Identity verification method and apparatus, electronic device, and storage medium

Abstract

The application discloses an identity verification method and device, electronic equipment and storage medium, and belongs to the field of image processing. The identity verification method comprises the following steps: acquiring a first finger vein image of a first user; performing image cutting on the first finger vein image to obtain a plurality of first finger vein sub-images of different regions, and the image cutting strategy of the first finger vein image is determined based on the texture offset degree of the first finger vein image; acquiring a first global feature vector of the first finger vein image, and acquiring a plurality of first local feature vectors of the plurality of first finger vein sub-images, wherein the first local feature vectors correspond to the first finger vein sub-images one by one; comparing the first global feature vector and the plurality of first local feature vectors with identity verification information of a second user registered in an identity library to obtain an identity verification result of the first user, and the identity verification information of the second user comprises a second global feature vector and a plurality of second local feature vectors. The method improves the finger vein recognition accuracy and efficiency.

Description

Technical Field

[0001] This application belongs to the field of biometric identification technology, and in particular relates to an authentication method, device, electronic device and storage medium. Background Technology

[0002] Finger vein recognition is a biometric identification technology that plays an important role in the security field. However, due to limitations in the acquisition equipment, the acquired finger vein features often exhibit a certain degree of rotation and translation. This rotation and translation reduces the accuracy of finger vein recognition, thereby decreasing its efficiency and resulting in a poor user experience. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes an authentication method, apparatus, electronic device, and storage medium that effectively improves the accuracy and efficiency of finger vein recognition.

[0004] Firstly, this application provides an authentication method, which includes:

[0005] Obtain the first finger vein image of the first user;

[0006] The first finger vein image is segmented to obtain multiple first finger vein sub-images in different regions. The image segmentation strategy of the first finger vein image is determined based on the texture offset degree of the first finger vein image.

[0007] Obtain the first global feature vector of the first finger vein image, and obtain multiple first local feature vectors of the multiple first finger vein sub-images. The first local feature vectors are compared with the first finger vein sub-images. Figure 1 One-to-one correspondence;

[0008] The first global feature vector and the plurality of first local feature vectors are compared with the authentication information of the second user registered in the identity database to obtain the authentication result of the first user. The authentication information of the second user includes the second global feature vector and the plurality of second local feature vectors.

[0009] According to the authentication method of this application, by cutting the image into different regions, the finger vein image is divided into multiple finger vein sub-images. Then, the overall image of the user to be verified is compared with the overall image of the registered user, and the multiple finger vein sub-images of the user to be verified are compared with the multiple finger vein sub-images of the registered user. This solves the problem of poor recognition performance of finger veins under rotation and translation, thereby improving the accuracy and efficiency of finger vein recognition.

[0010] According to one embodiment of this application, comparing the first global feature vector and the plurality of first local feature vectors with the authentication information of a second user registered in the identity database to obtain the authentication result of the first user includes:

[0011] Based on the first global feature vector and the second global feature vector, the global feature similarity between the first user and the second user is determined;

[0012] Based on the plurality of first local feature vectors and the plurality of second local feature vectors, the local feature similarity between the first user and the second user is determined;

[0013] The authentication result of the first user is obtained based on at least one of the global feature similarity and the local feature similarity.

[0014] According to one embodiment of this application, obtaining the authentication result of the first user based on at least one of the global feature similarity and the local feature similarity includes:

[0015] If the global feature similarity is greater than the first global threshold, the authentication result of the first user is determined to be the second user;

[0016] Alternatively, if the global feature similarity is less than or equal to a first global threshold and greater than a second global threshold, and the local feature similarity is greater than a first local threshold, the authentication result of the first user is determined to be the second user, where the second global threshold is less than the first global threshold.

[0017] According to one embodiment of this application, obtaining the authentication result of the first user based on at least one of the global feature similarity and the local feature similarity includes:

[0018] If the global feature similarity is less than or equal to the second global threshold, the authentication result of the first user is determined to be an unregistered user.

[0019] Alternatively, if the local feature similarity is less than or equal to the first local threshold, the authentication result of the first user is determined to be an unregistered user.

[0020] According to one embodiment of this application, obtaining the first global feature vector of the first finger vein map and obtaining multiple first local feature vectors of the multiple first finger vein sub-maps includes:

[0021] The first finger vein image is input into the first layer of the authentication model to obtain the plurality of first global feature vectors output by the first layer, and the plurality of first finger vein sub-images are input into the second layer of the authentication model to obtain the plurality of first local feature vectors output by the second layer.

