Method, device, equipment and medium for updating feature vector database

Abstract

The present application discloses a method, device, equipment and medium for updating a feature vector database, which belongs to the field of data processing. The method includes: obtaining a first biometric feature in a business request; obtaining a first feature vector and a second feature vector respectively through a first algorithm model and a second algorithm model according to the first biometric feature, and the first feature vector database includes a sample feature vector obtained based on the first algorithm model; verifying the validity of the second feature vector according to the associated feature vector of the first user corresponding to the first sample feature vector; if the validity verification of the second feature vector passes, obtaining a second sample feature vector of the first user based on the second feature vector, and storing the second sample feature vector in a second feature vector database configured by the second algorithm model. According to the embodiment of the present application, risks to the user's privacy data can be avoided in the process of obtaining the sample feature vector required for the feature vector database.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method, device, equipment and medium for updating a feature vector database. Background Art

[0002] Biometric identification technology is a technology that uses the physiological or behavioral characteristics inherent in the human body to identify individuals. With the development of biometric identification technology, the application fields of biometric identification technology are becoming more and more extensive. For example, biometric identification technology can be applied to payment, access, attendance and other fields.

[0003] When the network structure that biometric recognition technology relies on changes or the biometric recognition algorithm needs to be optimized, the biometric recognition algorithm model needs to be updated. However, the feature vectors of biometrics obtained by different versions of the algorithm model will be different, and a database of corresponding feature vectors needs to be generated for different versions of the algorithm model.

[0004] At present, the original biometrics of users when they register are stored. When it is necessary to generate databases corresponding to different versions of algorithm models, new feature vectors are generated using the original biometrics, and then a new database is formed using the new feature vectors. However, the original biometrics are the user's private data, and the security risk of storing the user's original biometrics is relatively high. In other words, the process of obtaining the feature vectors required for the new database will bring greater security risks to the user's private data. Summary of the invention

[0005] The embodiments of the present application provide a method, apparatus, device and medium for updating a feature vector database, which can avoid risks to user privacy data in the process of obtaining sample feature vectors required by the feature vector database.

[0006] In a first aspect, an embodiment of the present application provides a method for updating a feature vector database, comprising: obtaining a first biometric feature in a business request; obtaining a first feature vector and a second feature vector respectively according to the first biometric feature through a first algorithm model and a second algorithm model, the first algorithm model and the second algorithm model having different versions, the first algorithm model being configured with a first feature vector database, the first feature vector database including sample feature vectors obtained based on the first algorithm model; verifying the validity of the second feature vector according to an associated feature vector of the first user corresponding to the first sample feature vector, the first sample feature vector being a sample feature vector in the first feature vector database that successfully matches the first feature vector, the associated feature vector including at least part of the following: the first feature vector, the second feature vector, and at least part of the historical feature vectors corresponding to the first algorithm model; when the validity verification of the second feature vector passes, obtaining a second sample feature vector of the first user based on the second feature vector, and storing the second sample feature vector in a second feature vector database configured with the second algorithm model.

[0007] In a second aspect, an embodiment of the present application provides an updating device for a feature vector database, comprising: an acquisition module, used to acquire a first biometric feature in a business request; a calculation module, used to obtain a first feature vector and a second feature vector respectively according to the first biometric feature through a first algorithm model and a second algorithm model, the first algorithm model and the second algorithm model have different versions, the first algorithm model is configured with a first feature vector database, and the first feature vector database includes sample feature vectors obtained based on the first algorithm model; a validity verification module, used to verify the validity of the second feature vector according to an associated feature vector of the first user corresponding to the first sample feature vector, the first sample feature vector is a sample feature vector in the first feature vector database that successfully matches the first feature vector, and the associated feature vector includes at least part of the following: the first feature vector, the second feature vector, and at least part of the historical feature vector corresponding to the first algorithm model; a processing module, used to obtain a second sample feature vector of the first user based on the second feature vector when the validity verification of the second feature vector passes, and store the second sample feature vector in a second feature vector database configured with the second algorithm model.

[0008] In a third aspect, an embodiment of the present application provides a device for updating a feature vector database, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for updating the feature vector database of the first aspect is implemented.

[0009] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the method for updating the feature vector database of the first aspect is implemented.

