Fingerprint carrier state recognition method and apparatus, electronic device, and storage medium

Abstract

Embodiments of the present application provide a fingerprint carrier state recognition method and apparatus, an electronic device, and a storage medium. The fingerprint carrier state recognition method comprises: acquiring a current fingerprint image frame; inputting feature information of the current fingerprint image frame, time information corresponding to the current fingerprint image frame, and a latent variable corresponding to the current fingerprint image frame into a recurrent neural network model to obtain carrier state information outputted by the recurrent neural network model and a latent variable corresponding to a next fingerprint image frame, wherein a fingerprint carrier is a medium carrying a fingerprint corresponding to a fingerprint image; and determining the state of the fingerprint carrier on the basis of the carrier state information. In the fingerprint carrier state recognition method provided by the present application, carrier state recognition is performed with reference to the time information of the current fingerprint image frame and a latent variable obtained from a historical fingerprint image, so that a high recognition accuracy rate can be achieved.

Description

Fingerprint carrier status recognition method, device, electronic device and storage medium Technical Field

[0001] This application relates to the field of fingerprint recognition technology, and in particular to a fingerprint carrier status recognition method, device, electronic device and storage medium. Background Technology

[0002] Fingerprint recognition is a biometric technology. Because fingerprint recognition is widely used in various fields such as smartphone unlocking, financial transactions, and access control, it is crucial to prevent unauthorized unlocking caused by fake fingerprints and foreign object attacks, as well as the resulting information leaks, in everyday applications.

[0003] Currently, when a finger is pressed on the recognition area of ​​the fingerprint module, the system determines whether the fingerprint carrier corresponding to the fingerprint image is a counterfeit finger by acquiring a single frame or multiple frames of fingerprint images.

[0004] However, single-frame fingerprint image anti-counterfeiting solutions are limited by the information captured in a single session and are susceptible to interference from factors such as localized pressing or dirty fingers, leading to false positives. Multi-frame fingerprint image anti-counterfeiting solutions suffer from poor anti-counterfeiting effectiveness due to high repetition rates and lengthy authentication times, resulting in a poor user experience. Summary of the Invention

[0005] In view of the above, embodiments of this application provide a fingerprint carrier status recognition method, apparatus, electronic device, and computer storage medium to at least partially solve the above problems.

[0006] According to a first aspect of the embodiments of this application, a fingerprint carrier state recognition method is provided, comprising: acquiring a current frame fingerprint image; inputting feature information of the current frame fingerprint image, time information corresponding to the current frame fingerprint image, and latent variables corresponding to the current frame fingerprint image into a recurrent neural network model to obtain carrier state information output by the recurrent neural network model and latent variables corresponding to the next frame fingerprint image, wherein the fingerprint carrier is a medium carrying the fingerprint corresponding to the fingerprint image; and determining the state of the fingerprint carrier according to the carrier state information.

[0007] In one possible implementation, the feature information of the current frame fingerprint image includes at least one of the fingerprint signal output by the fingerprint sensor when the current frame fingerprint image is acquired, the current frame fingerprint image, and fingerprint features extracted from the current frame fingerprint image.

[0008] In one possible implementation, the time information corresponding to the current frame fingerprint image includes the time of acquiring the current frame fingerprint image, and / or the time interval between acquiring the current frame fingerprint image and acquiring the previous frame fingerprint image.

[0009] In one possible implementation, the state of the fingerprint carrier includes at least one of: whether the fingerprint carrier is a result of the bifurcation of a finger, the medium classification of the fingerprint carrier, and the physical state of the fingerprint carrier.

[0010] In one possible implementation, the recurrent neural network model is obtained by: acquiring multiple fingerprint image queues, wherein each fingerprint image queue includes multiple sample fingerprint images, and the sample fingerprint images included in each fingerprint image queue are obtained by fingerprint acquisition from at least one fingerprint carrier; determining the temporal information of each fingerprint image queue, wherein the temporal information includes the time information corresponding to each sample fingerprint image in the fingerprint image queue; determining the carrier state information of the last sample fingerprint image in each fingerprint image queue; constructing training samples corresponding to each fingerprint image queue, wherein the training samples include the fingerprint image queue, the temporal information corresponding to the fingerprint image queue, and the carrier state information of the last sample fingerprint image in the fingerprint image queue; and training the model to be trained using each training sample to obtain the recurrent neural network model.

[0011] In one possible implementation, acquiring multiple fingerprint image queues includes: acquiring sample fingerprint images corresponding to fingerprints carried by at least two fingerprint carriers to obtain at least two sample fingerprint image libraries; extracting multiple sample fingerprint images from the at least two sample fingerprint image libraries according to a target time interval to obtain multiple sets of sample fingerprint image queues, wherein the sample fingerprint image queues include multiple sample fingerprint images extracted from at least one sample fingerprint library; and sorting each set of sample fingerprint image queues according to the extraction order to obtain multiple fingerprint image queues.

[0012] In one possible implementation, training the model to be trained using each of the training samples to obtain the recurrent neural network model includes: inputting the fingerprint image queue and the corresponding time-series information of each training sample into the model to be trained to obtain the recognition result corresponding to each training sample, wherein the recognition result is used to indicate the carrier state of the last sample fingerprint image in the fingerprint image queue of the training sample; determining the matching result corresponding to each training sample based on the recognition result corresponding to each training sample and the carrier state information of the last sample fingerprint image in the training sample; and correcting the parameters of the model to be trained based on the matching results corresponding to each training sample until the matching results corresponding to each training sample meet the training completion conditions to obtain the recurrent neural network model.