[0022] According to one embodiment of this application, comparing the first global feature vector and the plurality of first local feature vectors with the authentication information of a second user registered in the identity database to obtain the authentication result of the first user includes:

[0023] The first global feature vector and the plurality of first local feature vectors are input into the third layer of the authentication model. The first global feature vector is compared with the second global feature vector, and the plurality of first local feature vectors are compared with the plurality of second local feature vectors respectively, to obtain the authentication result of the first user output by the third layer; the authentication model includes the identity database.

[0024] According to one embodiment of this application, the authentication model is trained through the following steps:

[0025] Edge detection is performed on the finger vein image sample to determine the region of interest corresponding to the finger vein image sample, and image enhancement is performed on the finger vein image sample to obtain a first image sample;

[0026] The first image sample is subjected to data augmentation operations to obtain multiple second image samples. The data augmentation operations include at least one of image cropping, local filling, image translation, and image rotation.

[0027] Based on the multiple second image samples, determine the sample training set corresponding to the finger vein image sample;

[0028] The sample training set is input into the identity verification model to be trained, and the model parameters of the identity verification model are updated to obtain the trained identity verification model.

[0029] According to one embodiment of this application, the authentication information of the second user is registered in the identity database through the following steps:

[0030] Obtain the second finger vein image of the second user;

[0031] The second finger vein image is segmented to obtain multiple second finger vein sub-images in different regions. The image segmentation strategy of the second finger vein image is determined based on the texture offset degree of the second finger vein image.

[0032] Obtain the second global feature vector of the second finger vein image, and obtain multiple second local feature vectors of the multiple second finger vein sub-images. The second local feature vectors are compared with the second finger vein sub-images. Figure 1 One-to-one correspondence;

[0033] The identity database stores the mapping relationship between the second global feature vector, the plurality of second local feature vectors, and the second user.

[0034] Secondly, this application provides an authentication device, which includes:

[0035] The first acquisition module is used to acquire the first finger vein image of the first user;

[0036] The first processing module is used to perform image segmentation on the first finger vein image to obtain multiple first finger vein sub-images in different regions. The image segmentation strategy of the first finger vein image is determined based on the texture offset degree of the first finger vein image.

[0037] The second acquisition model is used to acquire a first global feature vector of the first finger vein image and acquire multiple first local feature vectors of the multiple first finger vein sub-images, wherein the first local feature vectors are compared with the first finger vein sub-images. Figure 1 One-to-one correspondence;

[0038] The second processing module is used to compare the first global feature vector and the plurality of first local feature vectors with the identity verification information of the second user registered in the identity database to obtain the identity verification result of the first user. The identity verification information of the second user includes the second global feature vector and the plurality of second local feature vectors.

[0039] According to the authentication device of this application, by cutting the image into multiple finger vein sub-images according to different regions, and then comparing the overall image of the user to be verified with the overall image of the registered user, and comparing the multiple finger vein sub-images of the user to be verified with the multiple finger vein sub-images of the registered user, the problem of poor recognition performance of finger veins under rotation and translation is solved, thereby improving the accuracy and efficiency of finger vein recognition.

[0040] Thirdly, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the authentication method as described in the first aspect above.

[0041] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the authentication method as described in the first aspect above.

[0042] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0043] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0044] Figure 1 This is one of the flowcharts illustrating the authentication method provided in the embodiments of this application;

[0045] Figure 2 This is a second flowchart illustrating the authentication method provided in the embodiments of this application;

[0046] Figure 3 This is the third flowchart illustrating the authentication method provided in the embodiments of this application;

[0047] Figure 4 This is a schematic diagram of the structure of the identity verification device provided in the embodiments of this application;

[0048] Figure 5 This is a hardware schematic diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0049] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0050] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0051] The following is combined Figures 1-5 The authentication method, authentication device, electronic device, and readable storage medium provided in this application will be described in detail through specific embodiments and application scenarios.

[0052] like Figure 1 As shown, the authentication method provided in this application embodiment includes steps 110 to 140.

[0053] Step 110: Obtain the first finger vein image of the first user.

[0054] The first user is the user whose identity is to be verified, and the first finger vein image is the finger vein image of the first user.

[0055] The first finger vein image of the first user can be obtained through a finger vein acquisition device.

[0056] The finger vein acquisition device can obtain finger vein images by transmitting infrared light through the finger vein sample and using a camera, or it can obtain finger vein images by transmitting near-infrared light through the finger vein sample and using a camera.

[0057] The finger vein acquisition device for obtaining the first finger vein image of the first user can be a camera equipped with an infrared LED light. The infrared LED light shines infrared light above the finger, and the camera captures the finger vein image from the other side of the finger. The finger vein image is then acquired from the camera. After the finger vein image is preprocessed, a Region of Interest (ROI) sample is obtained. After data augmentation of the ROI sample, the first finger vein image can be obtained.

