the target medium type.

[0045] In this embodiment, since the fingerprint feature matrix is ​​a digital matrix data after the sensor completes the signal analog-to-electric conversion, it may not be possible to visually observe the corresponding fingerprint information without preprocessing. However, it contains all the original information that has not been filtered out, so it is more suitable for automatic feature extraction with deep learning technology.

[0046] Step S206, performing classification operation on the fingerprint feature matrix through the trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector;

[0047] Optionally, the medium classification network includes but is not limited to various types of neural networks, such as convolutional neural networks (CNN), recurrent neural networks (RNN), long short-term memory networks (LSTM), and graph neural networks (GNN). For example, when the medium classification network is a CNN network, the structure of the medium classification network includes but is not limited to the following: Figure 3 The network structure shown includes: a feature extraction layer and a classification layer. The fingerprint feature matrix is ​​input into the feature extraction layer through the CNN network, and the feature vector of the fingerprint feature matrix is ​​output through the feature extraction layer. The feature vector of the fingerprint feature matrix is ​​input into the classification layer, and the probability that the feature vector of the fingerprint feature matrix belongs to the classification labels of multiple media is calculated through the classification layer, and the fingerprint medium classification vector is output.

[0048] Optionally, multiple sub-network layers may be included in the feature extraction layer. For example, when the medium classification network is a CNN network, the feature extraction layer is composed of a Conv layer (convolution layer), a ReLU activation layer, a MaxPool maximum pooling layer, and a GlobalAvgPool layer (global pooling layer). Among them, the Conv layer is used to extract part of the image features. The ReLU activation layer is a nonlinear activation function used to introduce nonlinearity in the neural network so that the network can learn and simulate more complex functions. The MaxPool maximum pooling layer is used to reduce the spatial size of the features output by the convolution layer, thereby reducing the number of parameters and computational complexity, and also making feature detection more robust. The GlobalAvgPool global pooling layer is a special pooling operation that averages the entire feature. The number and stacking method of the feature extraction layer consisting of the Conv layer (convolution layer), the ReLU activation layer, the MaxPool maximum pooling layer, and the GlobalAvgPool layer (global pooling layer) can be adjusted. For example, the number and stacking method of the feature extraction layer consisting of Conv layer (convolution layer), ReLU activation layer, MaxPool maximum pooling layer and GlobalAvgPool layer (global pooling layer) are as follows: Figure 4As shown, it is used to extract a one-dimensional deep feature vector from a fingerprint feature matrix (for example, an ultrasonic fingerprint input image). For example, the fingerprint feature matrix is ​​input into the Conv layer, a convolution operation is performed using a 3x3 convolution kernel, and the matrix is ​​output. The ReLU activation layer performs a ReLU activation operation on the matrix output by the convolution layer, and then the maximum pooling kernel in the MaxPool maximum pooling layer is used to perform a maximum pooling operation on the output of the ReLU activation layer. Finally, the GlobalAvgPool layer is used to perform a global average pooling operation on the output of the maximum pooling layer to obtain the feature vector of the fingerprint feature matrix.

[0049] Optionally, the classification layer may include multiple sub-network layers. For example, when the medium classification network is a CNN network, Figure 4 As shown in the figure, the classification layer consists of an FC fully connected layer and a Softmax activation function layer, which is used to predict the probability scores of 7 media from the deep feature vector. Among them, the FC fully connected layer is used to linearly transform the features of the input data and then provide it to the subsequent layers for further processing. By stacking multiple fully connected layers, the neural network can learn more complex features and patterns. The Softmax activation function is a function that converts input data into a probability distribution. In multi-classification problems, the Softmax function is used in the output layer of the neural network to convert the output of the model into the probability of each category.

[0050] Optionally, the number of elements in the output fingerprint medium classification vector is related to the number of media preset in the medium classification network. For example, when FC is included in the classification layer, the FC output is a 1*N vector, and each bit corresponds to the probability of belonging to this category. The medium classification network outputs guide labels as 0-unknown, 1-gel, 2-PU, 3-latex, 4-silicone, 5-acrylic, 6-resin, where only the element value of the corresponding category is 1, and the element values ​​of other categories are all 0. For example, the output vector corresponding to the gel category is [0, 1, 0, 0, 0, 0, 0].

[0051] Step S208: determining the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0052] Optionally, the medium type of the fingerprint to be detected can be determined from the value corresponding to each element in the fingerprint medium classification vector. For example, when the fingerprint medium classification vector is [0, 1, 0, 0, 0, 0, 0], it can be determined that the medium type of the fingerprint to be detected is gel.

[0053] Optionally, the fingerprint detection method provided in this embodiment can be applied to various scenarios where fingerprint medium identification is required, including but not limited to scenarios of fingerprint medium testing and scenarios of true and false fingerprint identification.

[0054] Through the above steps, the native domain information of the fingerprint to be detected simulated on the target medium is obtained, and the fingerprint feature extraction is performed on the native domain information to obtain a fingerprint feature matrix, and the fingerprint feature matrix is ​​classified by the trained medium classification network to obtain a fingerprint medium classification vector, and finally the medium type of the fingerprint to be detected is analyzed based on the fingerprint medium classification vector. By using the trained medium classification network to perform classification operations on the fingerprint feature matrix, a more accurate and high-quality fingerprint medium classification vector can be generated, thereby improving the accuracy and efficiency of identifying the medium type, and effectively detecting the type of fingerprint medium. The technical effect of accurately and quickly identifying the type of fingerprint medium is achieved, thereby solving the problem that the type of fingerprint medium cannot be effectively detected in the related technology.