[0010] The embodiment of the present application provides a method, device, equipment and medium for updating a feature vector database. In the process of two different versions of algorithm models running in parallel, a first feature vector can be generated in real time by a first algorithm model according to the biometric features in the user's business request, and a second feature vector can be generated in real time by a second algorithm model. The feature vector obtained by the algorithm model that is not fully online in the two algorithm models, i.e., the second algorithm model, can be verified for validity based on at least part of the first feature vector, the second feature vector and the historical feature vector. If the validity verification passes, the feature vector is stored as a sample feature vector of the user in an unfinished feature vector database, i.e., the second feature vector database. In the process of obtaining the sample feature vector in the second feature vector database, it is not necessary to store the original biometric features registered by the user, thereby ensuring that the user's privacy data will not be at risk in the process of obtaining the sample feature vector required by the feature vector database, thereby improving the security of the user's privacy data. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0012] Figure 1 A schematic diagram of an example of an application scenario of the method for updating a feature vector database provided in an embodiment of the present application;

[0013] Figure 2 A flowchart of an embodiment of a method for updating a feature vector database provided by the present application;

[0014] Figure 3 A schematic diagram of an example of the angle of a feature vector in a high-dimensional space provided in an embodiment of the present application;

[0015] Figure 4 A flowchart of another embodiment of the method for updating the feature vector database provided by the present application;

[0016] Figure 5 A flowchart of another embodiment of the method for updating the feature vector database provided by the present application;

[0017] Figure 6 A flowchart of another embodiment of the method for updating the feature vector database provided by the present application;

[0018] Figure 7 A schematic diagram of the structure of an embodiment of a device for updating a feature vector database provided by the present application;

[0019] Figure 8 A schematic structural diagram of another embodiment of a device for updating a feature vector database provided by the present application;

[0020] Fig. 9 A structural schematic diagram of another embodiment of the device for updating the feature vector database provided by the present application;

[0021] Fig.10 A schematic structural diagram of yet another embodiment of the device for updating the feature vector database provided by the present application;

[0022] Fig.11 A schematic structural diagram of an embodiment of a device for updating a feature vector database provided in the present application. DETAILED DESCRIPTION

[0023] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0024] Biometric identification technology is a technology that uses the physiological or behavioral characteristics inherent in the human body to identify individuals. The application scope of biometric identification technology is very wide. For example, it can be applied to payment, access, attendance and other fields. When the network structure on which biometric identification technology relies changes or the biometric identification algorithm needs to be optimized, the algorithm model of biometric identification needs to be updated and upgraded, that is, migrated from one version of the algorithm model to another version of the algorithm model. However, the feature vectors of biometrics obtained by different versions of algorithm models will be different. For the same biometric, the feature vector obtained by one version of the algorithm model may not be used in another version of the algorithm model. In order to ensure that the updated and upgraded algorithm model can be used, the corresponding feature vector database can be generated for different versions of the algorithm model. The feature vectors in the feature vector database corresponding to the updated and upgraded algorithm model can be generated based on the original biometrics when the user registered. However, if the original biometrics when the user registered are to be obtained, the original biometrics of the user need to be stored after the user registered. However, the original biometrics belong to the user's privacy data, and the security risk of storing the original biometrics of the user is relatively high. In other words, the process of obtaining the feature vectors required for the new feature vector database will bring greater security risks to the user's privacy data.

[0025] For ease of explanation, in the embodiment of the present application, two algorithm models with different versions are referred to as the first algorithm model and the second algorithm model. The version of the first algorithm model is different from the version of the second algorithm model. The first algorithm model is the algorithm model used previously, and the first algorithm model is configured with a first feature vector database, and the first feature vector database includes relatively complete sample feature vectors of users. The sample feature vectors in the first feature vector database can be used as samples for identifying users in the process of biometric identification using the first algorithm model. The second algorithm model can be a migrated algorithm model. For example, the first algorithm model is the algorithm model before the update and upgrade, and the second algorithm model is the algorithm model after the update and upgrade. The second algorithm model is configured with a second feature vector database. The sample feature vectors in the second feature vector database can be used as samples for identifying users in the process of biometric identification using the second algorithm model. The sample feature vectors in the second feature vector database can be obtained using the feature vector database update method provided in the embodiment of the present application. Before obtaining the second feature vector database including relatively complete sample features of users corresponding to the second algorithm model, the first algorithm model and the first feature vector database can be used for biometric identification. After obtaining the second feature vector database including relatively complete sample features of users corresponding to the second algorithm model, the second algorithm model and the second feature vector database can be used for biometric identification.