[0013] In one possible implementation, the step of inputting the fingerprint image queues and corresponding time information of each training sample into the model to be trained to obtain the recognition result corresponding to each training sample includes: performing the following operations on the fingerprint image queues in each training sample: obtaining an initial variable; inputting the initial variable, the fingerprint image of the first sample in the fingerprint image queue, and the time information corresponding to the fingerprint image of the first sample in the fingerprint image queue into the model to be trained to obtain a first variable; inputting the (n-1)th variable, the fingerprint image of the nth sample in the fingerprint image queue, and the time information corresponding to the fingerprint image of the nth sample in the fingerprint image queue into the model to be trained to obtain an nth variable; inputting the (m-1)th variable, the fingerprint image of the last sample in the fingerprint image queue, and the time information corresponding to the fingerprint image of the last sample in the fingerprint image queue into the model to be trained to obtain the recognition result corresponding to the fingerprint image queue, where m is the number of sample fingerprint images included in the fingerprint image queue, and n is an integer less than or equal to m-1 and greater than or equal to 2.

[0014] In one possible implementation, the recurrent neural network model is obtained by: acquiring model parameters; and correcting the parameters in the model to be trained based on the model parameters to obtain the recurrent neural network model.

[0015] According to a second aspect of the embodiments of this application, a fingerprint carrier state recognition device is provided, comprising: an acquisition unit for acquiring a current frame fingerprint image; an input unit for inputting feature information of the current frame fingerprint image, time information corresponding to the current frame fingerprint image, and latent variables corresponding to the current frame fingerprint image into a recurrent neural network model to obtain carrier state information output by the recurrent neural network model and latent variables corresponding to the next frame fingerprint image, wherein the fingerprint carrier is a medium carrying the fingerprint corresponding to the fingerprint image; and a determination unit for determining the state of the fingerprint carrier based on the carrier state information.

[0016] According to a third aspect of the present application, an electronic device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; the memory is used to store at least one executable instruction, wherein the executable instruction causes the processor to perform an operation corresponding to the method described in the first aspect.

[0017] According to a fourth aspect of the embodiments of this application, a computer storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0018] According to the fingerprint carrier state recognition scheme provided in this application embodiment, the current frame fingerprint image is acquired. The feature information of the current frame fingerprint image, the time information corresponding to the current frame fingerprint image, and the latent variables corresponding to the current frame fingerprint image are input into a recurrent neural network model to determine the carrier state information corresponding to the current frame fingerprint image. Thus, the fingerprint carrier state corresponding to the current frame fingerprint image can be determined. Fingerprint anti-counterfeiting and derivation operations can be performed based on the determined fingerprint carrier state. Since the content input to the recurrent neural network model includes the latent variables of the current frame fingerprint image, and the latent variables of the current frame fingerprint image are determined based on historical fingerprint images, historical fingerprint images and the acquisition time of the current frame fingerprint image are referenced when performing carrier state recognition. Compared with the existing scheme of anti-counterfeiting recognition based on a single frame fingerprint image, since the fingerprint anti-counterfeiting result is not determined solely based on a single frame fingerprint image, the recognition accuracy can be improved. Compared with the existing scheme of anti-counterfeiting recognition based on multiple frames of fingerprint images, since the current frame fingerprint image is recognized through a recurrent neural network model, less time is consumed, which can improve the user experience. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0020] Figure 1 is a flowchart of a fingerprint carrier status recognition method provided in an embodiment of this application;

[0021] Figure 2 is a flowchart of a recurrent neural network model training method provided in an embodiment of this application;

[0022] Figure 3 is a schematic diagram of a fingerprint image queue construction process provided in an embodiment of this application;

[0023] Figure 4 is a flowchart of a recurrent neural network model training method provided in an embodiment of this application;

[0024] Figure 5 is a schematic diagram of a fingerprint carrier status recognition device provided in an embodiment of this application;

[0025] Figure 6 is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the technical solutions in the embodiments of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of this application.

[0027] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0028] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0029] As mentioned earlier, fingerprint recognition is a biometric technology. Since fingerprint recognition is widely used in various fields such as smartphone unlocking, financial transactions, and access control, it is crucial to prevent unauthorized unlocking caused by fake fingerprints and foreign object attacks, as well as the resulting information leakage issues, in daily applications. Currently, when a finger is pressed on the fingerprint module's recognition area, the system determines whether the fingerprint carrier corresponding to that image is a counterfeit finger by acquiring a single or multiple fingerprint images. However, single-frame fingerprint image anti-counterfeiting schemes are limited by the information available in a single acquisition and are susceptible to interference from factors such as partial pressing or dirty fingers, leading to false positives. For multi-frame fingerprint image anti-counterfeiting schemes, the high repetition rate of acquisition results in poor anti-counterfeiting effectiveness, and the long authentication time required results in a poor user experience for existing fingerprint anti-counterfeiting solutions.

[0030] This application provides a fingerprint carrier state recognition method. It acquires a current frame fingerprint image, inputs the feature information of the current frame fingerprint image, the corresponding time information, and the latent variables corresponding to the current frame fingerprint image into a recurrent neural network model to determine the carrier state information corresponding to the current frame fingerprint image. This allows for the determination of the fingerprint carrier state corresponding to the current frame fingerprint image, enabling fingerprint anti-counterfeiting and derivation operations. Since the input to the recurrent neural network model includes the latent variables of the current frame fingerprint image, and these latent variables are determined based on historical fingerprint images, the method references historical fingerprint images and the acquisition time of the current frame fingerprint image during carrier state recognition. Compared to existing anti-counterfeiting recognition schemes that use a single frame fingerprint image, this method improves recognition accuracy because it does not rely solely on a single frame fingerprint image to determine the anti-counterfeiting result. Compared to existing anti-counterfeiting recognition using multiple frames of fingerprint images, this method consumes less time and improves user experience because it uses a recurrent neural network model to recognize the current frame fingerprint image.

[0031] The fingerprint carrier status recognition method provided in this application is illustrated below through examples.