[0055] In an exemplary embodiment, before performing a classification operation on the fingerprint feature matrix through the trained medium classification network to obtain the fingerprint medium classification vector of the fingerprint to be detected, the method further includes:

[0056] Step S302, respectively collecting native domain information of simulated fingerprints simulated on multiple media to obtain multiple sets of simulated fingerprint native domain information, wherein the multiple sets of simulated fingerprint native domain information include: multiple native domain information of the same medium simulated fingerprint obtained after multiple information collections of the same medium simulated fingerprint;

[0057] Optionally, this embodiment is mainly a process of training a medium classification network. The various media include, but are not limited to, media made of materials such as gel, PU, ​​latex, silicone, acrylic, and resin. For example, under the same environment, different media processes such as gel, PU, ​​latex, silicone, acrylic, and resin are used to manufacture multiple counterfeit fingerprints corresponding to real fingers. The method of collecting the native domain information of the simulated fingerprints simulated on various media is the same as the method of collecting the native domain information of the fingerprint to be detected, which will not be repeated here.

[0058] Optionally, in order to enable the medium classification network to learn a more generalized feature representation, the training data should be as large as possible, and the amount of data for each medium should be as consistent as possible. In order to improve the advantages of the classification scheme over the previous manual feature modeling and extraction scheme, and to improve the medium classification system's tolerance to differences in fingers, temperature, and system environment, each medium collection should at least cover more fingers (e.g., no less than 20), temperature (e.g., the temperature range covers -25° to 55°), and machine range (at least 3 different sampling machines are used for sampling). If training from scratch is required, the data specifications involved in the training should be more than 100,000.

[0059] Step S304, performing fingerprint feature extraction on each of the simulated fingerprint native domain information included in the multiple simulated fingerprint native domain information sets to obtain multiple simulated fingerprint feature matrix sets;

[0060] Step S306: training a medium classification network using the simulated fingerprint feature matrix set and the classification labels of the multiple media to obtain the medium classification network, wherein the classification labels are used to represent the medium types of the multiple media.

[0061] Optionally, when collecting fingerprint native domain data corresponding to each medium, the classification label of the corresponding medium can be marked. For example, gel is marked as 1, PU is marked as 2, latex is marked as 3, silicone is marked as 4, acrylic is marked as 5, resin is marked as 6, and unknown type is marked as 0. If there are additional media to be classified, the cumulative label value can be increased.

[0062] This embodiment can collect information more comprehensively by collecting simulated fingerprints on multiple media and performing digital conversion, thereby improving the accuracy of the recognition system. Using multiple media for training helps to enhance the model's ability to generalize the characteristics of different media, so that it can still maintain a high recognition accuracy when facing unknown media. By training the medium classification network using a set of digital matrices and classification labels, automated learning and recognition of medium types are achieved, reducing manual intervention and improving work efficiency. Since multiple media are involved in the training process, the medium classification network can adapt to different media types, enhancing the applicability and flexibility of the algorithm. By collecting information multiple times on the same simulated fingerprint simulated on the same medium, the native domain information of multiple simulated fingerprints obtained can increase the richness of the data and help improve the model's ability to recognize subtle differences.

[0063] In an exemplary embodiment, in the above step S304, fingerprint feature extraction is performed on each of the simulated fingerprint native domain information included in the simulated fingerprint native domain information set to obtain a simulated fingerprint feature matrix set, including:

[0064] Step S3041, for each simulated fingerprint native domain information included in the simulated fingerprint native domain information set, the following operations are performed to obtain a simulated fingerprint feature matrix of each simulated fingerprint native domain information, wherein the simulated fingerprint feature matrix set includes the simulated fingerprint feature matrix of each simulated fingerprint native domain information: performing a signal conversion operation on the simulated fingerprint native domain information through a signal converter to obtain a group of digital signals of the simulated fingerprint native domain information; extracting digital features from the group of digital signals to obtain a group of fingerprint features of the simulated fingerprint native domain information; determining the simulated fingerprint feature matrix of the simulated fingerprint native domain information as a group of fingerprint features including the simulated fingerprint native domain information.

[0065] Optionally, the signal conversion of the simulated fingerprint native domain information can be converted in the following ways: Determine the type of the simulated fingerprint native domain information, whether it is an analog signal, a digital signal, or other forms of information. Select a suitable signal converter according to the type of the simulated fingerprint native domain information. For example, if an analog signal is converted to a digital signal, an ADC may be required. According to the characteristics and requirements of the signal, configure the relevant parameters of the converter, such as sampling rate, resolution, filtering, etc. Input the simulated fingerprint native domain information into the signal converter for signal conversion operations. After the conversion is completed, the signal converter will output a set of digital signals, which are digital representations of the simulated fingerprint native domain information.

[0066] For example, a capacitive sensor is used to obtain an image of the fingerprint to be detected. This image contains detailed information such as the ridges and valleys of the fingerprint. A signal converter is used to convert the fingerprint image, including mathematical methods such as Fourier transform and wavelet transform, to convert the image information into a set of digital signals. These digital signals contain the characteristic information of the fingerprint to be detected. Digital features are extracted from the converted digital signals. These features can be the ridge direction, ridge frequency, ridge curvature, etc. of the fingerprint. The extracted digital features are arranged in a certain order to form a digital matrix. This matrix contains all the key information of the fingerprint to be detected.

[0067] This embodiment uses a signal converter to perform signal conversion and digital feature extraction on native domain information, which can improve the efficiency of signal processing, enhance the readability of the signal, and more accurately describe the characteristics of the fingerprint, facilitating machine learning.

