A fingerprint image matching method, device, medium, and program product

Abstract

This application provides a fingerprint image matching method, device, medium, and program product, relating to the field of image processing technology, aiming to quickly and accurately match fingerprint images. The method includes: acquiring a fingerprint image to be identified, and reading a complete fingerprint image and its location information; inputting the fingerprint image to be identified and the complete fingerprint image into a fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified within the complete fingerprint image; determining a transformation matrix based on the location information of the complete fingerprint image and the predicted location information; transforming a first original fingerprint image according to the transformation matrix to generate a transformed image; matching the transformed image with a second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image; wherein, one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a fingerprint image matching method, device, medium, and program product. Background Technology

[0002] During the fingerprint recognition process, the collected fingerprint image needs to be matched with the complete fingerprint image to identify whether the collected fingerprint image is part of the complete fingerprint image and to perform finger biometric verification.

[0003] The fingerprint recognition methods provided by related technologies first require extracting feature points from both the complete fingerprint image and the fingerprint image acquired by the sensor. Then, based on the feature points of the complete fingerprint image and the acquired fingerprint image, an affine transformation matrix is ​​estimated. Finally, fingerprint recognition is performed based on the estimated affine transformation matrix. However, this method has drawbacks such as long feature point extraction and matching time, redundant components in the estimated affine transformation matrix, and low accuracy. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention provide a fingerprint image matching method, device, medium, and program product to overcome or at least partially solve the above problems.

[0005] A first aspect of the present invention provides a fingerprint image matching method, the method comprising:

[0006] Acquire a fingerprint image to be identified, and read the complete fingerprint image and the location information of the complete fingerprint image, wherein the area of ​​the fingerprint image to be identified is smaller than the area of ​​the complete fingerprint image;

[0007] The fingerprint image to be identified and the complete fingerprint image are input into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the complete fingerprint image;

[0008] Based on the location information of the complete fingerprint image and the predicted location information, a transformation matrix is ​​determined. The transformation matrix represents the translation and / or rotation relationship between the fingerprint image to be identified and the image corresponding to the fingerprint image to be identified in the complete fingerprint image.

[0009] The first original fingerprint image is transformed according to the transformation matrix to generate a transformed image;

[0010] The transformed image is matched with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image;

[0011] Wherein, one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image.

[0012] Optionally, obtaining the predicted location information of the fingerprint image to be identified within the complete fingerprint image includes:

[0013] The coordinates of the four corner points of the fingerprint image to be identified are obtained in the coordinate system of the complete fingerprint image.

[0014] Optionally, if multiple complete fingerprint images are read, the location information of each complete fingerprint image is read;

[0015] The fingerprint image to be identified and the complete fingerprint image are input into a fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified within the complete fingerprint image, including:

[0016] The fingerprint image processing model inputs the fingerprint image to be identified and each complete fingerprint image into a fingerprint image pair to obtain the predicted position information of the fingerprint image to be identified in each complete fingerprint image. Multiple transformation matrices are generated based on the predicted position information and the corresponding position information of the complete fingerprint image, thereby obtaining the matching degree between the fingerprint image to be identified and each complete fingerprint image.

[0017] Optionally, if multiple complete fingerprint images are read, the location information of each complete fingerprint image is read;

[0018] The fingerprint image to be identified and the complete fingerprint image are input into a fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified within the complete fingerprint image, including:

[0019] Take any one of the multiple complete fingerprint images as the current complete fingerprint image and perform the following steps:

[0020] The current complete fingerprint image and the fingerprint image to be identified are input into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the current complete fingerprint image. A transformation matrix is ​​generated based on the predicted position information and the position information of the current complete fingerprint image, thereby obtaining the matching degree between the fingerprint image to be identified and the current complete fingerprint image.

[0021] When the matching degree between the fingerprint image to be identified and the current complete fingerprint image is lower than a preset threshold, another complete fingerprint image among the multiple complete fingerprint images is taken as the current complete fingerprint image, and the process returns to the step of inputting the current complete fingerprint image and the fingerprint image to be identified into the fingerprint image processing model, until the multiple complete fingerprint images are traversed, or until the matching degree between the fingerprint image to be identified and any complete fingerprint image is higher than the preset threshold.

[0022] Optionally, the method further includes:

[0023] Obtain a complete fingerprint image sample;

[0024] The complete fingerprint image sample is translated by multiple distances and / or rotated by multiple angles;

[0025] Extract multiple sub-image samples of a target size from a complete fingerprint image sample after translation and / or rotation;

[0026] The target model is trained using the complete fingerprint image sample and multiple sub-image samples carrying location labels to obtain the fingerprint image processing model, wherein the location labels represent the location information of the sub-image samples in the complete fingerprint image sample.