[0032] Figure 1 is a flowchart of a fingerprint carrier state recognition method provided in an embodiment of this application. As shown in Figure 1, the fingerprint carrier state recognition method includes the following steps 101 to 103:

[0033] Step 101: Obtain the fingerprint image of the current frame.

[0034] When a fingerprint is pressed on the fingerprint sensor, the fingerprint sensor captures the current frame fingerprint image, which is a single frame fingerprint image.

[0035] Step 102: Input the feature information of the current frame fingerprint image, the time information corresponding to the current frame fingerprint image, and the latent variables corresponding to the current frame fingerprint image into the recurrent neural network model to obtain the carrier state information output by the recurrent neural network model and the latent variables corresponding to the next frame fingerprint image.

[0036] The fingerprint carrier is the medium that carries the fingerprint corresponding to the fingerprint image. The fingerprint carrier can be a finger, 2D printing material, 3D silicone material, etc. The latent variables corresponding to the current frame fingerprint image can be obtained from historically acquired fingerprint images. Specifically, when a fingerprint image is acquired for the first time, the latent variables corresponding to the current frame fingerprint image are preset values, for example, they can be set to 0. When a fingerprint image is acquired for the first time, the latent variables corresponding to the current frame fingerprint image are the latent variables output by the previous frame fingerprint image after being input into the recurrent neural network model.

[0037] The recurrent neural network model outputs the carrier state information corresponding to the current frame fingerprint image and the latent variables corresponding to the next frame fingerprint image, based on the feature information of the current frame fingerprint image, the time information corresponding to the current frame fingerprint image, and the latent variables corresponding to the current frame fingerprint image. The latent variables corresponding to the next frame fingerprint image can be used for carrier state recognition in the next frame fingerprint image. In one example, the recurrent neural network can be represented by the following formula: [Y t ,h t+1 ]=f(X t ,T t ,h t ), t=1,2,…,M, t is used to characterize the fingerprint image of the t-th frame, Y t h is used to characterize the carrier state information corresponding to the t-th frame fingerprint image. t X is used to characterize the latent variable corresponding to the fingerprint image of frame t. t T is used to characterize the feature information of the t-th frame fingerprint image. t Used to characterize the time information corresponding to the t-th frame fingerprint image.

[0038] In one example, the recurrent neural network model can employ various time-series signal processing recurrent architectures, including Recurrent Neural Network (RNN), Gate Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and Mamba.

[0039] Optionally, the latent variables corresponding to the next frame fingerprint image are obtained by updating the latent variables of the current frame fingerprint image based on the feature information and time information of the current frame fingerprint image input into the model. In one example, for a recurrent neural network with an RNN architecture, the latent variables can be updated using the following formula: b, U, and W are used to characterize the activation function, and b, U, and W are used to characterize the fully connected parameters. X is used to characterize the activation function. t T is used to characterize the feature information of the t-th frame fingerprint image. t Used to characterize the time information corresponding to the t-th frame fingerprint image.

[0040] Step 103: Determine the status of the fingerprint carrier based on the carrier status information.

[0041] The state of the fingerprint carrier is determined based on the carrier state information. For example, it can be determined based on the Y mentioned above. t Determine the state of the fingerprint carrier; for example, determine whether the fingerprint carrier is a finger.

[0042] In this embodiment, the current frame fingerprint image is acquired, and its feature information, corresponding time information, and latent variables are input into a recurrent neural network model to determine the carrier state information corresponding to the current frame fingerprint image. This allows for the determination of the fingerprint carrier state, enabling fingerprint anti-counterfeiting and derivation operations. Since the input to the recurrent neural network model includes the latent variables of the current frame fingerprint image, and these latent variables are determined based on historical fingerprint images, historical fingerprint images and the acquisition time of the current frame fingerprint image are referenced during carrier state identification. Compared to existing anti-counterfeiting identification schemes using a single frame fingerprint image, this approach improves accuracy because it doesn't rely solely on a single frame. Furthermore, compared to existing anti-counterfeiting identification using multiple frames of fingerprint images, this approach, using a recurrent neural network model for current frame fingerprint image identification, consumes less time and improves user experience.

[0043] In one possible implementation, the feature information of the current frame fingerprint image includes at least one of the fingerprint signal output by the fingerprint sensor when the current frame fingerprint image is acquired, the current frame fingerprint image, and fingerprint features extracted from the current frame fingerprint image.

[0044] The feature information of the current frame fingerprint image can be the fingerprint signal output by the fingerprint sensor when the current frame fingerprint image is acquired, such as the optical imaging result of an optical fingerprint module, the capacitive reactance value of a capacitive fingerprint module, or the ultrasonic echo of an ultrasonic fingerprint module. The feature information of the current frame fingerprint image can also be the fingerprint image itself, such as the fingerprint image identified by the processing unit based on the fingerprint signal output by the fingerprint sensor. Furthermore, the feature information of the current frame fingerprint image can be fingerprint features extracted from the current frame fingerprint image, such as fingerprint features obtained by performing operations such as convolution on the current frame fingerprint image.

[0045] In this embodiment, the feature information of the current frame fingerprint image includes at least one of the fingerprint signal output by the fingerprint sensor when the current frame fingerprint image is acquired, the current frame fingerprint image, and the fingerprint features extracted from the current frame fingerprint image. Thus, the features of the current frame fingerprint image can be accurately represented by the feature information of the current frame fingerprint image, enabling the recurrent neural network model to obtain the features of the current frame fingerprint image.

[0046] In one possible implementation, the time information corresponding to the current frame fingerprint image includes the time when the current frame fingerprint image was acquired, and / or the time interval between acquiring the current frame fingerprint image and acquiring the previous frame fingerprint image.