[0026] The method for updating a feature vector database provided in an embodiment of the present application can enable two different algorithm models of biometric feature recognition technology to run in parallel. In this process, feature vectors are generated in real time according to the biometric features in the user's service request. The feature vectors obtained by the two different algorithm models and at least part of the historical feature vectors can be used to verify the validity of the feature vectors obtained by the algorithm model that has not yet been fully launched. If the validity verification passes, the feature vector is stored as the user's sample feature vector in the unfinished feature vector database. In the process of obtaining the sample feature vector in the feature vector database, it is not necessary to store the original biometric features registered by the user, thereby ensuring that the user's privacy data will not be at risk in the process of obtaining the sample feature vector required by the feature vector database, thereby improving the security of the user's privacy data.

[0027] The updating method of the feature vector database provided in the embodiment of the present application can be applied to the migration process from an algorithm model that has not been updated or upgraded to an algorithm model that has been updated or upgraded. Figure 1 A schematic diagram of an example of an application scenario of the method for updating a feature vector database provided in an embodiment of the present application. Figure 1 As shown, the method for updating the feature vector database provided in the embodiment of the present application may involve a terminal device 11 and a server 12 .

[0028] The terminal device 11 is a front-end device that provides biometric identification services and can collect biometric features of users. The terminal device 11 can interact with the server 12. In some examples, the algorithm model can be set in the terminal device 11, and the terminal device 11 generates a feature vector based on the user's biometric features, sends the feature vector to the server 12, and the server 12 performs operations such as storage, matching, and identification on the feature vector. In other examples, the terminal device 11 sends the collected biometric features of the user to the server 12, and the server 12 generates a feature vector and performs operations such as storage, matching, and identification on the feature vector.

[0029] The server 12 is a background device for performing biometric identification, and can perform operations such as storage, matching, and identification on feature vectors generated based on biometrics. In some examples, the server 12 can receive a biometric vector generated by the terminal device 11 based on the user's biometrics. In other examples, the server 12 can receive the user's biometrics sent by the terminal device 11, and generate a corresponding feature vector based on the biometrics.

[0030] In some examples, the terminal device 11 and the server 12 may also be integrated into a device or a system to jointly implement the method for updating the feature vector database in the embodiment of the present application.

[0031] The present application provides a method for updating a feature vector database, which can be applied to a scenario where a biometric feature recognition service migrates from one version of an algorithm model to another version of an algorithm model. The method for updating a feature vector database can be executed by a feature vector database update device, which is not limited here. Figure 2 This is a flow chart of an embodiment of a method for updating a feature vector database provided by the present application. Figure 2 As shown, the method for updating the feature vector database may include steps S201 to S202.

[0032] In step S201, a first biometric feature in a service request is obtained.

[0033] During the migration from the first algorithm model to the second algorithm model, the first algorithm model and the second algorithm model run in parallel, and biometric identification can be performed normally. If the migration is not completed, the first algorithm model and the first feature vector database can be used to perform biometric identification of the user. During the migration from the first algorithm model to the second algorithm model, the first biometric in the business request is obtained. The business request is a request for a business that requires biometric identification. For example, the business request may include a payment business request using biometric identification, a pass business request using biometric identification, an attendance business request using biometric identification, etc., which are not limited here.

[0034] In the embodiment of the present application, the type of the first biometric feature is not limited. For example, the first biometric feature may include face, palm print, fingerprint, iris, voice print, etc.

[0035] In step S202, according to the first biological feature, a first feature vector and a second feature vector are obtained respectively through a first algorithm model and a second algorithm model.

[0036] The first algorithm model is different from the second algorithm model in version. In the embodiment of the present application, the first algorithm model is migrated to the second algorithm model. Correspondingly, the first algorithm model is the algorithm model before migration, and the second algorithm model is the algorithm model after migration. The first feature vector database is adapted to the first algorithm model. The first feature vector database includes sample feature vectors obtained based on the first algorithm model. The sample feature vector is a sample used for biometric recognition.

[0037] The first biometric feature is input into the first algorithm model, and the first algorithm model can output a first feature vector corresponding to the first biometric feature, that is, the first feature vector is a feature vector obtained based on the first algorithm model. Similarly, the first biometric feature is input into the second algorithm model, and the second algorithm model outputs a second feature vector corresponding to the first biometric feature, that is, the second feature vector is a feature vector obtained based on the second algorithm model.