[0027] Optionally, the fingerprint image processing model is obtained by training the target model using the complete fingerprint image sample and multiple sub-image samples carrying location labels, including:

[0028] The complete fingerprint image sample and multiple sub-image samples carrying location tags are input into the target model to obtain the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample.

[0029] The mean squared error loss function value is determined based on the difference between the location labels carried by each of the multiple sub-image samples and the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample.

[0030] With the goal of minimizing the difference, the model parameters of the target model are updated by stochastic gradient based on the mean squared error loss function value to obtain the fingerprint image processing model.

[0031] Optionally, after matching the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image, the method further includes:

[0032] When the matching degree between the fingerprint image to be identified and the complete fingerprint image is higher than a preset threshold, an unlock command is output.

[0033] In a second aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the fingerprint image matching method as described in the embodiments of this application.

[0034] A third aspect of the present invention provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the fingerprint image matching method as described in the embodiments of this application.

[0035] A fourth aspect of the present invention provides a computer program product, including a computer program / instructions, characterized in that, when the computer program / instructions are executed by a processor, they implement the fingerprint image matching method as described in the embodiments of this application.

[0036] A fifth aspect of the present invention provides a fingerprint image matching device, the device comprising:

[0037] A fingerprint image acquisition module is used to acquire a fingerprint image to be identified and to read a complete fingerprint image and the location information of the complete fingerprint image, wherein the area of ​​the fingerprint image to be identified is smaller than the area of ​​the complete fingerprint image;

[0038] The location information acquisition module is used to input the fingerprint image to be identified and the complete fingerprint image into the fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified in the complete fingerprint image;

[0039] The transformation matrix determination module is used to determine the transformation matrix based on the position information of the complete fingerprint image and the predicted position information. The transformation matrix represents the translation and / or rotation relationship between the fingerprint image to be identified and the image corresponding to the fingerprint image to be identified in the complete fingerprint image.

[0040] The transformed image generation module is used to transform the first original fingerprint image according to the transformation matrix to generate a transformed image;

[0041] A fingerprint image matching module is used to match the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image; wherein, one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image.

[0042] The embodiments of the present invention have the following advantages:

[0043] In this embodiment, the fingerprint image to be identified and the complete fingerprint image are directly input into the fingerprint image processing model, eliminating the need to extract and match feature points of the fingerprint image to be identified and the complete fingerprint image, thus saving time. The fingerprint image processing model can determine the predicted position information of the fingerprint image to be identified in the complete fingerprint image, and thus determine the transformation matrix based on the predicted position information and the position information of the complete fingerprint image. Compared with the related technology that estimates the transformation matrix based on the matching of feature points, the transformation matrix determined by this application based on the position information is more accurate. Considering the significant size difference between the fingerprint image to be identified and the complete fingerprint image, this application uses the transformation matrix to transform the first original fingerprint image to generate a transformed image with a size similar to the second original image. Compared with directly matching the complete fingerprint image and the fingerprint image to be identified, which have a large size difference, matching the transformed image and the second original image with a similar size can effectively improve the accuracy of matching the fingerprint image to be identified, especially when the size of the fingerprint image to be identified is small, it can significantly improve the matching accuracy. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart of the steps of a fingerprint image matching method according to an embodiment of the present invention;

[0046] Figure 2 This is an example diagram of a complete fingerprint image, a transformed image, and a fingerprint image to be identified in an embodiment of the present invention;

[0047] Figure 3 This is an example diagram of a complete fingerprint image sample and corresponding sub-image samples in an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of information interaction in a fingerprint image matching device according to an embodiment of the present invention;

[0049] Figure 5 This is a schematic diagram of the structure of a fingerprint image matching device according to an embodiment of the present invention. Detailed Implementation

[0050] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] To address the issues of time-consuming fingerprint image matching methods, redundant components in the estimated affine transformation matrix, and low accuracy in related technologies, the applicant proposes the following: An end-to-end method is used to determine the predicted location information of the fingerprint image to be identified within the complete fingerprint image in a single inference step. Based on this predicted location information and the location information of the complete fingerprint image, a transformation matrix is ​​determined. The first original image is then transformed using the transformation matrix to obtain a transformed image. This transformed image is then matched with a second original image, where one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image.