[0047] It should be noted that if the current fingerprint image is acquired within a very short time after the carrier state identification of the previous fingerprint image, the probability of the fingerprint carrier changing is low. Therefore, the proportion of latent variables for the current fingerprint image in the recurrent neural network model is relatively large. That is, when the time interval between acquiring the current fingerprint image and acquiring the previous fingerprint image is small, the historical fingerprint image has high reference significance. Conversely, if the current fingerprint image is acquired after a relatively long time following the carrier state identification of the previous fingerprint image, the probability of the fingerprint carrier corresponding to the fingerprint image changing is high. Therefore, the proportion of latent variables for the current fingerprint image in the recurrent neural network model is relatively small. That is, when the time interval between acquiring the current fingerprint image and acquiring the previous fingerprint image is large, the historical fingerprint image has low reference significance.

[0048] In this embodiment, the time information corresponding to the current frame fingerprint image includes the time of acquiring the current frame fingerprint image and / or the time interval between acquiring the current frame fingerprint image and acquiring the previous frame fingerprint image. Thus, the time information of the current frame fingerprint image can be input into the recurrent neural network model, so that the recurrent neural network model refers to the time information when performing fingerprint carrier state recognition, thereby determining the reference weight of the latent variable and improving the recognition accuracy of the fingerprint carrier state.

[0049] In one possible implementation, the state of the fingerprint carrier includes at least one of the following: whether the fingerprint carrier is a result of the bifurcation of a finger, the medium classification of the fingerprint carrier, and the physical state of the fingerprint carrier.

[0050] The state of the fingerprint carrier, determined based on the carrier state information, can be a binary result of whether the fingerprint carrier is a finger, i.e., output whether the fingerprint carrier is a finger, output 1 if it is, otherwise output 0. The state of the fingerprint carrier can also be the medium classification of the fingerprint carrier, such as: the carrier is numerical, the carrier is 3D silicone, etc. The state of the fingerprint carrier can also be the physical state of the fingerprint carrier, for example: taking the fingerprint carrier as a finger, it can output the dryness of the finger skin surface, the skin condition of the finger, etc.

[0051] In one example, continuous skin condition detection (such as moisture content, stratum corneum condition, and degree of burns or scalds) can be performed based on the state of the fingerprint carrier. Accurate skin condition classification can be achieved by determining the carrier state of the fingerprint carrier multiple times over time.

[0052] In the embodiments of this application, the state of the fingerprint carrier includes at least one of whether the fingerprint carrier is a result of the binary division of a finger, the medium classification of the fingerprint carrier, and the physical state of the fingerprint carrier. Thus, fingerprint anti-counterfeiting identification or derivative operations can be performed based on the current frame fingerprint image, which can meet different user needs.

[0053] Figure 2 is a flowchart of a recurrent neural network model training method provided in an embodiment of this application. As shown in Figure 2, the recurrent neural network model can be obtained through steps 201 to 205 as shown below.

[0054] Step 201: Obtain multiple fingerprint image queues.

[0055] Multiple fingerprint image queues are acquired, each fingerprint image queue includes multiple sample fingerprint images. The sample fingerprint images included in each fingerprint image queue are obtained by fingerprint acquisition on at least one fingerprint carrier. That is, each fingerprint image queue includes fingerprint images corresponding to fingerprints carried by at least one fingerprint carrier. Different fingerprint image queues may include different numbers of sample fingerprint images and / or different numbers of sample fingerprint images corresponding to fingerprint carriers.

[0056] Step 202: Determine the timing information of each fingerprint image queue.

[0057] The timing information includes the time information corresponding to each sample fingerprint image in the fingerprint image queue. In one example, the timing information can be preset timing information. The time information corresponding to each sample fingerprint image can indicate the time corresponding to the sample fingerprint image, such as the time interval between the sample fingerprint image and the previous frame sample fingerprint image, the acquisition time corresponding to the sample fingerprint image, etc.

[0058] Step 203: Determine the carrier status information of the last sample fingerprint image in each fingerprint image queue.

[0059] Each fingerprint image queue includes multiple sample fingerprint images, which are arranged to form a fingerprint image queue. The carrier state information of the last sample fingerprint image in each fingerprint image queue is determined. It should be noted that since the recurrent neural network model outputs the carrier state information of the last sample fingerprint image in the fingerprint queue, determining the carrier state information of the last sample fingerprint image can inform the model to be trained of the correct result.

[0060] Step 204: Construct training samples for each fingerprint image queue.

[0061] Each training sample includes a fingerprint image queue, the corresponding time sequence information of the fingerprint image queue, and the carrier state information of the last sample fingerprint image in the fingerprint image queue.

[0062] Step 205: Train the model to be trained using each training sample to obtain a recurrent neural network model.

[0063] In this embodiment, multiple fingerprint image queues are acquired. Based on each fingerprint image queue, the corresponding time sequence information, and the carrier state information of the last sample fingerprint image in the queue, multiple training samples are determined. The recurrent neural network model is trained using these multiple training samples, thus realizing the model training process. Since the time sequence information corresponding to the fingerprint image queue is used as a variable input into the model during training, the trained recurrent neural network model can refer to the time condition when recognizing carrier state information and refer to historical fingerprint images when recognizing the carrier state information of the current frame fingerprint image, thereby improving the recognition accuracy of the recurrent neural network model.

[0064] In one possible implementation, when acquiring multiple fingerprint image queues, sample fingerprint images corresponding to fingerprints carried by at least two fingerprint carriers can be collected respectively to obtain at least two sample fingerprint image libraries. From the at least two sample fingerprint image libraries, multiple sample fingerprint images are extracted according to a target time interval to obtain multiple sets of sample fingerprint image queues. The sample fingerprint image queue includes multiple sample fingerprint images extracted from at least one sample fingerprint library. Each set of sample fingerprint image queues is sorted according to the extraction order to obtain multiple fingerprint image queues.