[0058] Step 120: Perform image segmentation on the first finger vein image to obtain multiple first finger vein sub-images in different regions. The image segmentation strategy of the first finger vein image is determined based on the degree of texture offset of the first finger vein image.

[0059] Image segmentation is a technique that divides an image into several specific regions.

[0060] The first finger vein sub-image is a number of different first finger vein regions obtained after image segmentation of the first finger vein image; the image segmentation strategy determines the segmentation rules of the first finger vein image based on different situations; the texture offset degree of the first finger vein image is the degree of offset of the first finger vein image compared with the upright sample in the horizontal and vertical directions.

[0061] Among them, the upright sample is a finger vein image obtained by illuminating the finger with near-infrared light and using a CCD camera when the finger is facing upwards.

[0062] In this embodiment, when performing image segmentation on the first finger vein image, the first finger vein image and the upright sample are first compared in the horizontal and vertical directions to determine the degree of texture offset of the first finger vein image. Based on the degree of texture offset of the first finger vein image, an image segmentation strategy for the first finger vein image is determined. Based on the determined image segmentation strategy, the first finger vein image is segmented to obtain multiple first finger vein sub-images.

[0063] Step 130: Obtain the first global feature vector of the first finger vein map, and obtain multiple first local feature vectors of multiple first finger vein sub-maps, wherein the first local feature vectors and the first finger vein sub-maps are... Figure 1 One-to-one correspondence.

[0064] The first global feature vector is a representation of the global features of the first finger vein image of the first user, and the first local feature vector is a representation of the local features of the first finger vein image of the first user, obtained after cutting according to the above cutting rules.

[0065] There are multiple first finger vein sub-images and multiple first local feature vectors, and each first finger vein sub-image corresponds to a first local feature vector.

[0066] Step 140: Compare the first global feature vector and multiple first local feature vectors with the identity verification information of the second user registered in the identity database to obtain the identity verification result of the first user. The identity verification information of the second user includes the second global feature vector and multiple second local feature vectors.

[0067] The identity database includes authentication information for multiple registered users. The authentication information includes the global feature vector and multiple local feature vectors of the corresponding registered user.

[0068] In this embodiment, after obtaining the first global feature vector and multiple first local feature vectors of the user whose identity is to be verified, they are compared with the second user authentication information in the identity database. Specifically, the first global feature vector is compared with the second global feature vector, and the first local feature vector is compared with the second local feature vector, thereby determining the authentication result of the first user.

[0069] In related technologies, a method for finger vein feature extraction and recognition based on terrain point classification has been proposed. This method uses a multi-scale Gaussian filter to collect finger vein image information at different scales, performs image processing on the finger vein images at different scales to obtain image features, and compares the user to be registered with the registered user through the image features to obtain the finger vein recognition result. However, this method cannot solve the problem of the performance degradation of finger veins under rotation and translation, and the accuracy and efficiency of finger vein recognition are low, with high time complexity.

[0070] In this embodiment, multiple first finger vein sub-images are obtained by sub-region segmentation, and finger vein recognition is performed based on sub-region mutual matching technology. This method can solve the problem of deterioration of finger vein performance under rotation and translation, improve the accuracy and efficiency of finger vein recognition, and also has good scalability. The database will only increase the amount of computation by a small amount when the number of registered samples increases, and has low time complexity.

[0071] According to the authentication method of this application, by cutting the image into different regions, the finger vein image is divided into multiple finger vein sub-images. Then, the overall image of the user to be verified is compared with the overall image of the registered user, and the multiple finger vein sub-images of the user to be verified are compared with the multiple finger vein sub-images of the registered user. This solves the problem of poor recognition performance of finger veins under rotation and translation, thereby improving the accuracy and efficiency of finger vein recognition.

[0072] In some embodiments, step 140, comparing the first global feature vector and multiple first local feature vectors with the authentication information of the second user registered in the identity database to obtain the authentication result of the first user, may include:

[0073] Based on the first global feature vector and the second global feature vector, determine the global feature similarity between the first user and the second user;

[0074] Based on multiple first local feature vectors and multiple second local feature vectors, the local feature similarity between the first user and the second user is determined.

[0075] Global feature similarity is a numerical value that describes the degree of similarity between the first global feature vector and the second global feature vector, while local feature similarity is a numerical value that describes the degree of similarity between the first local feature vector and the second local feature vector.

[0076] In this embodiment, the first global feature vector and multiple first local feature vectors are represented together by a set, which yields the first feature vector set. in This is the first global feature vector. These are multiple first local feature vectors.

[0077] Representing the second global feature vector and multiple second local feature vectors together as a set yields the set of second feature vectors. in This is the second global feature vector. These are multiple second local feature vectors.