[0052] Reference Figure 1 As shown, a flowchart illustrating the steps of a fingerprint image matching method according to an embodiment of the present invention is presented, as follows: Figure 1 As shown, this fingerprint image matching method can be used on electronic devices such as electronic locks, computers, mobile phones, and tablets. It can also be used to unlock application clients installed on electronic devices that require authentication, such as file vaults and diary clients. The following explanation uses the fingerprint image matching method to unlock electronic devices as an example. This fingerprint image matching method includes the following steps:

[0053] Step S11: Obtain the fingerprint image to be identified, and read the complete fingerprint image and the location information of the complete fingerprint image, wherein the area of ​​the fingerprint image to be identified is smaller than the area of ​​the complete fingerprint image.

[0054] The fingerprint image to be identified is a fingerprint image obtained by capturing an image of the finger in the acquisition area. The acquisition area is above the sensor, such as part of a mobile phone screen, or the metal layer above the sensor. The sensor can be integrated into the electronic device or independent of the electronic device, and can communicate with the electronic device to transmit the fingerprint image to be identified. When acquiring the fingerprint image to be identified, due to factors such as the shape of the sensor, the rotation angle of the finger, and the position of the finger, the acquired fingerprint image may be a partial fingerprint image with rotation angle and displacement distance. The displacement distance refers to the distance between the corresponding point on the finger in the partial fingerprint image and the complete fingerprint image.

[0055] A complete fingerprint image refers to a registered fingerprint image that can unlock an electronic device. A complete fingerprint image can be obtained by fusing multiple partial fingerprint images captured during registration, or by updating the registered complete fingerprint image based on the recognized fingerprint image during use. Generally, the area of ​​the fingerprint image to be identified is smaller than the area of ​​the complete fingerprint image. A coordinate system can be established with a point in the fingerprint image acquisition area as the origin, and two mutually perpendicular directions as the x-axis and y-axis. The positional information of the complete fingerprint image can be represented by four coordinates in this coordinate system, corresponding to the positional information of the four corner points of the complete fingerprint image.

[0056] Once the fingerprint image to be identified is a partial fingerprint image corresponding to a complete fingerprint image, and after biometric verification, the electronic device can be successfully unlocked. The complete fingerprint image can be stored in the electronic device or in the corresponding fingerprint recognition server; this invention does not limit this.

[0057] Step S12: Input the fingerprint image to be identified and the complete fingerprint image into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the complete fingerprint image.

[0058] This application proposes an end-to-end method to determine the predicted location information of the fingerprint image to be identified within the complete fingerprint image through a single inference process. This predicted location information represents the position of the mapped image of the fingerprint image to be identified relative to the complete fingerprint image. The end-to-end method is implemented through a fingerprint image processing model. Specifically, by inputting the fingerprint image to be identified and the complete fingerprint image into the fingerprint image processing model, the predicted location information of the fingerprint image to be identified within the complete fingerprint image can be obtained from the model's output. The fingerprint image processing model typically refers to a neural network model.

[0059] Optionally, obtaining the predicted position information of the fingerprint image to be identified in the complete fingerprint image includes: obtaining the coordinates of the four corner points of the fingerprint image to be identified in the coordinate system of the complete fingerprint image.

[0060] Both the fingerprint image to be identified and the complete fingerprint image can be acquired through the same acquisition area. The positional information of the complete fingerprint image can be represented by coordinates in the coordinate system of the acquisition area. Therefore, the positional information of the four corner coordinates of the fingerprint image to be identified can be represented by four coordinates in this coordinate system. The four corner coordinates of the fingerprint image to be identified and the complete fingerprint image can be the coordinates of the upper left corner, the upper right corner, the lower left corner, and the lower right corner. Multiple corner coordinates included in the positional information of the fingerprint image to be identified and the complete fingerprint image can be recorded in a preset order, for example, the coordinates of the upper left corner, the upper right corner, the lower right corner, and the lower left corner can be recorded in a clockwise order. Each corner coordinate can be represented by two floating-point numbers.

[0061] After obtaining the position information of the fingerprint image to be identified in the coordinate system of the complete fingerprint image, the predicted position information of the fingerprint image to be identified in the complete fingerprint image can be obtained based on the position information of the fingerprint image to be identified and the complete fingerprint image in the coordinate system.

[0062] Optionally, the positional information of the fingerprint image to be identified and the complete fingerprint image in the coordinate system can be represented by only two diagonal coordinates (e.g., the coordinates of the upper left corner and the lower right corner).

[0063] The method for training the fingerprint image processing model will be described in detail later.