[0065] Sample fingerprint images are collected from at least two different fingerprint carriers, including but not limited to fingers, 2D printing materials, and 3D silicone materials. Sample fingerprint images can be obtained through continuous acquisition, interval acquisition, and virtual stitching. In one example, when constructing a sample fingerprint image library based on the sample fingerprint images, the images can be preprocessed. For example, preprocessing methods such as rotation, affine transformation, translation, flipping, denoising, enhancement, and position swapping can be applied to increase the diversity of sample fingerprint images included in the library. It should be noted that a sample fingerprint image library includes sample fingerprint images corresponding to only one fingerprint carrier.

[0066] The following explanation uses two fingerprint image databases as an example, that is, two fingerprint carriers as an example. Figure 3 is a schematic diagram of a fingerprint image queue construction process provided by an embodiment of this application. As shown in Figure 3, fingerprint images are constructed from at least two sample fingerprint image databases (e.g., sample fingerprint image database 1 and sample fingerprint image database 2) according to a target time interval (e.g., t1, t2 to t3 in Figure 3). n Multiple sample fingerprint images are extracted to obtain multiple sample fingerprint image queues. Each sample fingerprint image queue includes multiple sample fingerprint images extracted from at least one sample fingerprint image library. For example, queue 1 includes sample fingerprint images corresponding to one fingerprint carrier, and queue 2 includes sample fingerprint images corresponding to two fingerprint carriers. The sample fingerprint images in each group of sample fingerprint image queues are sorted according to the extraction time to obtain multiple fingerprint image queues, such as queues 1 to queue n in Figure 3.

[0067] It should be noted that the target time interval can be set manually or generated randomly. The target time interval is the timing information of the corresponding fingerprint image queue.

[0068] In this embodiment, sample fingerprint images corresponding to fingerprints carried by at least two fingerprint carriers are collected respectively, and at least two sample fingerprint image libraries are constructed. Sample fingerprint images are extracted from the at least two sample fingerprint image libraries according to a preset time sequence. The extracted sample fingerprint images are sorted according to the extraction time to obtain multiple fingerprint image queues. This realizes the construction of multiple fingerprint image queues. Since the sample fingerprint images in the fingerprint image queues are sorted according to the extraction time, the process of collecting fingerprint images in daily use can be simulated. Furthermore, since the fingerprint image queues can include sample fingerprint images corresponding to at least two fingerprint carriers, continuous real finger pressing, continuous fake fingerprint pressing, and alternating pressing of real and fake fingerprints can be simulated. Therefore, the recurrent neural network model trained based on the fingerprint image queues can have a high recognition accuracy in daily use.

[0069] Figure 4 is a flowchart of a recurrent neural network model training method provided in an embodiment of this application. As shown in Figure 4, when training the model to be trained using various training samples to obtain a recurrent neural network model, the following steps 401 to 405 can be executed:

[0070] Step 401: Input the fingerprint image queue and the corresponding time information of each training sample into the model to be trained to obtain the recognition result corresponding to each training sample.

[0071] A model to be trained is constructed by inputting the fingerprint image queues and corresponding time-series information of each training sample into the constructed model. The recognition result can indicate the carrier state of the last sample fingerprint image in the fingerprint image queue of the training sample.

[0072] Step 402: Determine the matching result corresponding to each training sample based on the recognition result corresponding to each training sample and the carrier state information of the fingerprint image of the last sample in the training sample.

[0073] Determine whether the recognition result output by the model to be trained matches the carrier state information of the fingerprint image of the last sample in the training sample. If they match, it proves that the recognition result output by the model to be trained is correct; if they do not match, it proves that the recognition result output by the model to be trained is incorrect.

[0074] Step 403: Determine whether the matching results corresponding to each training sample meet the training completion conditions. If yes, proceed to step 404; otherwise, proceed to step 405.

[0075] Determine whether the matching results for each training sample meet the training completion conditions. The training completion conditions can be set as needed. For example, the training completion conditions can be set as follows: the proportion of matching training samples is greater than a proportion threshold, the number of matching training samples exceeds a quantity threshold, etc.

[0076] Step 404: Determine the model to be trained as a recurrent neural network model.

[0077] When the matching results corresponding to each training sample meet the training completion conditions, the training is completed, and the trained model is determined as a recurrent neural network model.

[0078] Step 405: Based on the matching results corresponding to each training sample, correct the parameters of the model to be trained, and then execute step 401.

[0079] When the matching results corresponding to each training sample do not meet the training completion conditions, such as when the number of matching training samples is too low or the proportion is too low, the parameters of the model to be trained are corrected according to the matching results corresponding to each training sample. After modifying the parameters of the model, the identification is performed again until the matching results corresponding to each training sample meet the training completion conditions. It should be noted that when re-executing step 401, identification can be performed using a different fingerprint image queue than the previous identification to obtain matching results again. Identification can also be performed using the same fingerprint image queue, which is not limited here.

[0080] In this embodiment, the fingerprint image queue and the corresponding temporal information of each training sample are input into the model to be trained to obtain the recognition result of each training sample. The matching status of the recognition result with the carrier state information of the last sample fingerprint image is determined. Based on the matching status of each training sample, the parameters of the model to be trained are corrected until the matching result of each training sample meets the training completion condition, thus obtaining a recurrent neural network model. This enables the training of the recurrent neural network model. Since the temporal information of the fingerprint queue is referenced during model training, the trained recurrent neural network will refer to the temporal information when recognizing the fingerprint image of the current frame. Furthermore, since the training ends when the matching result of each training sample meets the training completion condition, the accuracy of the recurrent neural network model can be guaranteed. Therefore, the recurrent neural network model trained based on the fingerprint image queue can have a high recognition accuracy in daily use.