[0038] In some embodiments, in order to ensure the security of user biometric identification, liveness detection may be performed on the user requesting the service. In some examples, after the liveness detection is passed, the first feature vector and the second feature vector may be obtained through the first algorithm model and the second algorithm model according to the first biometric. In other examples, after the liveness detection is passed, the validity verification of the second feature vector may be continued.

[0039] In step S203, the validity of the second feature vector is verified according to the associated feature vector of the first user corresponding to the first sample feature vector.

[0040] The first sample feature vector is a sample feature vector that successfully matches the first feature vector in the first feature vector database. Specifically, after obtaining the first feature vector, a sample feature vector that successfully matches the first feature vector can be searched in the first feature vector database. Successful matching of two feature vectors means that the similarity of the two feature vectors is greater than or equal to a preset judgment threshold. The judgment threshold is a threshold value used to determine that the users indicated by the two feature vectors are the same user. If the similarity of the two feature vectors is greater than or equal to the preset judgment threshold, the two feature vectors are successfully matched, indicating that the users indicated by the two feature vectors are the same user. The first feature vector database stores the correspondence between the sample feature vector and the user, and the first user corresponding to the first sample feature vector can be determined based on the first sample feature vector.

[0041] The associated feature vector of the first user includes a feature vector obtained based on the biometric feature of the first user. The associated feature vector includes at least part of the following: the first feature vector, the second feature vector, and at least part of the historical feature vector corresponding to the first algorithm model. At least part of the historical feature vector corresponding to the first algorithm model is at least part of the historical feature vector obtained by the first algorithm model. Correspondingly, the associated feature vector of the first user includes at least part of the following: the first feature vector corresponding to the first biometric feature, the second feature vector corresponding to the first biometric feature, and at least part of the historical feature vector of the first user generated by the first algorithm model.

[0042] The first user's associated features can be verified individually or in association to determine whether the second biometric feature vector corresponding to the first biometric feature is a valid feature vector for the first user, that is, to verify the validity of the second feature vector. If the validity verification is successful, it means that the second biometric feature vector corresponding to the first biometric feature is a valid feature vector for the first user. If the validity verification is successful, it means that the second biometric feature vector corresponding to the first biometric feature may not be a valid feature vector for the first user, and further verification can be performed.

[0043] In some examples, validity verification may include one or more of the following: reliability verification, first difference verification, and second difference verification.

[0044] Reliability verification can be the verification of the reliability of the second feature vector using the first feature vector generated based on the same biometric feature. The first difference verification can be the verification of the error difference between the second feature vector and the first feature vector using the error offset of the second algorithm model and the error offset of the first algorithm model. The second difference verification can be the verification of the error difference of the second feature vector of the same user using multiple second feature vectors obtained based on multiple service requests of the same user.

[0045] In the case where the validity verification includes one of the reliability verification, the first difference verification, and the second difference verification, if this verification passes, the validity verification is considered to have passed. In the case where the validity verification includes the reliability verification, and the validity verification also includes at least one of the first difference verification and the second difference verification, the reliability verification can be given priority. If the reliability verification fails, but at least one of the first difference verification and the second difference verification included in the validity verification passes, the validity verification can also be considered to have passed. In the case where the validity does not include the reliability verification, and the validity verification also includes at least one of the first difference verification and the second difference verification, the validity verification can only be considered to have passed if the first difference verification and the second difference verification included in the validity verification pass.

[0046] The reliability verification, the first difference verification and the second difference verification are described in detail below.

[0047] Reliability verification requires associated feature vectors including the first feature vector and the historical feature vector corresponding to the first algorithm model. The historical feature vector corresponding to the first algorithm model of the first user is the feature vector of the first user generated by the first algorithm model before the first feature vector of the first user is generated this time. Reliability verification may include steps a1 and a2.

[0048] Step a1, respectively calculating a first similarity between the first feature vector and at least a portion of the historical feature vectors corresponding to the first algorithm model.

[0049] At least part of the historical feature vectors corresponding to the first algorithm model may include N feature vectors generated by the first algorithm model based on the biometric features in the most recent N acquired service requests. N is a positive integer, and the value of N can be set according to scenarios, requirements, experience, etc., and is not limited here. For example, N can be 5.