[0064] Step S13: Determine the transformation matrix based on the position information of the complete fingerprint image and the predicted position information. The transformation matrix represents the translation and / or rotation relationship between the fingerprint image to be identified and the image corresponding to the fingerprint image in the complete fingerprint image.

[0065] In fingerprint image matching, the fingerprint image to be identified only involves rotation, displacement, and truncation relative to the complete fingerprint image, without stretching or distortion. Therefore, after obtaining the predicted position information of the fingerprint image to be identified within the complete fingerprint image, the transformation matrix can be determined by combining the position information of the complete fingerprint image. This transformation matrix can be an isometric transformation matrix (including translation and / or rotation transformations). The transformation matrix represents the translation and / or rotation relationship between the fingerprint image to be identified and its corresponding image within the complete fingerprint image. The method for determining the transformation matrix based on the position information of the complete fingerprint image and the predicted position information can refer to the methods for determining transformation matrices in plane linear algebra.

[0066] Step S14: Transform the first original fingerprint image according to the transformation matrix to generate a transformed image.

[0067] Step S15: Match the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image; wherein, one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image.

[0068] Because the fingerprint image to be identified only corresponds to a partial fingerprint image of the complete fingerprint image, directly matching the fingerprint image to be identified with the complete fingerprint image has the disadvantages of high computational cost and inaccurate matching results. Instead, the complete fingerprint image can be transformed according to a transformation matrix to obtain a transformed image that is approximately the same size and angle as the fingerprint image to be identified. This transformed image can then be matched with the fingerprint image to be identified to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image.

[0069] Alternatively, the fingerprint image to be identified can be transformed according to the transformation matrix to obtain a transformed image that is approximately the same size and angle as the complete fingerprint image. This transformed image is then matched with the complete fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image. However, since the transformed image obtained by transforming the fingerprint image to be identified will only be partially similar to the complete fingerprint image, when matching the transformed image with the complete fingerprint image, only the matching degree of the regions in the transformed image and the complete fingerprint image corresponding to the transformed image is calculated as the matching degree between the fingerprint image to be identified and the complete fingerprint image.

[0070] Figure 2 An example image is shown, comprising a complete fingerprint image, a transformed image, and a fingerprint image to be identified. The fingerprint image on the left is the complete fingerprint image, the middle fingerprint image is the transformed image, and the fingerprint image on the right is the fingerprint image to be identified. The transformed image is generated by transforming the complete fingerprint image based on its positional information within the complete fingerprint image, obtaining a transformation matrix, and then transforming the complete fingerprint image. Combined with... Figure 2 It can be seen that the transformed image of the complete fingerprint image generated by the transformation matrix is ​​similar in size and rotation angle to the fingerprint image to be identified. Therefore, after generating the transformed image of the complete fingerprint image by the transformation matrix, matching the transformed image with the fingerprint image to be identified can effectively reduce the amount of computation and improve the accuracy of the matching results.

[0071] After determining the transformation matrix, multiplying the complete fingerprint image by the transformation matrix yields the transformed image. The matching degree between the transformed image and the fingerprint image to be identified can then be determined using methods such as Hamming distance or the exponential method.

[0072] The technical solution of this application directly inputs the fingerprint image to be identified and the complete fingerprint image into the fingerprint image processing model, eliminating the need to extract and match feature points of the fingerprint image to be identified and the complete fingerprint image, thus saving time. The fingerprint image processing model can determine the predicted position information of the fingerprint image to be identified in the complete fingerprint image, and thus determine the transformation matrix based on the predicted position information and the position information of the complete fingerprint image. Compared with the related technology that estimates the transformation matrix based on the matching of feature points, the transformation matrix determined by this application based on the position information is more accurate. Considering the significant size difference between the fingerprint image to be identified and the complete fingerprint image, this application uses the transformation matrix to transform the first original fingerprint image to generate a transformed image with a size similar to the second original image. Compared with directly matching the complete fingerprint image and the fingerprint image to be identified, which have a large size difference, matching the transformed image and the second original image with a similar size can effectively improve the accuracy of matching the fingerprint image to be identified, especially when the size of the fingerprint image to be identified is small, it can significantly improve the matching accuracy.

[0073] Optionally, as an embodiment, after matching the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image, the method further includes: when the matching degree between the fingerprint image to be identified and the complete fingerprint image is higher than a preset threshold, outputting an unlock command.

[0074] The preset threshold is a value that indicates whether an unlock command can be output. An unlock command is output when the matching degree between the fingerprint image to be recognized and the complete fingerprint image is higher than the preset threshold; no unlock command is output when the matching degree between the fingerprint image to be recognized and the complete fingerprint image is lower than the preset threshold.