[0081] In one possible implementation, when inputting the fingerprint image queues and corresponding time-series information of each training sample into the model to be trained to obtain the recognition results for each training sample, the following operation can be performed on the fingerprint image queues in each training sample:

[0082] To obtain the first variable, input the initial variable, the fingerprint image of the first sample in the fingerprint image queue, and the time information corresponding to the first sample fingerprint image in the fingerprint image queue into the model to be trained. To obtain the first variable, input the (n-1)th variable, the fingerprint image of the nth sample in the fingerprint image queue, and the time information corresponding to the nth sample fingerprint image into the model to be trained. To obtain the nth variable, input the (m-1)th variable, the last sample fingerprint image in the fingerprint image queue, and the time information corresponding to the last sample fingerprint image into the model to be trained. To obtain the recognition result corresponding to the fingerprint image queue, where m is the number of sample fingerprint images included in the fingerprint image queue, and n is an integer less than or equal to m-1 and greater than or equal to 2.

[0083] Since the fingerprint image ranked first in the fingerprint image queue is the first fingerprint image input to the model to be trained, the latent variable corresponding to the fingerprint image ranked first is the initial variable. The initial variable can be set manually. After inputting the initial variable, the fingerprint image ranked first in the fingerprint image queue, and the time information corresponding to the fingerprint image ranked first in the fingerprint image queue into the model to be trained, the training model will output the recognition result corresponding to the fingerprint image ranked first and the first variable updated with the initial variable.

[0084] Input the (n-1)th variable, the nth ranked sample fingerprint image in the fingerprint image queue, and the time information corresponding to the nth ranked sample fingerprint image in the fingerprint image queue into the model to be trained. This will obtain the recognition result corresponding to the nth sample fingerprint image output by the model to be trained and the nth variable after updating the (n-1)th variable. n is an integer less than or equal to m-1 and greater than or equal to 2, and m is the number of sample fingerprint images included in the fingerprint image queue. For example, when n=2, input the second ranked sample fingerprint image of the first variable and the time information corresponding to the second ranked sample fingerprint image into the model to be trained to obtain the second variable. When n=5, input the fourth variable, the fifth ranked sample fingerprint image, and the time information corresponding to the fifth ranked sample fingerprint image into the model to be trained to obtain the fifth variable.

[0085] Input the (m-1)th variable, the last sample fingerprint image in the fingerprint image queue, and the time information corresponding to the last sample fingerprint image in the fingerprint image queue into the model to be trained to obtain the recognition result corresponding to the fingerprint image queue, and the mth variable obtained after updating the m-1 variable, where m is the number of sample fingerprint images in the fingerprint image queue, thereby obtaining the recognition result corresponding to the last sample fingerprint image in the fingerprint image queue.

[0086] In this embodiment, for each training sample, the fingerprint images of each sample in the fingerprint queue, the time information corresponding to each sample fingerprint image, and the latent variables updated by the model to be trained based on the information corresponding to the previous sample fingerprint image are sequentially input into the model to be trained. This obtains the recognition result of the last sample fingerprint image in each fingerprint queue. Since the sample fingerprint images in the fingerprint queue are sorted according to the extraction time, the carrier state information of the current frame fingerprint image can be determined based on the latent variables corresponding to the previous frame fingerprint image during daily use, thereby simulating daily use.

[0087] In one possible implementation, a recurrent neural network model can be obtained by acquiring model parameters and then correcting the parameters in the model to be trained based on those parameters.

[0088] When obtaining a recurrent neural network model, model parameters can also be set based on experience. For example, model parameters can be manually input, and the model parameters of the constructed training model can be corrected by manually inputting model parameters to obtain a recurrent neural network model.

[0089] In this embodiment, model parameters are obtained, and the parameters in the model to be trained are corrected according to the model parameters to obtain a recurrent neural network model. This realizes the acquisition of the recurrent neural network model. Since there is no need to train the model to be trained, the time required to obtain the recurrent neural network by manually inputting model parameters is short, which can improve the efficiency of model acquisition.

[0090] Figure 5 is a schematic diagram of a fingerprint carrier status recognition device provided in an embodiment of this application. As shown in Figure 5, the device 500 includes:

[0091] Acquisition unit 501 is used to acquire the fingerprint image of the current frame;

[0092] The input unit 502 is used to input the feature information of the current frame fingerprint image, the time information corresponding to the current frame fingerprint image, and the latent variables corresponding to the current frame fingerprint image into the recurrent neural network model to obtain the carrier state information and the latent variables corresponding to the next frame fingerprint image output by the recurrent neural network model. The fingerprint carrier is the medium that carries the fingerprint corresponding to the fingerprint image.

[0093] The determining unit 503 is used to determine the status of the fingerprint carrier based on the carrier status information.

[0094] In this embodiment of the application, the acquisition unit 501 can be used to execute step 101 in the above method embodiment, the input unit 502 can be used to execute step 102 in the above method embodiment, and the determination unit 503 can be used to execute step 103 in the above method embodiment.

[0095] In one possible implementation, the feature information of the current frame fingerprint image includes at least one of the fingerprint signal output by the fingerprint sensor when the current frame fingerprint image is acquired, the current frame fingerprint image, and fingerprint features extracted from the current frame fingerprint image.

[0096] In one possible implementation, the time information corresponding to the current frame fingerprint image includes the time when the current frame fingerprint image was acquired, and / or the time interval between acquiring the current frame fingerprint image and acquiring the previous frame fingerprint image.

[0097] In one possible implementation, the state of the fingerprint carrier includes at least one of the following: whether the fingerprint carrier is a result of the bifurcation of a finger, the medium classification of the fingerprint carrier, and the physical state of the fingerprint carrier.

[0098] In one possible implementation, the recurrent neural network model is obtained by: acquiring multiple fingerprint image queues, wherein each fingerprint image queue includes multiple sample fingerprint images, and the sample fingerprint images included in each fingerprint image queue are obtained by fingerprint acquisition from at least one fingerprint carrier; determining the temporal information of each fingerprint image queue, wherein the temporal information includes the time information corresponding to each sample fingerprint image in the fingerprint image queue; determining the carrier state information of the last sample fingerprint image in each fingerprint image queue; constructing training samples corresponding to each fingerprint image queue, wherein the training samples include the fingerprint image queue, the temporal information corresponding to the fingerprint image queue, and the carrier state information of the last sample fingerprint image in the fingerprint image queue; and training the model to be trained using each training sample to obtain the recurrent neural network model.