[0068] In an exemplary embodiment, the above step S306 uses the simulated fingerprint feature matrix set and the classification labels of multiple media to train the pre-trained media classification network to obtain the trained media classification network, including:

[0069] Step S3061, inputting the simulated fingerprint feature matrix set into the medium classification network to perform a training operation, wherein the training operation includes: inputting the simulated fingerprint feature matrix into the feature extraction layer included in the medium classification network, outputting the feature vector of the simulated fingerprint feature matrix through the feature extraction layer, inputting the feature vector of the digital matrix into the classification layer included in the medium classification network, calculating the probability that the feature vector of the simulated fingerprint feature matrix belongs to the classification labels of multiple media through the classification layer, and outputting the medium classification vector, wherein the simulated fingerprint feature matrix is ​​the fingerprint feature matrix included in the digital matrix set;

[0070] Optionally, in step S3061, the medium classification network includes but is not limited to various types of neural networks, such as CNN, RNN, LSTM, GNN. For example, when the medium classification network is a CNN network, the structure of the CNN network includes but is not limited to the following: Figure 3 The network structure shown includes a feature extraction layer and a classification layer. Each digital matrix in the digital matrix set is input into the feature extraction layer through the CNN network, and the feature vector of each digital matrix is ​​output through the feature extraction layer. The feature vector of each digital matrix is ​​input into the classification layer, and the probability that the feature vector of each digital matrix belongs to the classification labels of multiple media is calculated through the classification layer, and the medium classification vector is output.

[0071] Optionally, multiple sub-network layers may be included in the feature extraction layer. For example, when the medium classification network is a CNN network, the feature extraction layer is composed of a Conv layer (convolution layer), a ReLU activation layer, a MaxPool maximum pooling layer, and a GlobalAvgPool layer (global pooling layer). Among them, the Conv layer is used to extract part of the image features. The ReLU activation layer is a nonlinear activation function used to introduce nonlinearity in the neural network so that the network can learn and simulate more complex functions. The MaxPool maximum pooling layer is used to reduce the spatial size of the features output by the convolution layer, thereby reducing the number of parameters and computational complexity, and also making feature detection more robust. The GlobalAvgPool global pooling layer is a special pooling operation that averages the entire feature. The number and stacking method of the feature extraction layer consisting of the Conv layer (convolution layer), the ReLU activation layer, the MaxPool maximum pooling layer, and the GlobalAvgPool layer (global pooling layer) can be adjusted. For example, the number and stacking method of the feature extraction layer consisting of Conv layer (convolution layer), ReLU activation layer, MaxPool maximum pooling layer and GlobalAvgPool layer (global pooling layer) are as follows: Figure 4 As shown, it is used to extract a one-dimensional depth feature vector from a digital matrix (eg, ultrasonic fingerprint Raw domain whole image data).

[0072] Optionally, the classification layer may include multiple sub-network layers. For example, when the medium classification network is a CNN network, Figure 4As shown in the figure, the classification layer consists of an FC fully connected layer and a Softmax activation function layer, which is used to predict the probability scores of 7 media from the deep feature vector. Among them, the FC fully connected layer is used to linearly transform the features of the input data and then provide it to the subsequent layers for further processing. By stacking multiple fully connected layers, the neural network can learn more complex features and patterns. The Softmax activation function is a function that converts input data into a probability distribution. In multi-classification problems, the Softmax function is usually used in the output layer of the neural network to convert the output of the model into the probability of each category.

[0073] Step S3062, determining a loss value output by a loss function of the medium classification network based on the medium classification vector and a classification label of a target medium, wherein the target medium is a medium corresponding to the digital matrix;

[0074] Optionally, the number of elements in the output medium classification vector is related to the number of media in the medium classification network. For example, when FC is included in the classification layer, the FC output is a 1*N vector, and each bit corresponds to the probability of belonging to this category. The medium classification network output guide labels are 0-unknown, 1-gel, 2-PU, 3-latex, 4-silicone, 5-acrylic, 6-resin. For the convenience of training, it is converted into one-hot encoding. Each category corresponds to a one-hot encoding vector, in which only the element value of the corresponding category is 1, and the element values ​​of other categories are all 0. For example, the gel category corresponds to [0, 1, 0, 0, 0, 0, 0].

[0075] Step S3064, when the loss value satisfies the preset training end condition, the training is ended to obtain the medium classification network;

[0076] Optionally, before training the medium classification network, you also need to configure the training configuration of the medium classification network. For example, the optimizer is AdamW, and the initial learning rate is 1×10 -3 The learning rate scheduling strategy selects cosine annealing scheduling, dynamically adjusts the learning rate, and trains for 100 epochs. During the training process, parameters need to be continuously updated. For example, ultrasonic fingerprint images are randomly extracted from the digital matrix set and input into the medium classification network to obtain the probability distribution of each medium category. and the true distribution Y = {y i ,i∈[0,6]} calculates the cross entropy (CE) loss, uses the back propagation algorithm to update the network parameters, and gradually reduces the loss function. The loss function CE loss calculation method in step S3062 includes:

[0077] Step S3066, when the above loss value does not meet the preset training end condition, adjust the value of the parameter of the above medium classification network to reduce the loss value output by the above loss function of the above medium classification network in the next training.

[0078] Optionally, the training end condition preset in step S3066 includes: whether the maximum iteration epoch is reached, if yes, then the optimization is stopped and the network training is completed, otherwise, the above steps are repeated until the maximum iteration epoch is reached. Optionally, after the media classification network training is completed, the obtained CNN classification network can be directly applied to the fingerprint media detection task. In practical applications, the pressed frame data is processed according to Figure 4 The input is shown in the CNN classification network, and the CNN classification network outputs the prediction scores of the five media respectively. The range is (0, 1), the maximum The category corresponding to the value subscript is the medium category of the input predicted by the CNN classification network.

[0079] Optionally, in step S3066, when the loss value does not meet the preset training end condition, the following methods can be used to adjust the parameters of the medium classification network to reduce the output value of the loss function: adjust the learning rate, increase the network depth or width, adjust the optimizer, use regularization technology, use data enhancement, adjust the loss function, adjust the hyperparameters, etc. It should be noted that the adjustment of the parameters of the medium classification network needs to be tried and adjusted according to the specific task and data set. By comprehensively considering multiple methods, the output value of the loss function can be effectively reduced and the classification performance can be improved.