[0075] Optionally, as an embodiment, the method further includes the step of training a fingerprint image processing model: acquiring a complete fingerprint image sample; translating the complete fingerprint image sample by multiple distances and / or rotating it by multiple angles; extracting multiple sub-image samples of a target size from the translated and / or rotated complete fingerprint image sample; training a target model using the complete fingerprint image sample and the multiple sub-image samples carrying location labels to obtain the fingerprint image processing model, wherein the location labels characterize the position information of the sub-image samples in the complete fingerprint image sample.

[0076] To obtain a well-trained fingerprint image processing model, training samples need to be constructed, using complete fingerprint image samples and sub-image samples generated based on the complete fingerprint image samples as training samples.

[0077] When a user attempts to unlock the device with their finger, there may be rotation angles, translation distances, and translation directions between the fingerprint image to be identified captured by the sensor and the stored complete fingerprint image. Furthermore, the fingerprint image captured by the sensor is only a portion of the complete fingerprint image. Therefore, when constructing training samples, a complete fingerprint image sample is first obtained; this complete fingerprint image sample is the actual fingerprint image captured. Then, the complete fingerprint image sample is translated in various directions and distances, and / or rotated at various angles. Finally, sub-images of the target size are cropped from the translated and / or rotated complete fingerprint image sample, resulting in multiple sub-image samples. Figure 3 An example image is shown, depicting a complete fingerprint image sample and its corresponding sub-image sample, with the complete fingerprint image sample on the left and the corresponding sub-image sample on the right.

[0078] The target size can be the size of a rectangular area within the sensor's acquisition region, or it can be any preset size. When the target size is arbitrary, a fingerprint image processing model suitable for different sensors can be trained.

[0079] During the cropping process, the translation distance and direction, rotation angle, and target size corresponding to each sub-image sample are recorded. Then, the position information of the sub-image sample in the complete fingerprint image is determined based on planar linear algebra. The position label of the sub-image sample is then generated based on the position information. The position label represents the position information of the sub-image sample in the complete fingerprint image sample.

[0080] In practice, the above translation, rotation, and cropping operations can be automatically executed by the program to generate multiple sub-image samples of the complete fingerprint image sample. The program can also automatically record the values ​​processed at each step and automatically label the location of the sub-image samples. Alternatively, the location label can be directly calculated manually for each sub-image sample.

[0081] The fingerprint image processing model is obtained by using complete fingerprint image samples and multiple sub-image samples carrying location labels as training samples and inputting the training samples into the target model.

[0082] Optionally, as an embodiment, the complete fingerprint image sample and multiple sub-image samples carrying location tags are input into the target model to obtain the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample; the mean squared error loss function value is determined based on the difference between the location tags carried by each of the multiple sub-image samples and the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample; with the goal of minimizing the difference, the model parameters of the preset model are updated by stochastic gradient based on the mean squared error loss function value to obtain the fingerprint image processing model.

[0083] The complete fingerprint image sample and multiple sub-image samples carrying location labels are input into the target model, and the target model outputs the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample.

[0084] The mean squared error loss function value is determined based on the difference between the location labels carried by each of the multiple sub-image samples and the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample.

[0085] With the goal of minimizing the discrepancy, the model parameters of the target model are updated using stochastic gradients based on the mean squared error loss function value to obtain the fingerprint image processing model. The trained fingerprint image processing model outputs more accurate location information for both the fingerprint image to be identified and the complete fingerprint image.

[0086] The technical solution of this application embodiment obtains a fingerprint image processing model by training complete fingerprint image samples and sub-image samples. The fingerprint image processing model transforms the fingerprint matching process into an end-to-end inference process. Then, only a simple image transformation is needed on the complete fingerprint image based on the transformation matrix to obtain the transformed image. This completes the fingerprint matching process between the image to be identified and the transformed image, greatly saving the time spent on fingerprint matching.

[0087] Optionally, as an embodiment, when multiple complete fingerprint images are read, the position information of each complete fingerprint image is read; the fingerprint image to be identified and the complete fingerprint images are input into a fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the complete fingerprint images, including: inputting a fingerprint image pair composed of the fingerprint image to be identified and each complete fingerprint image into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in each complete fingerprint image, so as to generate multiple transformation matrices based on the predicted position information and the corresponding position information of the complete fingerprint image, thereby obtaining the matching degree between the fingerprint image to be identified and each complete fingerprint image.

[0088] When there are multiple complete fingerprint images stored, in order to improve the matching speed, the fingerprint image to be identified can be matched with multiple complete fingerprint images.