[0078] Set S r and set S t Calculate the cosine similarity to obtain K = {K} G ,K L}

[0079] in, The global feature similarity score represents the highest match between the authentication information of the first user and the authentication information of the second user. The local feature similarity is the highest matching degree between the authentication information of the first user and the authentication information of the second user, and the second user corresponding to this local feature similarity is the registered user in the identity database with the highest matching degree with the authentication information of the first user.

[0080] Where M represents the number of second users in the identity database, N represents the number of sub-regions, and distance(·) represents the distance function that measures the distance between two feature vectors. This application uses cosine similarity as the distance function.

[0081] The segmentation strategy for the sub-region is determined by the degree of texture offset of the finger vein sample.

[0082] For example, such as Figure 2 As shown, assuming the width and height of the finger vein sample are w and h respectively, and the offset of the finger vein sample relative to the upright sample in the horizontal and vertical directions are Δw and Δh respectively, and four sub-regions are selected for cutting, then the cut sub-regions can be as follows: L1(0,0,w-Δw,h-Δh), L2(Δw,Δh,w-Δw,h-Δh), L3(Δw,0,w-Δw,h-Δh), and L4(0,Δh,w-Δw,h-Δh). The coordinates within parentheses represent the starting point coordinates and length of the sub-region in the original finger vein image, respectively: starting position in the X direction, starting position in the Y direction, width in the X direction, and width in the Y direction.

[0083] In practice, since different finger vein acquisition devices produce different texture offsets, the coordinates of the sub-region cutting and the number of sub-regions can be modified by changing the configuration file, thereby obtaining different numbers and sizes of sub-regions.

[0084] In some embodiments, the authentication result of the first user is obtained based on at least one of global feature similarity and local feature similarity.

[0085] At least one of them indicates that the authentication result of the first user can be obtained based on both global feature similarity and local feature similarity.

[0086] In this embodiment, the result can be determined by global feature similarity alone, or by a combination of global feature similarity and local feature similarity.

[0087] In some embodiments, the authentication result of the first user is obtained based on at least one of global feature similarity and local feature similarity, including:

[0088] If the global feature similarity is greater than the first global threshold, the authentication result of the first user is obtained as the second user;

[0089] Alternatively, if the global feature similarity is less than or equal to the first global threshold and greater than the second global threshold, and the local feature similarity is greater than the first local threshold, the authentication result of the first user is the second user, where the second global threshold is less than the first global threshold.

[0090] The first global threshold is a critical value for measuring the global feature similarity. When the global feature similarity of the first user is greater than the first global threshold, it means that the finger vein feature of the first user is very similar to that of the second user as a whole. At this time, the authentication result of the first user is the second user.

[0091] The second global threshold is also a critical value for measuring global feature similarity. The second global threshold is less than the first global threshold. When the global feature similarity of the first user is less than or equal to the first global threshold and greater than the second global threshold, it means that the first user is not very similar to the finger vein features of the second user as a whole, and additional local feature similarity needs to be considered.

[0092] The first local threshold is a critical value for measuring the degree of similarity of local features. When the global feature similarity of the first user is less than or equal to the first global threshold and greater than the second global threshold, and at the same time the local feature similarity of the first user is greater than the first local threshold, the authentication result of the first user is also the second user.

[0093] In some embodiments, the authentication result of the first user is obtained based on at least one of global feature similarity and local feature similarity, including:

[0094] If the global feature similarity is less than or equal to the second global threshold, the authentication result of the first user is an unregistered user.

[0095] Alternatively, if the local feature similarity is less than or equal to the first local threshold, the authentication result of the first user is determined to be an unregistered user.

[0096] An unregistered user indicates that the user did not register in the identity database before identity verification, and no matching second user could be found in the identity database.

[0097] The following is a specific implementation example for determining whether the first user is a second user who has already registered.

[0098] For example, when the first global threshold is 80, the second global threshold is 60, and the first local threshold is 75.

[0099] If the global feature similarity between the first user and the second user is 90 and the local feature similarity is 50, then the global feature similarity of 90 between the first user and the second user is greater than the first global threshold of 80, and the authentication result of the first user is the second user.

[0100] If the global feature similarity between the first user and the second user is 70 and the local feature similarity is 85, then since the global feature similarity of 75 between the first user and the second user is less than the first global threshold and greater than the second global threshold, and the local feature similarity of 85 between the first user and the second user is greater than the first local threshold, then the identity verification result of the first user is the second user.

[0101] If the global feature similarity between the first user and the second user is 65 and the local feature similarity is 70, since the global feature similarity of 75 between the first user and the second user is less than the first global threshold and greater than the second global threshold, but the local feature similarity of 70 between the first user and the second user is less than the first local threshold, then the authentication result of the first user is an unregistered user.