[0080] This embodiment can effectively extract the features of the digital matrix and improve the accuracy of classification by inputting the digital matrix set into the medium classification network for training. By inputting the feature vector of the digital matrix into the classification layer, the probability of the classification label of the medium can be calculated to provide a basis for subsequent classification decisions. Based on the medium classification vector and the classification label of the target medium, the loss value output by the loss function of the medium classification network can be determined to provide feedback information for subsequent training. When the loss value meets the preset training end condition, the training can be ended to obtain a trained medium classification network, thereby improving the training efficiency. When the loss value does not meet the preset training end condition, the loss value in the next training can be reduced by adjusting the parameter value of the medium classification network, thereby improving the flexibility and adaptability of the training. Through multiple iterative training, the performance of the medium classification network can be continuously optimized to improve the accuracy and robustness of the classification. In general, the above steps can effectively train a medium classification network with excellent performance by inputting a digital matrix set, extracting features, calculating loss values, adjusting parameters, and other operations, thereby improving the accuracy and efficiency of the classification task.

[0081] In an exemplary embodiment, after inputting the simulated fingerprint feature matrix set into the medium classification network to perform a training operation, the method further includes:

[0082] Step S402, extracting a feature vector of each simulated fingerprint feature matrix included in the simulated fingerprint feature matrix set output by the feature extraction layer;

[0083] Step S404, mapping the feature vectors of each simulated fingerprint feature matrix to a coordinate space of a preset dimension, and performing dimensionality reduction processing on the feature vectors of each simulated fingerprint feature matrix in the coordinate space to obtain feature vectors of a target dimension;

[0084] Step S406: Perform a visualization operation on the feature vectors of the target dimension of each simulated fingerprint feature matrix to display them on the target interface.

[0085] Optionally, this embodiment is applicable to the scenario of verifying the effectiveness of the medium classification network during the training of the medium classification network. In this embodiment, in order to effectively demonstrate the effectiveness of the constructed medium classification network for fingerprint medium classification, nonlinear dimensionality reduction technology (T-distributed Stochastic Neighbor Embedding, referred to as T-SNE) can be used to visualize the deep learning features of different media. The dimensionality reduction visualization based on T-SNE is used for material feature analysis. T-SNE is used to embed high-dimensional data into a low-dimensional space (the coordinate space of the preset dimension can be a two-dimensional or three-dimensional coordinate space) to facilitate the visualization of deep learning features, thereby revealing the similarities and differences between data points.

[0086] For example, in the qualitative analysis phase, the TSNE algorithm maps the eigenvectors of each digital matrix into a two-dimensional coordinate space and visualizes the output of the TSNE algorithm. Figure 5 As shown in the figure, the visual distribution of the 7-dimensional deep learning features after dimensionality reduction can be achieved through methods such as scatter plots, where the position of each data point represents its coordinates in the low-dimensional space, and different colors or shapes can be used to distinguish different media. The deep learning features of fingerprint data of 6 different media, including gel, silica gel, and resin, show strong separability and aggregability according to their physical properties, proving the feasibility of guiding the learning of implicit physical properties in the native domain of different media based on deep learning technology.

[0087] For example, in the quantitative analysis stage, the comparison of quantitative results of media detection is shown in Table 1. The classical statistical method of evaluating the performance and accuracy of multi-classification tasks is to use the classification accuracy confusion matrix of five different pressing media. The values ​​on the diagonal of the confusion matrix represent the probability that each media category is accurately classified to the correct result. These values ​​can be used to calculate the classification accuracy and other related indicators. Among them, the classification accuracy confusion matrix is ​​used to evaluate the performance of the classification network. For the classification task of five different pressing media, the confusion matrix will be a 5x5 matrix, where each row represents the actual category and each column represents the predicted category. For example, there are five pressing media: A, B, C, D, and E. The confusion matrix will include the following cells: True Positives (TP): The number of samples predicted to be in this category and actually in this category. False Positives (FP): The number of samples predicted to be in this category but actually not in this category. False Negatives (FN): The number of samples predicted not to be in this category but actually in this category. True Negatives (TN): For a binary classification problem, this can be the number of samples predicted to be not in this category and actually not in this category. Actual category: AB CDE; predicted category: AB ED A. The confusion matrix can be like this:

[0088] “AB CDE

[0089] A TP1 0 0 0 0FN1

[0090] B 0TP2 0FN2 0

[0091] C 0 0TP3 0 0

[0092] D 0FN3 0TP4 0

[0093] E FN4 0 0 0TP5";

[0094] Among them, `TP1` is used to represent the number of samples that are actually class A and correctly predicted to be class A, and `FN1` is used to represent the number of samples that are actually class A but incorrectly predicted to be other classes.

[0095] In this embodiment, the confusion matrix provides a comprehensive view of the classification performance, including the details of the correct classification and the wrong classification. Through the confusion matrix, you can have a more comprehensive understanding of the performance of the classifier and improve the model by adjusting the model parameters or using different classification algorithms.

[0096] Table 1:

[0097]

[0098] By observing the visualization results, this embodiment can analyze how the features of different fingerprint media are distributed in the low-dimensional space. If the features of the fingerprint medium form different clusters in the visualization, this means that the data of different media are distinguishable in the feature space. According to the visualization results, it may be necessary to return to the network training or feature extraction step to improve the performance of the network or select more appropriate features. Qualitative analysis of the correlation and difference of deep learning features of different media is achieved. It can effectively realize deep anti-counterfeiting feature extraction based on native domain features, and materials of the same medium are effectively aggregated in the reduced dimensional space. At the same time, different media also show strong separability even in two-dimensional space.