[0089] The location information of each complete fingerprint image is read. The fingerprint image processing model is then input with the fingerprint image to be identified and each complete fingerprint image. The model generates predicted location information for the fingerprint image to be identified within each complete fingerprint image. After obtaining the predicted location information of the fingerprint image to be identified within each complete fingerprint image, the transformation matrix of the fingerprint image to be identified and each complete fingerprint image can be generated by combining the location information of each complete fingerprint image.

[0090] The first original fingerprint image is transformed according to each transformation matrix to generate a transformed image corresponding to each first original image. The transformed image corresponding to each first original image is then matched with the corresponding second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and each complete fingerprint image. When the matching degree between any complete fingerprint image and the fingerprint image to be identified is higher than a preset threshold, an unlock command is output.

[0091] Optionally, as an embodiment, when multiple complete fingerprint images are read, the position information of each complete fingerprint image is read; the fingerprint image to be identified and the complete fingerprint images are input into a fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the complete fingerprint images, including: taking any one of the multiple complete fingerprint images as the current complete fingerprint image, and performing the following steps: inputting the current complete fingerprint image and the fingerprint image to be identified into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the current complete fingerprint image, generating a transformation matrix based on the predicted position information and the position information of the current complete fingerprint image, and then obtaining the matching degree between the fingerprint image to be identified and the current complete fingerprint image; when the matching degree between the fingerprint image to be identified and the current complete fingerprint image is lower than a preset threshold, taking another complete fingerprint image among the multiple complete fingerprint images as the current complete fingerprint image, and returning to the step of inputting the current complete fingerprint image and the fingerprint image to be identified into the fingerprint image processing model, until the multiple complete fingerprint images are traversed, or until the matching degree between the fingerprint image to be identified and any one of the complete fingerprint images is higher than the preset threshold.

[0092] When storing a large number of complete fingerprint images, matching each complete fingerprint image with the fingerprint image to be identified consumes significant computational resources. Therefore, multiple complete fingerprint images can be matched one by one with the fingerprint image to be identified. Optionally, the matching order of each complete fingerprint image can be planned based on the actual number of times each complete fingerprint image has been successfully matched.

[0093] Take any one of the multiple complete fingerprint images as the current complete fingerprint image, and perform the following step A: Input the current complete fingerprint image and the fingerprint image to be identified into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the current complete fingerprint image, so as to generate the transformation matrix of the current complete fingerprint image based on the predicted position information and the position information of the current complete fingerprint image; transform the first original fingerprint image according to the transformation matrix to obtain the transformed image, and match the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the current complete fingerprint image.

[0094] Optionally, the location information of the current complete fingerprint image can be read only when matching each current complete fingerprint image with the fingerprint image to be identified, instead of reading the location information of all complete fingerprint images at once.

[0095] If the matching degree between the current complete fingerprint image and the fingerprint image to be identified is lower than a preset threshold, another unmatched complete fingerprint image from among multiple complete fingerprint images is taken as the current complete fingerprint image, and step A above is repeated until all multiple complete fingerprint images have been traversed; or, until the matching degree between any complete fingerprint image and the fingerprint image to be identified is higher than the preset threshold. If the matching degree is higher than the preset threshold, an unlock command is output, and therefore no further matching is required. This method of matching one by one consumes relatively few computational resources.

[0096] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0097] This invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the fingerprint image matching method as described in the embodiments of this application.

[0098] This invention also provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the fingerprint image matching method as described in the embodiments of this application.

[0099] This invention also provides a computer program product, including a computer program / instructions, characterized in that, when the computer program / instructions are executed by a processor, they implement the fingerprint image matching method as proposed in the embodiments of this application.

[0100] This invention also provides a fingerprint image matching device. (Refer to...) Figure 4 The diagram illustrates the information interaction of a fingerprint image matching device according to an embodiment of the present invention. The fingerprint image matching device includes a sensor, a memory, and a processor. The sensor acquires a fingerprint image to be identified, the memory stores a complete fingerprint image, and the sensor, the memory, and the processor interact with each other to execute the fingerprint image matching method as proposed in the embodiments of this application.

[0101] A complete fingerprint image is pre-acquired by a sensor, registered by a processor, and stored in memory. Later, during fingerprint matching of the image to be identified, the sensor acquires the image and transmits it to the processor. The processor retrieves the complete fingerprint image stored in memory and performs a fingerprint matching process between the image to be identified and the complete fingerprint image to determine if a match is successful.