[0102] If the global feature similarity between the first user and the second user is 50 and the local feature similarity is 80, since the global feature similarity between the first user and the second user is less than the second global threshold, and the overall similarity between the first user and the second user is too low, the local feature similarity can be ignored, and the identity verification result of the first user can be obtained as an unregistered user.

[0103] In some embodiments, obtaining a first global feature vector of a first finger vein map and obtaining multiple first local feature vectors of multiple first finger vein sub-maps includes:

[0104] The first finger vein map is input into the first layer of the authentication model to obtain multiple first global feature vectors output by the first layer, and multiple first finger vein sub-maps are input into the second layer of the authentication model to obtain multiple first local feature vectors output by the second layer.

[0105] The identity verification model is obtained by training the recognition network using multiple finger vein pattern samples.

[0106] The recognition network can be MobileNetV1, ShuffleNet, or MobilefaceNet. Since the MobileNetV1 and ShuffleNet recognition networks use average pooling layers, the accuracy of finger vein recognition decreases because each unit has the same weight.

[0107] The MobileFaceNet recognition network uses global depth convolutional layers instead of average pooling layers, which allows different locations in the finger vein image to have different levels of importance during finger vein recognition, thus improving the accuracy of finger vein recognition.

[0108] In this embodiment, the authentication model can be the MobileFaceNet identification network.

[0109] The first layer of the authentication model is used to obtain and output the first global feature vector of the first finger vein image. The second layer of the authentication model is used to obtain and output multiple first local feature vectors of multiple first finger vein sub-images.

[0110] In actual implementation, the first finger vein image is input into the first layer of the authentication model, and the output of the first layer of the authentication model is the first global feature vector. Multiple first finger vein sub-images are input into the second layer of the authentication model, and the multiple outputs of the second layer of the authentication model are the multiple first local feature vectors.

[0111] In some embodiments, the first global feature vector and multiple first local feature vectors are compared with the authentication information of a second user registered in the identity database to obtain the authentication result of the first user, including:

[0112] The first global feature vector and multiple first local feature vectors are input into the third layer of the authentication model. The first global feature vector is compared with the second global feature vector, and the corresponding multiple first local feature vectors are compared with the multiple second local feature vectors respectively. The authentication result of the first user is obtained from the output of the third layer.

[0113] The authentication model includes an identity database.

[0114] In this embodiment, the third layer of the identity verification model includes: a global feature similarity calculation module, a local feature similarity calculation module, and a comparison module.

[0115] The global feature similarity calculation module is used to obtain the global feature similarity between the first user and the user that best matches the identity database, while the local feature similarity calculation module is used to obtain the local feature similarity between the first user and the user that best matches the identity database.

[0116] The comparison module is used to determine the magnitude of global feature similarity with the first global threshold or the second global threshold, and the magnitude of local feature similarity with the first local threshold. Based on the determination results, the authentication result of the first user is output.

[0117] In some embodiments, such as Figure 3 As shown, the authentication model is trained through the following steps:

[0118] Edge detection is performed on the finger vein image samples to determine the corresponding regions of interest, and image enhancement is performed on the finger vein image samples to obtain the first image sample.

[0119] Edge detection is based on identifying points with significant brightness changes in the finger vein image sample, thereby removing irrelevant information and preserving the important structural attributes of the image, thus significantly reducing the amount of data. The region of interest is the area retained after edge detection of the finger vein image sample, where irrelevant information has been removed.

[0120] There are various edge detection operators, including the 100G operator, the Laplacian operator, and the Canny operator. In this embodiment, the Canny operator is used for edge detection.

[0121] Image enhancement aims to improve the visual effect of an image by intentionally emphasizing its overall or local characteristics, making an originally unclear image clearer, highlighting certain features or regions of interest, amplifying the differences between features of different objects in the image, and suppressing features of little interest. This improves image quality, enriches information, enhances image interpretation and recognition, and meets the needs of certain special analyses.

[0122] Data augmentation is performed on the first image sample to obtain multiple second image samples. The data augmentation operation includes at least one of image cropping, local filling, image translation, and image rotation.

[0123] Data augmentation uses geometric transformations to generate multiple second image samples based on a first image sample, thereby increasing the training data, improving the model's generalization ability, and avoiding sample imbalance.

[0124] Data augmentation can include only image cropping, or it can include four types: image cropping, local filling, image translation, and image selection.

[0125] Based on multiple second image samples, determine the sample training set corresponding to the finger vein image samples.

[0126] The sample training set is input into the authentication model to be trained, the model parameters of the authentication model are updated, and the trained authentication model is obtained.

[0127] In actual execution, after acquiring the original infrared transmission finger vein image, Canny edge detection is performed on the finger vein image to remove the background of non-finger vein areas in the image. Then, histogram equalization is used to enhance the finger vein image, improving the texture contrast of the finger vein image, and obtaining the first image sample.