[0050] The first similarity is the similarity between the first feature vector and the historical feature vector. A first similarity can be calculated between the first feature vector and each of at least some of the historical feature vectors. For example, at least some of the historical feature vectors corresponding to the first algorithm model include 5 historical feature vectors A1, historical feature vector A2, historical feature vector A3, historical feature vector A4 and historical feature vector A5, then the first similarity between the first feature vector and the historical feature vector A1, the first similarity between the first feature vector and the historical feature vector A2, the first similarity between the first feature vector and the historical feature vector A3, the first similarity between the first feature vector and the historical feature vector A4 and the first similarity between the first feature vector and the historical feature vector A5 can be calculated.

[0051] Step a2: If there is no first similarity less than the first determination threshold, it is determined that the reliability verification is passed.

[0052] The first judgment threshold is a threshold value for judging whether the user corresponding to the first feature vector and the user corresponding to the historical feature vector are the same user, which can be set according to scenarios, needs, experience, etc., and is not limited here. If the first similarity is greater than or equal to the first judgment threshold, it means that the user corresponding to the first feature vector and the user corresponding to the historical feature vector are the same user. If the first similarity is less than the first judgment threshold, it means that the user corresponding to the first feature vector and the user corresponding to the historical feature vector are not the same user. In the embodiment of the present application, the reliability verification is determined to have passed only when all first similarities are greater than or equal to the first judgment threshold.

[0053] If there is a first similarity threshold value that is smaller than the first determination threshold value, it is determined that the reliability verification has failed.

[0054] The first difference verification is performed, and the required associated feature vectors include the first feature vector and the second feature vector. The first difference verification may include step b1 and step b2.

[0055] Step b1, using a plurality of second feature vectors obtained based on a plurality of service requests of the first user, calculate and obtain the maximum angle between the plurality of second feature vectors.

[0056] Each time a second feature vector is obtained through a second algorithm model based on the biometric features in a service request of the first user, the second feature vector can be cached. When the number of second feature vectors of the same user reaches the verification requirement, the second feature vectors that reach the verification requirement are used to calculate the angle between every two second feature vectors, and the maximum angle among the multiple angles is selected.

[0057] Figure 3 A schematic diagram of an example of the angle of a feature vector in a high-dimensional space provided in an embodiment of the present application. Figure 3 Four eigenvectors are shown, namely, eigenvector 1, eigenvector 2, eigenvector 3 and eigenvector 4. Figure 4 As shown, there is an angle between each of the eigenvector 1, the eigenvector 2, the eigenvector 3 and the eigenvector 4, wherein the maximum angle between the multiple eigenvectors is the angle between the eigenvector 1 and the eigenvector 4.

[0058] To calculate the angle between two eigenvectors, the cosine of the angle between the two eigenvectors can be calculated first, and then the angle can be calculated based on the cosine of the angle. For example, the cosine of the angle between the two eigenvectors can be calculated based on the following relationship (1):

[0059]

[0060] in, is a feature vector; is another eigenvector; is the feature vector and the eigenvector The cosine of the angle between is the feature vector Model; is the feature vector Model.

[0061] Step b2: using the multiple first feature vectors obtained based on the multiple service requests of the first user, calculate the maximum angle between the multiple first feature vectors.

[0062] Each time a first feature vector is obtained through a first algorithm model based on the biometric features in a service request of a first user, the first feature vector can be cached. When the number of first feature vectors of the same user reaches a verification requirement, the first feature vectors that reach the verification requirement are used to calculate the angle between every two first feature vectors, and the maximum angle among the multiple angles is selected.

[0063] The calculation of the angle between the first eigenvectors can refer to the above-mentioned contents related to the calculation of the angle between the eigenvectors and the cosine of the angle, which will not be repeated here.

[0064] Step b3: If the maximum angle between the plurality of second eigenvectors is smaller than the maximum angle between the plurality of first eigenvectors, it is determined that the first difference verification is passed.