[0099] In one possible implementation, acquiring multiple fingerprint image queues includes: acquiring sample fingerprint images corresponding to fingerprints carried by at least two fingerprint carriers to obtain at least two sample fingerprint image libraries; extracting multiple sample fingerprint images from the at least two sample fingerprint image libraries according to a target time interval to obtain multiple sets of sample fingerprint image queues, wherein each sample fingerprint image queue includes multiple sample fingerprint images extracted from at least one sample fingerprint library; and sorting each set of sample fingerprint image queues according to the extraction order to obtain multiple fingerprint image queues.

[0100] In one possible implementation, a recurrent neural network model is obtained by training the model to be trained using each training sample. This includes: inputting the fingerprint image queue and corresponding time-series information of each training sample into the model to be trained to obtain the recognition result for each training sample, wherein the recognition result indicates the carrier state of the last fingerprint image in the fingerprint image queue of that training sample; determining the matching result for each training sample based on the recognition result and the carrier state information of the last fingerprint image in that training sample; and correcting the parameters of the model to be trained based on the matching results for each training sample until the matching results for each training sample meet the training completion conditions, thereby obtaining the recurrent neural network model.

[0101] In one possible implementation, the fingerprint image queues and their corresponding time-series information included in each training sample are input into the model to be trained to obtain the recognition result corresponding to each training sample. This includes performing the following operations on the fingerprint image queues in each training sample: obtaining initial variables; inputting the initial variables, the fingerprint image of the first sample in the fingerprint image queue, and the time information corresponding to the first sample in the fingerprint image queue into the model to be trained to obtain the first variable; inputting the (n-1)th variable, the fingerprint image of the nth sample in the fingerprint image queue, and the time information corresponding to the nth sample in the fingerprint image queue into the model to be trained to obtain the nth variable; inputting the (m-1)th variable, the fingerprint image of the last sample in the fingerprint image queue, and the time information corresponding to the last sample in the fingerprint image queue into the model to be trained to obtain the recognition result corresponding to the fingerprint image queue, where m is the number of fingerprint images included in the fingerprint image queue, and n is an integer less than or equal to m-1 and greater than or equal to 2.

[0102] In one possible implementation, the recurrent neural network model is obtained by: acquiring the model parameters; and correcting the parameters in the model to be trained based on the model parameters to obtain the recurrent neural network model.

[0103] Referring to FIG6, a schematic diagram of the structure of an electronic device according to an embodiment of the present application is shown. The specific embodiments of the present application do not limit the specific implementation of the electronic device.

[0104] As shown in Figure 6, the electronic device may include: a processor 602, a communications interface 604, a memory 606, and a communications bus 608.

[0105] in:

[0106] The processor 602, communication interface 604, and memory 606 communicate with each other via communication bus 608.

[0107] Communication interface 604 is used for communication with other electronic devices or servers.

[0108] The processor 602 is used to execute program 610, which can specifically execute the relevant steps in the above-described fingerprint carrier state recognition method embodiment.

[0109] Specifically, program 610 may include program code that includes computer operation instructions.

[0110] Processor 602 may be a central processing unit (CPU), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs; one or more GPUs; or they may be processors of different types, such as one or more CPUs, one or more GPUs, and one or more ASICs.

[0111] Memory 606 is used to store program 610. Memory 606 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0112] Specifically, program 610 can be used to cause processor 602 to execute the fingerprint carrier status recognition method in any of the foregoing embodiments.

[0113] The specific implementation of each step in program 610 can be found in the corresponding steps and units described in any of the foregoing embodiments of the fingerprint carrier state recognition method, and will not be repeated here. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the devices and modules described above can be referred to the corresponding process descriptions in the foregoing method embodiments, and will not be repeated here.

[0114] In this embodiment, the current frame fingerprint image is acquired, and its feature information, corresponding time information, and latent variables are input into a recurrent neural network model to determine the carrier state information corresponding to the current frame fingerprint image. This allows for the determination of the fingerprint carrier state, enabling fingerprint anti-counterfeiting and derivation operations. Since the input to the recurrent neural network model includes the latent variables of the current frame fingerprint image, and these latent variables are determined based on historical fingerprint images, historical fingerprint images and the acquisition time of the current frame fingerprint image are referenced during carrier state identification. Compared to existing anti-counterfeiting identification schemes using a single frame fingerprint image, this approach improves accuracy because it doesn't rely solely on a single frame. Furthermore, compared to existing anti-counterfeiting identification using multiple frames of fingerprint images, this approach, using a recurrent neural network model for current frame fingerprint image identification, consumes less time and improves user experience.

[0115] This application also provides a computer program product, including computer instructions that instruct a computing device to perform an operation corresponding to any of the methods in the above-described multiple method embodiments.

[0116] It should be noted that, depending on the implementation needs, the various components / steps described in the embodiments of this application can be broken down into more components / steps, or two or more components / steps or parts of the operation of components / steps can be combined into new components / steps to achieve the purpose of the embodiments of this application.

[0117] The methods described in the embodiments of this application can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD-ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or as computer code downloaded over a network that is originally stored in a remote recording medium or a non-transitory machine-readable medium and will be stored in a local recording medium. Thus, the methods described herein can be stored as software processing on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components (e.g., RAM, ROM, flash memory, etc.) capable of storing or receiving software or computer code that, when accessed and executed by the computer, processor, or hardware, implements the fingerprint carrier state recognition method described herein. Furthermore, when a general-purpose computer accesses code used to implement the fingerprint carrier state recognition method shown herein, the execution of the code transforms the general-purpose computer into a dedicated computer for executing the fingerprint carrier state recognition method shown herein.