[0099] In an exemplary embodiment, in step S204, fingerprint feature extraction is performed on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected, including:

[0100] Step S2041, constructing a native domain information set, wherein the native domain information set includes a plurality of native domain information of the fingerprint to be detected acquired in a current time period, or the native domain information set includes a plurality of historical native domain information acquired in a historical time period and the native domain information acquired in the current time period;

[0101] Optionally, the above-mentioned multiple native domain information of the above-mentioned fingerprint to be detected obtained in the current time period refers to pressing the same fingerprint medium on the fingerprint collector multiple times to obtain more comprehensive and clear fingerprint information. Thereby increasing the number of fingerprint feature points collected, the accuracy of recognition can be improved. Continuous pressing can reduce the number of repeated collections due to collection failures, thereby improving the overall collection efficiency. The fingerprints corresponding to the multiple historical native domain information obtained in the historical time period can be the same fingerprint as the fingerprint corresponding to the native domain information obtained in the current time period, or they can be different fingerprints.

[0102] Step S2042, converting all native domain information included in the native domain information set into digital signals through a signal converter to obtain multiple groups of digital signals;

[0103] Optionally, in this embodiment, the native domain information is converted into a digital signal through a signal converter, and the digital feature signal is extracted, which is the same as the above-mentioned processing method of the native domain information, and will not be repeated here.

[0104] Step S2043, extracting digital features from each group of the digital signals to obtain fingerprint features of each group of the native domain information;

[0105] Step S2044, generating a matrix including fingerprint features of each group of the native domain information according to the signal acquisition channel of the target sensor for acquiring the native domain information included in the native domain information set, to obtain multiple groups of fingerprint feature matrices;

[0106] Optionally, according to the signal acquisition channel of the target sensor for acquiring the native domain information included in the above-mentioned native domain information set, generating a digital matrix including the digital features of each group of the above-mentioned digital signals includes but is not limited to being implemented in the following manner: defining the matrix dimension: determining the number of rows and columns of the generated digital matrix. Generally, the number of rows represents the number of time series or samples, and the number of columns represents different signal channels. Initializing the matrix: creating an empty matrix with appropriate dimensions for storing signal data collected from each channel. Signal acquisition: using a sensor to collect signals from each channel. Each channel may correspond to different types of signals, such as temperature, pressure, fingerprint patterns, etc. Data formatting: formatting the collected signal data into numerical values, which may be digital sampling values ​​of analog signals. Filling the matrix: filling the signal data of each channel into the corresponding columns of the matrix in sequence. For example, if the data of the first channel is temperature, it will be filled into the first column of the matrix; if the data of the second channel is pressure, it will be filled into the second column, and so on.

[0107] Step S2045, combining the multiple groups of fingerprint feature matrices to obtain the fingerprint feature matrix to be detected.

[0108] Optionally, the method of combining the above-mentioned multiple groups of digital matrices includes but is not limited to merging multiple groups of digital matrices into one digital matrix, for example, merging two matrices into a larger matrix. Specifically including: vertical merging (merging by row): placing two matrices up and down to form a new matrix, the number of whose rows is the sum of the number of rows of the two matrices, and the number of columns remains unchanged. Horizontal merging (merging by column): placing two matrices side by side to form a new matrix, the number of whose columns is the sum of the number of columns of the two matrices, and the number of rows remains unchanged. If the dimensions of the two matrices allow, they can also be placed in specific positions of a larger matrix to form a block matrix.

[0109] This embodiment collects multi-modal fingerprint data to identify the fingerprint medium, and can collect fingerprint features at different angles, thereby reducing the uncertainty that may exist in a single fingerprint feature, thereby improving the accuracy of fingerprint medium identification.

[0110] In an exemplary embodiment, in the above step S204, the fingerprint feature matrix is ​​classified by the trained medium classification network to obtain the fingerprint medium classification vector of the fingerprint to be detected, including:

[0111] Step S2041, inputting the fingerprint feature matrix to be detected into the feature extraction layer included in the medium classification network, and extracting the feature vectors of each group of fingerprint feature matrices included in the fingerprint feature matrix to be detected through the feature extraction layer;

[0112] Step S2042, inputting the feature vectors of each group of fingerprint feature matrices outputted by the feature extraction layer into the classification layer included in the medium classification network, and calculating the probability that the feature vectors of each group of fingerprint feature matrices belong to the classification labels of the plurality of media respectively by the classification layer;

[0113] Step S2043, combining the multiple classification vectors output by the above classification layer to obtain the above fingerprint medium classification vector.

[0114] Optionally, the fingerprint feature matrix is ​​input into the feature extraction layer included in the above-mentioned medium classification network, and the feature extraction layer extracts each layer of the digital matrix in the fingerprint feature matrix layer by layer to obtain the feature vector of each layer. In the same way, the classification layer also calculates the probability that each feature vector belongs to each category. The probability of each category is converted into a vector distribution, and finally multiple groups of medium vectors are obtained. Finally, the medium vector corresponding to the current time period of the bottom layer is determined as the medium vector of the fingerprint medium to be detected.

[0115] In this embodiment, by using the combined data modality as the input of the medium classification network, different data modalities may capture different characteristics of the fingerprint medium to be detected. Combining these modalities can improve the performance of the classification layer. In addition, combining multiple modalities can increase the diversity of fingerprint features, thereby reducing the risk of overfitting. In general, the performance and adaptability of the medium classification network in the medium detection task can be significantly improved.

[0116] The present application is further described below in conjunction with specific embodiments:

[0117] This specific embodiment uses a deep learning network to build a medium detection model based on the differences in the physical properties of different media in the simulated fingerprint (including but not limited to capacitance impedance value, relative depth of valley ridges, acoustic impedance, optical polarization characteristics, etc.), collects data for training and optimization, and constructs a deep learning network CNN model. The detection and classification of fingerprint media in the under-screen fingerprint system is used as an example for explanation. Figure 6 The present specific embodiment comprises the following steps:

[0118] S601, collect training data, that is, collect the native domain information set of simulated fingerprints. For example, in the same environment, different media processes such as gel, PU, ​​latex, silicone, acrylic, resin, etc. are used to manufacture fake fingerprints corresponding to real fingers. Collect the native domain data of fingerprints corresponding to each medium.