[0128] Based on the obtained first image sample, multiple second image samples are obtained through data augmentation, and a sample training set corresponding to the finger vein image sample is obtained based on the multiple second image samples.

[0129] Then, the sample training set is input into the authentication model to be trained for training, the model parameters of the authentication model are updated, and the trained authentication model is obtained.

[0130] In this embodiment, the identity verification model to be trained is the MobileFaceNet recognition network, and the loss function used is ArcFace.

[0131] The Arcface loss function is shown below.

[0132]

[0133] Where m is the number of images in a batch in the training set, n is the number of classes in the training set, and y i Let θ represent the category of the i-th sample in the current batch, s be the scaling factor, and θ be the scaling factor. j η is the parameter for category j, and η is the set interval size.

[0134] The Arcface loss function is used to evaluate the degree to which the predicted values ​​of the authentication model differ from the actual values. The larger the value of the loss function, the better the performance of the authentication model.

[0135] In some embodiments, such as Figure 3 As shown, the second user's authentication information is registered in the identity database through the following steps:

[0136] Obtain the second finger vein image of the second user.

[0137] The finger vein acquisition device for obtaining the first finger vein image of the second user can be a camera equipped with an infrared LED light. The infrared LED light shines infrared light above the finger, and the camera captures the finger vein image from the other side of the finger. The finger vein image is then acquired from the camera. After the finger vein image is preprocessed, a Region of Interest (ROI) sample is obtained. After data augmentation of the ROI sample, the second finger vein image can be obtained.

[0138] The second finger vein image is segmented to obtain multiple second finger vein sub-images in different regions. The image segmentation strategy of the second finger vein image is determined based on the degree of texture offset of the second finger vein image.

[0139] When performing image segmentation on the second finger vein image, the second finger vein image and the upright sample are first compared in the horizontal and vertical directions to determine the degree of texture offset of the second finger vein image. Based on the degree of texture offset of the second finger vein image, the image segmentation strategy of the first finger vein image is determined. The second finger vein image is then segmented according to the determined image segmentation strategy to obtain multiple second finger vein sub-images.

[0140] Obtain the second global feature vector of the second finger vein image, and obtain multiple second local feature vectors of multiple second finger vein sub-images. The second local feature vectors are then compared with the second finger vein sub-images. Figure 1 One-to-one correspondence.

[0141] By incorporating the second finger vein map and multiple second finger vein sub-maps into the authentication model, the second global feature vector of the second finger vein map and the second local feature vectors of multiple second finger vein sub-maps can be obtained.

[0142] Among them, the second local feature vector and the second finger vein sub-vector Figure 1 One-to-one correspondence.

[0143] The identity database stores the mapping relationship between the second global feature vector, multiple second local feature vectors, and the second user.

[0144] The mapping relationship describes the correspondence between the second user and the second global feature vector, and multiple second local feature vectors.

[0145] In this embodiment, by searching for the second user in the identity database, and after obtaining the second user, the second global feature vector and multiple second local feature vectors corresponding to the second user can be obtained based on the mapping relationship between the second user and the second global feature vector and multiple second local feature vectors.

[0146] Similarly, a second user can be obtained by searching for a second global feature vector and multiple second local feature vectors in the identity database.

[0147] The authentication method can be applied to the terminal, and can be executed by the hardware or software in the terminal.

[0148] The terminal includes, but is not limited to, portable communication devices such as mobile phones or tablets with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but rather a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).

[0149] The following embodiments describe a terminal including a display and a touch-sensitive surface. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and joystick.

[0150] The authentication method provided in this application embodiment can be executed by an electronic device or a functional module or entity in an electronic device that can implement the authentication method. The electronic devices mentioned in this application embodiment include, but are not limited to, mobile phones, tablets, computers, cameras, and wearable devices. The authentication method provided in this application embodiment is described below using an electronic device as the execution subject.

[0151] The authentication method provided in this application can be executed by an authentication device. This application uses an authentication device executing the authentication method as an example to illustrate the authentication device provided in this application.

[0152] This application also provides an identity verification device.

[0153] like Figure 4 As shown, the authentication device includes: a first acquisition module 410, a first processing module 420, a second acquisition module 430, and a second processing module 440.

[0154] The first acquisition module 410 is used to acquire the first finger vein image of the first user;

[0155] The first processing module 420 is used to perform image segmentation on the first finger vein map to obtain multiple first finger vein sub-maps in different regions. The image segmentation strategy of the first finger vein map is determined based on the texture offset degree of the first finger vein map.