[0118] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this application.

[0119] The above embodiments are only used to illustrate the embodiments of this application, and are not intended to limit the embodiments of this application. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the embodiments of this application. Therefore, all equivalent technical solutions also fall within the scope of the embodiments of this application, and the patent protection scope of the embodiments of this application should be defined by the claims.

Claims

1. A method for recognizing the state of a fingerprint carrier, characterized in that, include: Obtain the fingerprint image of the current frame; The feature information of the current frame fingerprint image, the time information corresponding to the current frame fingerprint image, and the latent variables corresponding to the current frame fingerprint image are input into the recurrent neural network model to obtain the carrier state information and the latent variables corresponding to the next frame fingerprint image output by the recurrent neural network model. The fingerprint carrier is the medium that carries the fingerprint corresponding to the fingerprint image. The status of the fingerprint carrier is determined based on the carrier status information.

2. The method according to claim 1, characterized in that, The feature information of the current frame fingerprint image includes at least one of the fingerprint signal output by the fingerprint sensor when the current frame fingerprint image is acquired, the current frame fingerprint image, and fingerprint features extracted from the current frame fingerprint image.

3. The method according to claim 1, characterized in that, The time information corresponding to the current frame fingerprint image includes the time when the current frame fingerprint image was acquired, and / or the time interval between acquiring the current frame fingerprint image and acquiring the previous frame fingerprint image.

4. The method according to claim 1, characterized in that, The state of the fingerprint carrier includes at least one of the following: whether the fingerprint carrier is a result of the differentiation of a finger, the medium classification of the fingerprint carrier, and the physical state of the fingerprint carrier.

5. The method according to claim 1, characterized in that, The recurrent neural network model was obtained through the following method: Multiple fingerprint image queues are acquired, wherein each fingerprint image queue includes multiple sample fingerprint images, and the sample fingerprint images included in each fingerprint image queue are obtained by fingerprint acquisition from at least one fingerprint carrier; Determine the timing information for each of the fingerprint image queues, wherein the timing information includes the time information corresponding to each sample fingerprint image in the fingerprint image queue; Determine the carrier state information of the last sample fingerprint image in each fingerprint image queue; Construct training samples corresponding to each fingerprint image queue, wherein the training samples include the fingerprint image queue, the temporal information corresponding to the fingerprint image queue, and the carrier state information of the last sample fingerprint image in the fingerprint image queue; The recurrent neural network model is obtained by training the model to be trained using the training samples.

6. The method according to claim 5, characterized in that, The acquisition of multiple fingerprint image queues includes: Collect sample fingerprint images corresponding to fingerprints carried by at least two fingerprint carriers to obtain at least two sample fingerprint image libraries; Multiple sample fingerprint images are extracted from the at least two sample fingerprint image databases according to a target time interval to obtain multiple sample fingerprint image queues, wherein the sample fingerprint image queues include multiple sample fingerprint images extracted from at least one sample fingerprint database. The fingerprint image queues of each group of samples are sorted according to the extraction order to obtain multiple fingerprint image queues.

7. The method according to claim 5, characterized in that, The step of training the model to be trained using each of the training samples to obtain the recurrent neural network model includes: The fingerprint image queues and corresponding time-series information of each training sample are input into the model to be trained to obtain the recognition results corresponding to each training sample. The recognition results are used to indicate the carrier state of the fingerprint image of the last sample in the fingerprint image queue of the training sample. Based on the recognition result corresponding to each training sample and the carrier state information of the fingerprint image of the last sample in the training sample, the matching result corresponding to the training sample is determined. Based on the matching results corresponding to each training sample, the parameters of the model to be trained are corrected until the matching results corresponding to each training sample meet the training completion conditions, thereby obtaining the recurrent neural network model.

8. The method according to claim 7, characterized in that, The step of inputting the fingerprint image queues and corresponding time-series information of each training sample into the model to be trained to obtain the recognition results corresponding to each training sample includes: For each of the fingerprint image queues in the training samples, the following operations are performed: Get the initial variables; The initial variable, the fingerprint image of the first sample in the fingerprint image queue, and the time information corresponding to the fingerprint image of the first sample in the fingerprint image queue are input into the model to be trained to obtain the first variable; The (n-1)th variable, the nth sorted sample fingerprint image in the fingerprint image queue, and the time information corresponding to the nth sorted sample fingerprint image in the fingerprint image queue are input into the model to be trained to obtain the nth variable; The (m-1)th variable, the last sample fingerprint image in the fingerprint image queue, and the time information corresponding to the last sample fingerprint image in the fingerprint image queue are input into the model to be trained to obtain the recognition result corresponding to the fingerprint image queue. Here, m is the number of sample fingerprint images included in the fingerprint image queue, and n is an integer less than or equal to m-1 and greater than or equal to 2.

9. The method according to claim 1, characterized in that, The recurrent neural network model was obtained through the following method: Obtain model parameters; The recurrent neural network model is obtained by correcting the parameters in the model to be trained based on the model parameters.

10. A fingerprint carrier status recognition device, characterized in that, include: The acquisition unit is used to acquire the fingerprint image of the current frame; The input unit is used to input the feature information of the current frame fingerprint image, the time information corresponding to the current frame fingerprint image, and the latent variables corresponding to the current frame fingerprint image into the recurrent neural network model to obtain the carrier state information and the latent variables corresponding to the next frame fingerprint image output by the recurrent neural network model, wherein the fingerprint carrier is the medium that carries the fingerprint corresponding to the fingerprint image; A determining unit is used to determine the state of the fingerprint carrier based on the carrier state information.

11. An electronic device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the fingerprint carrier state recognition method as described in any one of claims 1-9.

12. A computer storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the fingerprint carrier state recognition method as described in any one of claims 1-9.

Images