[0119] S602, construct a CNN model. For example, Figure 4 The CNN model shown in Figure 1. The input is the whole image data of the ultrasonic fingerprint Raw domain, and the output is the medium classification result. The network architecture mainly consists of two parts: the feature extraction layer and the classification layer. The feature extraction layer is composed of a stack of Conv layer, ReLU activation layer, MaxPool maximum pooling layer, and GlobalAvgPool layer, which is used to extract a one-dimensional deep feature vector from the ultrasonic fingerprint input image. The classification layer is composed of FC and Softmax activation function layers, which are used to predict the probability scores of 7 media from the deep feature vector. The CNN model outputs guide labels as 0-unknown, 1-gel, 2-PU, 3-latex, 4-silicone, 5-acrylic, and 6-resin. For the convenience of training, it is converted into one-hot encoding. Each category corresponds to a one-hot encoding vector, in which only the element value of the corresponding category is 1, and the element values ​​of other categories are all 0.

[0120] S603, training and optimizing the CNN model, using the training set constructed in S601 to train and optimize the CNN model constructed in S602, so that it is suitable for the medium detection task.

[0121] S604, forged fingerprint detection. After the training is completed, the CNN model of S603 can be directly applied to the fingerprint medium detection task. In actual application, the press frame data is input into the CNN model, and the CNN model outputs the predicted scores of the five media respectively. The range is (0,1), the maximum The category corresponding to the value subscript is the medium category of the input image predicted by the model. The detection range includes but is not limited to the detection of gel, PU, ​​latex, silicone, acrylic, resin and unknown materials of simulated fingerprints.

[0122] This specific embodiment takes the ultrasonic under-screen fingerprint system as an example to demonstrate the differences in physical properties implied in the features extracted by deep learning technology for different media, and constructs a deep learning convolutional neural network model for medium detection. This reduces the algorithm calculation complexity without relying on additional complex sensors, thereby ensuring high accuracy of medium detection.

[0123] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0124] In this embodiment, a fingerprint detection device is also provided, which is used to implement the above embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0125] Figure 7 is a structural block diagram of a fingerprint detection device according to an embodiment of the present application, such as Figure 7 As shown, the device comprises:

[0126] A first acquisition module 72 is used to acquire native domain information of the fingerprint to be detected, wherein the native domain information is original data collected when fingerprint verification is performed on the fingerprint to be detected;

[0127] A first extraction module 74, configured to perform fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected;

[0128] A first classification module 76 is used to perform a classification operation on the fingerprint feature matrix through a trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector;

[0129] The first determination module 78 is used to determine the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0130] In an exemplary embodiment, the above-mentioned device also includes: a first acquisition unit, which is used to perform a classification operation on the fingerprint feature matrix through a trained medium classification network, and before obtaining the fingerprint medium classification vector of the fingerprint to be detected, the simulated fingerprint native domain information of the simulated fingerprint simulated on multiple media is respectively collected to obtain a simulated fingerprint native domain information set, wherein the part of the simulated fingerprint native domain information included in the simulated fingerprint native domain information set is: multiple simulated fingerprint native domain information obtained after multiple information collections are performed on the same simulated fingerprint simulated on the same medium; a first execution unit, which is used to perform fingerprint feature extraction on each of the simulated fingerprint native domain information included in the simulated fingerprint native domain information set to obtain a simulated fingerprint feature matrix set; a first training unit, which is used to train the medium classification network using the simulated fingerprint feature matrix set and the classification labels of the multiple media to obtain the medium classification network, wherein the classification labels are used to represent the media types of the multiple media.

[0131] In an exemplary embodiment, the first execution unit includes: a first execution submodule, which is used to perform the following operations for each simulated fingerprint native domain information included in the simulated fingerprint native domain information set to obtain a simulated fingerprint feature matrix of each simulated fingerprint native domain information, wherein the simulated fingerprint feature matrix set includes the simulated fingerprint feature matrix of each simulated fingerprint native domain information; perform a signal conversion operation on the simulated fingerprint native domain information through a signal converter to obtain a group of digital signals of the simulated fingerprint native domain information; extract digital features from the group of digital signals to obtain a group of fingerprint features of the simulated fingerprint native domain information; determine the simulated fingerprint feature matrix of the simulated fingerprint native domain information as a group of fingerprint features including the simulated fingerprint native domain information.

[0132] In an exemplary embodiment, the first training unit includes: a first input submodule, which is used to input the set of simulated fingerprint feature matrices into the medium classification network to perform a training operation, wherein the training operation includes: inputting the simulated fingerprint feature matrix into the feature extraction layer included in the medium classification network, outputting the feature vector of the simulated fingerprint feature matrix through the feature extraction layer, inputting the feature vector of the digital matrix into the classification layer included in the medium classification network, calculating the probability that the feature vector of the simulated fingerprint feature matrix belongs to the classification labels of multiple media through the classification layer, and outputting the medium classification vector, wherein the simulated fingerprint feature matrix is The fingerprint feature matrix included in the digital matrix set; a first determination submodule, used to determine the loss value output by the loss function of the medium classification network based on the medium classification vector and the classification label of the target medium, wherein the target medium is the medium corresponding to the digital matrix; a first processing submodule, used to terminate the training to obtain the medium classification network when the loss value meets the preset training end condition; a first adjustment submodule, used to adjust the parameter value of the medium classification network when the loss value does not meet the preset training end condition, so as to reduce the loss value output by the loss function of the medium classification network in the next training.