[0102] This invention also provides a fingerprint image matching device, referring to... Figure 5 The diagram illustrates a fingerprint image matching device according to an embodiment of the present invention. The device includes a fingerprint image acquisition module, a location information acquisition module, a transformation matrix determination module, a transformation image generation module, and a fingerprint image matching module, wherein:

[0103] A fingerprint image acquisition module is used to acquire a fingerprint image to be identified and to read a complete fingerprint image and the location information of the complete fingerprint image, wherein the area of ​​the fingerprint image to be identified is smaller than the area of ​​the complete fingerprint image;

[0104] The location information acquisition module is used to input the fingerprint image to be identified and the complete fingerprint image into the fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified in the complete fingerprint image;

[0105] The transformation matrix determination module is used to determine the transformation matrix based on the position information of the complete fingerprint image and the predicted position information. The transformation matrix represents the translation and / or rotation relationship between the fingerprint image to be identified and the image corresponding to the fingerprint image to be identified in the complete fingerprint image.

[0106] The transformed image generation module is used to transform the first original fingerprint image according to the transformation matrix to generate a transformed image;

[0107] A fingerprint image matching module is used to match the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image;

[0108] Wherein, one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image.

[0109] Optionally, as an embodiment, the location information acquisition module includes:

[0110] The coordinate acquisition unit is used to obtain the coordinates of the four corner points of the fingerprint image to be identified in the coordinate system of the complete fingerprint image.

[0111] Optionally, as an embodiment, when multiple complete fingerprint images are read, the fingerprint image acquisition module reads the location information of each complete fingerprint image;

[0112] The location information acquisition module includes:

[0113] The first location information acquisition unit is used to input the fingerprint image pair consisting of the fingerprint image to be identified and each complete fingerprint image into the fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified in each complete fingerprint image, so as to generate multiple transformation matrices based on the predicted location information and the corresponding location information of the complete fingerprint image, and then obtain the matching degree between the fingerprint image to be identified and each complete fingerprint image.

[0114] Optionally, as an embodiment, when multiple complete fingerprint images are read, the fingerprint image acquisition module reads the location information of each complete fingerprint image;

[0115] The location information acquisition module includes:

[0116] The second location information acquisition unit is used to select any one of multiple complete fingerprint images as the current complete fingerprint image and perform the following steps:

[0117] The current complete fingerprint image and the fingerprint image to be identified are input into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the current complete fingerprint image. A transformation matrix is ​​generated based on the predicted position information and the position information of the current complete fingerprint image, thereby obtaining the matching degree between the fingerprint image to be identified and the current complete fingerprint image.

[0118] When the matching degree between the fingerprint image to be identified and the current complete fingerprint image is lower than a preset threshold, another complete fingerprint image among the multiple complete fingerprint images is taken as the current complete fingerprint image, and the process returns to the step of inputting the current complete fingerprint image and the fingerprint image to be identified into the fingerprint image processing model, until the multiple complete fingerprint images are traversed, or until the matching degree between the fingerprint image to be identified and any complete fingerprint image is higher than the preset threshold.

[0119] Optionally, as an embodiment, the apparatus further includes:

[0120] The sample acquisition module is used to acquire complete fingerprint image samples;

[0121] A translation / rotation module is used to translate the complete fingerprint image sample by multiple distances and / or rotate it by multiple angles;

[0122] The cropping module is used to crop multiple sub-image samples of a target size from a complete fingerprint image sample after translation and / or rotation;

[0123] The model training module is used to train the target model using the complete fingerprint image sample and multiple sub-image samples carrying location labels to obtain the fingerprint image processing model, wherein the location labels represent the location information of the sub-image samples in the complete fingerprint image sample.

[0124] Optionally, as an embodiment, the model training module includes:

[0125] The predicted location information acquisition unit is used to input the complete fingerprint image sample and multiple sub-image samples carrying location tags into the target model to obtain the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample;

[0126] The loss function value determination unit is used to determine the mean squared error loss function value based on the difference between the location labels carried by each of the multiple sub-image samples and the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample.

[0127] The gradient update unit is used to perform stochastic gradient updates on the model parameters of the preset model based on the mean squared error loss function value with the goal of minimizing the difference, so as to obtain the fingerprint image processing model.

[0128] Optionally, as an embodiment, after matching the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image, the device further includes:

[0129] The unlocking module is used to output an unlocking command when the matching degree between the fingerprint image to be identified and the complete fingerprint image is higher than a preset threshold.