[0156] The second acquisition module 430 is used to acquire the first global feature vector of the first finger vein map and acquire multiple first local feature vectors of multiple first finger vein sub-maps, wherein the first local feature vectors are compared with the first finger vein sub-maps. Figure 1 One-to-one correspondence;

[0157] The second processing module 440 is used to compare the first global feature vector and multiple first local feature vectors with the identity verification information of the second user registered in the identity database to obtain the identity verification result of the first user. The identity verification information of the second user includes the second global feature vector and multiple second local feature vectors.

[0158] According to the authentication device of this application, by cutting the image into multiple finger vein sub-images according to different regions, and then comparing the overall image of the user to be verified with the overall image of the registered user, and comparing the multiple finger vein sub-images of the user to be verified with the multiple finger vein sub-images of the registered user, the problem of poor recognition performance of finger veins under rotation and translation is solved, thereby improving the accuracy and efficiency of finger vein recognition.

[0159] In some embodiments, the second processing module 440 is further configured to:

[0160] Based on the first global feature vector and the second global feature vector, determine the global feature similarity between the first user and the second user;

[0161] Based on multiple first local feature vectors and multiple second local feature vectors, the local feature similarity between the first user and the second user is determined.

[0162] The authentication result of the first user is obtained based on at least one of global feature similarity and local feature similarity.

[0163] In some embodiments, the second processing module 440 is further configured to:

[0164] If the global feature similarity is greater than the first global threshold, the authentication result of the first user is obtained as the second user;

[0165] Alternatively, if the global feature similarity is less than or equal to the first global threshold and greater than the second global threshold, and the local feature similarity is greater than the first local threshold, the authentication result of the first user is obtained as the second user, wherein the second global threshold is less than the first global threshold.

[0166] In some embodiments, the second processing module 440 is further configured to:

[0167] If the global feature similarity is less than or equal to the second global threshold, the authentication result of the first user is an unregistered user.

[0168] Alternatively, if the local feature similarity is less than or equal to the first local threshold, the authentication result of the first user is determined to be an unregistered user.

[0169] In some embodiments, the second acquisition module 430 is further configured to:

[0170] The first finger vein map is input into the first layer of the authentication model to obtain multiple first global feature vectors output by the first layer, and multiple first finger vein sub-maps are input into the second layer of the authentication model to obtain multiple first local feature vectors output by the second layer.

[0171] The first global feature vector and multiple first local feature vectors are input into the third layer of the authentication model. The first global feature vector is compared with the second global feature vector, and the multiple first local feature vectors are compared with the multiple second local feature vectors respectively. The authentication result of the first user is obtained from the output of the third layer.

[0172] The authentication model includes an identity database;

[0173] The authentication model is trained based on finger vein image samples.

[0174] In some embodiments, the first acquisition module 410 is used for:

[0175] Edge detection is performed on the finger vein image sample to determine the region of interest corresponding to the finger vein image sample, and image enhancement is performed on the finger vein image sample to obtain a first image sample;

[0176] The first image sample is subjected to data augmentation operations to obtain multiple second image samples. The data augmentation operations include at least one of image cropping, local filling, image translation, and image rotation.

[0177] Based on the multiple second image samples, determine the sample training set corresponding to the finger vein image sample;

[0178] The sample training set is input into the identity verification model to be trained, and the model parameters of the identity verification model are updated to obtain the trained identity verification model.

[0179] In some embodiments, the second acquisition module 430 is used to acquire the second finger vein image of the second user;

[0180] The second processing module 440 is used to perform image segmentation on the second finger vein map to obtain multiple second finger vein sub-maps in different regions. The image segmentation strategy of the second finger vein map is determined based on the degree of texture offset of the second finger vein map.

[0181] The second acquisition module 430 is used to acquire the second global feature vector of the second finger vein map and acquire multiple second local feature vectors of multiple second finger vein sub-maps. The second local feature vectors are compared with the second finger vein sub-maps. Figure 1 One-to-one correspondence;

[0182] The second processing module 440 is used to store the mapping relationship between the second global feature vector, multiple second local feature vectors and the second user in the identity database.

[0183] The authentication device in this application embodiment can be an electronic device or a component within an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television set (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the scope of the device.

[0184] The authentication device in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit the specific operating system used.

[0185] The authentication device provided in this application embodiment can achieve... Figures 1 to 5 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0186] In some embodiments, such as Figure 5 As shown, this application embodiment also provides an electronic device 500, including a processor 501, a memory 502, and a computer program stored on the memory 502 and executable on the processor 501. When the program is executed by the processor 501, it implements the various processes of the above-described authentication method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0187] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0188] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described authentication method embodiments and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0189] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0190] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. 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 apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0191] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0192] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0193] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0194] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. An authentication method, characterized in that, include: Obtain the first finger vein image of the first user; The first finger vein image is segmented to obtain multiple first finger vein sub-images in different regions. The image segmentation strategy of the first finger vein image is determined based on the texture offset degree of the first finger vein image. Specifically, when segmenting the first finger vein image, the first finger vein image and the upright sample are first compared in the horizontal and vertical directions to determine the texture offset degree of the first finger vein image. Obtain the first global feature vector of the first finger vein image, and obtain multiple first local feature vectors of the multiple first finger vein sub-images, wherein the first local feature vectors correspond one-to-one with the first finger vein sub-images; The first global feature vector and the plurality of first local feature vectors are compared with the authentication information of the second user registered in the identity database to obtain the authentication result of the first user. The authentication information of the second user includes the second global feature vector and the plurality of second local feature vectors.