[0133] In an exemplary embodiment, the above-mentioned device also includes: a first extraction module, which is used to extract the feature vector of each simulated fingerprint feature matrix included in the simulated fingerprint feature matrix set output by the feature extraction layer after inputting the simulated fingerprint feature matrix set into the medium classification network to perform a training operation; a first mapping module, which is used to map the feature vectors of each simulated fingerprint feature matrix to a coordinate space of a preset dimension, and perform dimensionality reduction processing on the feature vectors of each simulated fingerprint feature matrix in the coordinate space to obtain feature vectors of a target dimension; and a first operation module, which is used to perform visualization operations on the feature vectors of the target dimension of each simulated fingerprint feature matrix to display them on a target interface.

[0134] In an exemplary embodiment, the above-mentioned first extraction module includes: a first construction unit, which is used to construct a native domain information set, wherein the native domain information set includes multiple native domain information of the fingerprint to be detected obtained in the current time period, or the native domain information set includes multiple historical native domain information obtained in the historical time period and the native domain information obtained in the current time period; a first conversion unit, which is used to convert the native domain information included in the native domain information set into digital signals through a signal converter to obtain multiple groups of digital signals; a first extraction unit, which is used to extract digital features from each group of digital signals to obtain fingerprint features of each group of native domain information; a first generation unit, which is used to generate a matrix including fingerprint features of each group of native domain information according to the signal acquisition channel of the target sensor that obtains the native domain information included in the native domain information set, to obtain multiple groups of fingerprint feature matrices; a first combination unit, which is used to combine the multiple groups of fingerprint feature matrices to obtain the fingerprint feature matrix to be detected.

[0135] In an exemplary embodiment, the above-mentioned first classification module includes: a first input unit, used to input the fingerprint feature matrix to be detected into the feature extraction layer included in the medium classification network, and extract the feature vectors of each group of fingerprint feature matrices included in the fingerprint feature matrix to be detected through the feature extraction layer; a second input unit, used to input the feature vectors of each group of fingerprint feature matrices outputted through the feature extraction layer into the classification layer included in the medium classification network, and calculate the probability that the feature vectors of each group of fingerprint feature matrices belong to the classification labels of multiple media through the classification layer; a second combination unit, used to combine the multiple classification vectors outputted by the above-mentioned classification layer to obtain the above-mentioned fingerprint medium classification vector.

[0136] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0137] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0138] According to one aspect of the present application, a computer program product is provided, the computer program product comprising a computer program / instruction, the computer program / instruction comprising a method for executing a process Figure 8In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the central processor 801, various functions provided by the embodiments of the present application are executed.

[0139] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0140] Figure 8 The structure block diagram of a computer system for implementing an electronic device according to an embodiment of the present application is schematically shown.

[0141] It should be noted that Figure 8 The computer system 800 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0142] like Figure 8 As shown, the computer system 800 includes a central processing unit 801 (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory 802 (ROM) or the program loaded from the storage part 808 to the random access memory 803 (RAM). Various programs and data required for system operation are also stored in the random access memory 803. The central processing unit 801, the read-only memory 802 and the random access memory 803 are connected to each other through a bus 804. The input / output interface 805 (Input / Output interface, i.e., I / O interface) is also connected to the bus 804.

[0143] The following components are connected to the input / output interface 805: an input section 1006 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a local area network card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the input / output interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed so that a computer program read therefrom is installed into the storage section 808 as needed.

[0144] In particular, according to an embodiment of the present application, the process described in each method flow chart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program contains a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the central processor 801, various functions defined in the system of the present application are executed.

[0145] According to another aspect of the embodiment of the present application, an electronic device for implementing the above-mentioned fingerprint detection method is also provided. The electronic device may be Figure 1 The terminal device or server shown in the figure. This embodiment is described by taking the electronic device as a terminal device as an example. Fig. 9 As shown, the electronic device includes a memory 902 and a processor 904. The memory 902 stores a computer program, and the processor 904 is configured to execute the steps in any of the above method embodiments through the computer program.

[0146] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0147] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0148] S1, obtaining native domain information of the fingerprint to be detected, wherein the native domain information is the original data collected when the fingerprint to be detected is verified;

[0149] S2, performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected;

[0150] S3, performing classification operation on the fingerprint feature matrix through the trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector;

[0151] S4, determining the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0152] Alternatively, a person skilled in the art may understand that: Fig. 9 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (MID), a PAD, and other terminal devices. Fig. 9 The electronic device and the electronic equipment described above are not limited in structure. Fig. 9 More or fewer components (such as network interfaces, etc.) as shown in, or with Fig. 9 Different configurations shown.

[0153] Among them, the memory 902 can be used to store software programs and modules, such as the program instructions / modules corresponding to the fingerprint detection method and device in the embodiment of the present application. The processor 904 executes various functional applications and data processing by running the software programs and modules stored in the memory 902, that is, to implement the above-mentioned fingerprint detection method. The memory 902 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 902 may further include a memory remotely located relative to the processor 904, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Among them, the memory 902 can specifically be, but is not limited to, used to store information such as codes of items. As an example, such as Fig. 9 As shown, the memory 902 may include, but is not limited to, the first acquisition module 72, the first extraction module 74, the first classification module 76, and the first determination module 78 in the fingerprint detection device. In addition, it may also include, but is not limited to, other module units in the fingerprint detection device, which will not be repeated in this example.

[0154] Optionally, the transmission device 906 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wired network and a wireless network. In one example, the transmission device 906 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers via a network cable so as to communicate with the Internet or a local area network. In one example, the transmission device 906 is a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0155] In addition, the electronic device further comprises: a display 908 for displaying the initial code; and a connection bus 910 for connecting various module components in the electronic device.

[0156] In other embodiments, the terminal device or server may be a node in a distributed system, wherein the distributed system may be a blockchain system, and the blockchain system may be a distributed system formed by connecting the multiple nodes through network communication. The nodes may form a peer-to-peer network, and any form of computing device, such as a server, terminal or other electronic device, may become a node in the blockchain system by joining the peer-to-peer network.

[0157] According to one aspect of the present application, a computer-readable storage medium is provided, and a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the fingerprint detection method provided in various optional implementations of the above-mentioned fingerprint medium type determination.