[0130] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0131] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0132] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, electronic devices, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0133] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0135] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0136] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0137] The fingerprint image matching method, device, medium, and program product provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A fingerprint image matching method, characterized in that, The method includes: Acquire a fingerprint image to be identified, and read the complete fingerprint image and the location information of the complete fingerprint image, wherein the area of ​​the fingerprint image to be identified is smaller than the area of ​​the complete fingerprint image; The fingerprint image to be identified and the complete fingerprint image are input into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the complete fingerprint image; Based on the location information of the complete fingerprint image and the predicted location information, a transformation matrix is ​​determined. The transformation matrix represents the translation and / or rotation relationship between the fingerprint image to be identified and the image corresponding to the fingerprint image to be identified in the complete fingerprint image. The first original fingerprint image is transformed according to the transformation matrix to generate a transformed image; The transformed image is matched with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image; Wherein, one of the complete fingerprint image and the fingerprint image to be identified is the first original fingerprint image, and the other is the second original fingerprint image; Obtaining the predicted location information of the fingerprint image to be identified within the complete fingerprint image includes: The coordinates of the four corner points of the fingerprint image to be identified are obtained in the coordinate system of the complete fingerprint image.

2. The method according to claim 1, characterized in that, If multiple complete fingerprint images are read, the location information of each complete fingerprint image is read. The fingerprint image to be identified and the complete fingerprint image are input into a fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified within the complete fingerprint image, including: The fingerprint image processing model inputs the fingerprint image to be identified and each complete fingerprint image into a fingerprint image pair to obtain the predicted position information of the fingerprint image to be identified in each complete fingerprint image. Multiple transformation matrices are then generated based on the predicted position information and the corresponding position information of the complete fingerprint image, thereby obtaining the matching degree between the fingerprint image to be identified and each complete fingerprint image.

3. The method according to claim 1, characterized in that, If multiple complete fingerprint images are read, the location information of each complete fingerprint image is read. The fingerprint image to be identified and the complete fingerprint image are input into a fingerprint image processing model to obtain the predicted location information of the fingerprint image to be identified within the complete fingerprint image, including: Take any one of the multiple complete fingerprint images as the current complete fingerprint image and perform the following steps: The current complete fingerprint image and the fingerprint image to be identified are input into the fingerprint image processing model to obtain the predicted position information of the fingerprint image to be identified in the current complete fingerprint image. A transformation matrix is ​​generated based on the predicted position information and the position information of the current complete fingerprint image, thereby obtaining the matching degree between the fingerprint image to be identified and the current complete fingerprint image. When the matching degree between the fingerprint image to be identified and the current complete fingerprint image is lower than a preset threshold, another complete fingerprint image among multiple complete fingerprint images is taken as the current complete fingerprint image, and the process returns to the step of inputting the current complete fingerprint image and the fingerprint image to be identified into the fingerprint image processing model, until multiple complete fingerprint images are traversed, or until the matching degree between the fingerprint image to be identified and any complete fingerprint image is higher than the preset threshold.

4. The method according to claim 1, characterized in that, The method further includes: Obtain a complete fingerprint image sample; The complete fingerprint image sample is translated by multiple distances and / or rotated by multiple angles; Extract multiple sub-image samples of a target size from a complete fingerprint image sample after translation and / or rotation; The target model is trained using the complete fingerprint image sample and multiple sub-image samples carrying location labels to obtain the fingerprint image processing model, wherein the location labels represent the location information of the sub-image samples in the complete fingerprint image sample.

5. The method according to claim 4, characterized in that, The fingerprint image processing model is obtained by training a target model using the complete fingerprint image sample and multiple sub-image samples carrying location tags, including: The complete fingerprint image sample and multiple sub-image samples carrying location tags are input into the target model to obtain the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample. The mean squared error loss function value is determined based on the difference between the location labels carried by each of the multiple sub-image samples and the predicted location information of each of the multiple sub-image samples in the complete fingerprint image sample. With the goal of minimizing the difference, the model parameters of the target model are updated by stochastic gradient based on the mean squared error loss function value to obtain the fingerprint image processing model.

6. The method according to any one of claims 1-5, characterized in that, After matching the transformed image with the second original fingerprint image to obtain the matching degree between the fingerprint image to be identified and the complete fingerprint image, the method further includes: When the matching degree between the fingerprint image to be identified and the complete fingerprint image is higher than a preset threshold, an unlock command is output.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the fingerprint image matching method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instruction is executed by the processor, it implements the fingerprint image matching method as described in any one of claims 1 to 6.

9. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the fingerprint image matching method as described in any one of claims 1 to 6.

Images