2. The authentication method according to claim 1, characterized in that, The step of comparing the first global feature vector and the plurality of first local feature vectors with the authentication information of the second user registered in the identity database to obtain the authentication result of the first user includes: Based on the first global feature vector and the second global feature vector, the global feature similarity between the first user and the second user is determined; Based on the plurality of first local feature vectors and the plurality of second local feature vectors, the local feature similarity between the first user and the second user is determined; The authentication result of the first user is obtained based on at least one of the global feature similarity and the local feature similarity.

3. The authentication method according to claim 2, characterized in that, The step of obtaining the authentication result of the first user based on at least one of the global feature similarity and the local feature similarity includes: If the global feature similarity is greater than the first global threshold, the authentication result of the first user is determined to be the second user; Alternatively, if the global feature similarity is less than or equal to a first global threshold and greater than a second global threshold, and the local feature similarity is greater than a first local threshold, the authentication result of the first user is determined to be the second user, where the second global threshold is less than the first global threshold.

4. The authentication method according to claim 2, characterized in that, The step of obtaining the authentication result of the first user based on at least one of the global feature similarity and the local feature similarity includes: If the global feature similarity is less than or equal to the second global threshold, the authentication result of the first user is determined to be an unregistered user. Alternatively, if the local feature similarity is less than or equal to the first local threshold, the authentication result of the first user is determined to be an unregistered user.

5. The authentication method according to any one of claims 1-4, characterized in that, The step of obtaining the first global feature vector of the first finger vein image and obtaining the first local feature vectors of the plurality of first finger vein sub-images includes: The first finger vein image is input into the first layer of the authentication model to obtain the plurality of first global feature vectors output by the first layer, and the plurality of first finger vein sub-images are input into the second layer of the authentication model to obtain the plurality of first local feature vectors output by the second layer. The authentication model is trained based on finger vein image samples.

6. The authentication method according to claim 5, characterized in that, The authentication model is trained through the following steps: Edge detection is performed on the finger vein image sample to determine the region of interest corresponding to the finger vein image sample, and image enhancement is performed on the finger vein image sample to obtain a first image sample; The first image sample is subjected to data augmentation operations to obtain multiple second image samples. The data augmentation operations include at least one of image cropping, local filling, image translation, and image rotation. Based on the multiple second image samples, determine the sample training set corresponding to the finger vein image sample; The sample training set is input into the identity verification model to be trained, and the model parameters of the identity verification model are updated to obtain the trained identity verification model.

7. The authentication method according to claim 1, characterized in that, The second user's authentication information is registered in the identity database through the following steps: Obtain the second finger vein image of the second user; The second finger vein image is segmented to obtain multiple second finger vein sub-images in different regions. The image segmentation strategy of the second finger vein image is determined based on the texture offset degree of the second finger vein image. Obtain the second global feature vector of the second finger vein image, and obtain multiple second local feature vectors of the multiple second finger vein sub-images, wherein the second local feature vectors correspond one-to-one with the second finger vein sub-images; The identity database stores the mapping relationship between the second global feature vector, the plurality of second local feature vectors, and the second user.

8. An authentication device, characterized in that, include: The first acquisition module is used to acquire the first finger vein image of the first user; The first processing module is used to perform image segmentation on the first finger vein image to obtain multiple first finger vein sub-images in different regions. The image segmentation strategy of the first finger vein image is determined based on the texture offset degree of the first finger vein image. Specifically, when performing image segmentation on the first finger vein image, the first finger vein image and the upright sample are first compared in the horizontal and vertical directions to determine the texture offset degree of the first finger vein image. The second acquisition module is used to acquire the first global feature vector of the first finger vein image and acquire multiple first local feature vectors of the multiple first finger vein sub-images, wherein the first local feature vectors correspond one-to-one with the first finger vein sub-images. The second processing module is used to compare the first global feature vector and the plurality of first local feature vectors with the identity verification information of the second user registered in the identity database to obtain the identity verification result of the first user. The identity verification information of the second user includes the second global feature vector and the plurality of second local feature vectors.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the authentication method as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the authentication method as described in any one of claims 1-7.

Images