[0158] Optionally, in this embodiment, the computer-readable storage medium may be configured to store a computer program for performing the following steps:

[0159] S1, obtaining native domain information of the fingerprint to be detected, wherein the native domain information is the original data collected when the fingerprint to be detected is verified;

[0160] S2, performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected;

[0161] S3, performing classification operation on the fingerprint feature matrix through the trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector;

[0162] S4, determining the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

[0163] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing hardware related to the terminal device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0164] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0165] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0166] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0167] In the several embodiments provided in the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0168] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0169] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0170] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A fingerprint detection method, characterized in that: include: Acquire native domain information of the fingerprint to be detected, wherein the native domain information is original data collected when fingerprint verification is performed on the fingerprint to be detected; Performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected; Performing a classification operation on the fingerprint feature matrix through the trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple media types in the form of a vector; The medium type of the fingerprint to be detected is determined based on the fingerprint medium classification vector.

2. The method according to claim 1, characterized in that Before performing a classification operation on the fingerprint feature matrix through the trained medium classification network to obtain the fingerprint medium classification vector of the fingerprint to be detected, the method further includes: The native domain information of the simulated fingerprints simulated on the multiple media are collected respectively to obtain multiple sets of the native domain information of the simulated fingerprints, wherein the multiple sets of the native domain information of the simulated fingerprints include: the native domain information of multiple simulated fingerprints of the same medium obtained after collecting information of the simulated fingerprint of the same medium multiple times; Performing fingerprint feature extraction on each of the simulated fingerprint native domain information included in the multiple simulated fingerprint native domain information sets to obtain multiple simulated fingerprint feature matrix sets; The medium classification network is trained using the simulated fingerprint feature matrix set and the classification labels of the multiple media to obtain the medium classification network, wherein the classification labels are used to represent the medium types of the multiple media.

3. The method according to claim 2, characterized in that Using the simulated fingerprint feature matrix set and the classification labels of the multiple media, training the pre-trained medium classification network to obtain the trained medium classification network includes: Inputting the set of simulated fingerprint feature matrices into the medium classification network to perform a training operation, wherein the training operation includes: inputting the simulated fingerprint feature matrix into a feature extraction layer included in the medium classification network, outputting a feature vector of the simulated fingerprint feature matrix through the feature extraction layer, inputting the feature vector of the simulated fingerprint feature matrix into a classification layer included in the medium classification network, calculating the probability that the feature vector of the simulated fingerprint feature matrix belongs to a plurality of classification labels of the medium through the classification layer, and outputting a medium classification vector; Determine a loss value output by a loss function of the medium classification network based on the medium classification vector and the classification label of the medium; When the loss value satisfies a preset training end condition, the training is ended to obtain the trained medium classification network; When the loss value does not satisfy the preset training end condition, the parameter values ​​of the medium classification network are adjusted to reduce the loss value output by the loss function of the medium classification network in the next training.

4. The method according to claim 3, characterized in that After inputting the simulated fingerprint feature matrix set into the medium classification network to perform a training operation, the method further includes: Extracting a feature vector of each simulated fingerprint feature matrix included in the simulated fingerprint feature matrix set output by the feature extraction layer; Mapping the feature vectors of each simulated fingerprint feature matrix to a coordinate space of a preset dimension, and performing dimensionality reduction processing on the feature vectors of each simulated fingerprint feature matrix in the coordinate space to obtain feature vectors of a target dimension; The feature vectors of the target dimension of each simulated fingerprint feature matrix are visualized to be displayed on the target interface.

5. The method according to claim 1, characterized in that Performing fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected includes: Constructing a native domain information set, wherein the native domain information set includes a plurality of native domain information of the fingerprint to be detected acquired in a current time period, or the native domain information set includes a plurality of historical native domain information acquired in a historical time period and the native domain information acquired in the current time period; The native domain information included in the native domain information set is converted into digital signals by a signal converter to obtain multiple groups of digital signals; Extracting digital features from each group of digital signals to obtain fingerprint features of each group of native domain information; According to the signal acquisition channel of the target sensor for acquiring the native domain information included in the native domain information set, a matrix including fingerprint features of each group of the native domain information is generated to obtain multiple groups of fingerprint feature matrices; The multiple groups of fingerprint feature matrices are combined to obtain the fingerprint feature matrix to be detected.

6. The method according to claim 5, characterized in that Performing a classification operation on the fingerprint feature matrix through the trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected includes: Inputting the fingerprint feature matrix to be detected into the feature extraction layer included in the medium classification network, and extracting the feature vectors of each group of fingerprint feature matrices included in the fingerprint feature matrix to be detected through the feature extraction layer; Inputting the feature vectors of each group of fingerprint feature matrices outputted by the feature extraction layer into the classification layer included in the medium classification network, and respectively calculating the probabilities that the feature vectors of each group of fingerprint feature matrices belong to classification labels of multiple media by the classification layer; The fingerprint medium classification vector is obtained by combining multiple classification vectors output by the classification layer.

7. A fingerprint detection device, characterized in that: include: A first acquisition module is used to acquire native domain information of a fingerprint to be detected, wherein the native domain information is original data collected when fingerprint verification is performed on the fingerprint to be detected; A first extraction module, configured to perform fingerprint feature extraction on the native domain information to obtain a fingerprint feature matrix of the fingerprint to be detected; A first classification module is used to perform a classification operation on the fingerprint feature matrix through a trained medium classification network to obtain a fingerprint medium classification vector of the fingerprint to be detected, wherein the fingerprint medium classification vector is used to represent the probability that the fingerprint to be detected belongs to multiple medium types in the form of a vector; The first determination module is used to determine the medium type of the fingerprint to be detected based on the fingerprint medium classification vector.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 7 when executed by a processor.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 7 are implemented.

Images