[0065] Since the second algorithm model is a migrated algorithm model, that is, the version of the second algorithm model is higher than that of the first algorithm model, the difference fluctuation between the second feature vectors of the same user obtained by the second algorithm model should be smaller than the difference fluctuation between the first feature vectors of the same user obtained by the first algorithm model, so as to control the error of the second algorithm model within the error of the first algorithm model. The angle between different feature vectors can reflect the difference fluctuation between different feature vectors. The maximum angle between the second feature vectors and the maximum angle between the first feature vectors can be compared to determine whether the error of the second algorithm model is within the error of the first algorithm model. If the maximum angle between the second feature vectors is smaller than the maximum angle between the first feature vectors, it means that the error of the second algorithm model is within the error of the first algorithm model, and it is considered that the first difference verification has been passed. If the maximum angle between the second feature vectors is larger than the maximum angle between the first feature vectors, it means that the error of the second algorithm model exceeds the error of the first algorithm model, and it is considered that the first difference verification has not been passed.

[0066] The second difference verification is performed, and the required associated feature vector includes the second feature vector. The second difference verification may include steps c1 to c3.

[0067] Step c1, obtaining multiple second feature vectors based on multiple service requests of the first user.

[0068] In the case where the second feature vector is obtained through the second algorithm model each time according to the biometric feature in the service request of the first user, the second feature vector can be cached, thereby obtaining multiple second feature vectors corresponding to the first user.

[0069] Step c2: respectively calculating the second similarity between every two second eigenvectors in the plurality of second eigenvectors.

[0070] When the number of second feature vectors corresponding to the first user obtained reaches a certain number, the similarity of the second feature vectors in the second feature vectors that reach a certain number is calculated. The second similarity is the similarity of every two second feature vectors in the plurality of second feature vectors. For example, if the plurality of second feature vectors include a second feature vector B1, a second feature vector B2, and a second feature vector B3, the similarity between the second feature vector B1 and the second feature vector B2, the similarity between the second feature vector B1 and the second feature vector B3, and the similarity between the second feature vector B2 and the second feature vector B3 can be calculated.

[0071] Step c3: if there is no second similarity less than the second determination threshold, it is determined that the second difference verification is passed.

[0072] The second determination threshold is a threshold value for determining whether the difference between multiple second feature vectors is within an acceptable range, which can be set according to scenarios, requirements, experience, etc., and is not limited here. If the second similarity of two second feature vectors is greater than or equal to the second determination threshold, it means that the difference between the two second feature vectors is within an acceptable range. If the second similarity of the two second feature vectors is less than the second determination threshold, it means that the difference between the two second feature vectors is large and is not within an acceptable range. There may be a problem with the collected biometrics or fluctuations in the second algorithm model. In an embodiment of the present application, the second difference verification is determined to be passed only when all second similarities are greater than or equal to the second determination threshold.

[0073] In step S204, when the validity verification of the second feature vector is passed, a second sample feature vector of the first user is obtained based on the second feature vector, and the second sample feature vector is stored in a second feature vector database configured by the second algorithm model.

[0074] The validity verification of the second feature vector passes, indicating that the second feature vector is valid, and a second sample feature vector can be obtained according to the second feature vector. The second sample feature vector can be stored in the second feature vector database as a sample feature vector of the first user in the second feature vector database.

[0075] In some examples, the second feature vector may be directly determined as the second sample feature vector. For example, the validity verification includes reliability verification, and if the reliability verification passes, the second feature vector corresponding to the first feature vector may be directly used as the second sample feature vector of the first user.

[0076] In some examples, a mean vector of multiple second feature vectors obtained based on multiple service requests of the first user may be calculated, and the mean vector may be determined as the second sample feature vector. The calculation of the mean vector of multiple second feature vectors is the same as the calculation of the mean vector of multiple feature vectors, and the mean vector of the feature vector may be calculated according to the following equations (2) to (4):

[0077] X i =(x i1 ,x i2 ,…,x ip ) T ,i=1,2,…,N (2)

[0078]

[0079]

[0080] Among them, X i is the i-th eigenvector; x i1 is the first element in the i-th eigenvector; x ip is the pth element in the i-th eigenvector; is the mean vector; is the jth element in the mean vector.

[0081] For example, when the validity verification includes the first difference verification and / or the second difference verification, if the validity verification passes, the mean vector of multiple second feature vectors obtained based on multiple service requests of the first user can be used as the second sample feature vector of the first user.

[0082] In some other examples, the second feature vector most recently obtained among the multiple second feature vectors obtained from the multiple service requests of the first user may be determined as the second sample feature vector. For example, in the case where the validity verification includes the first difference verification and / or the second difference verification, if the validity verification passes, the second feature vector most recently obtained among the multiple second feature vectors obtained based on the multiple service requests of the first user may be used as the second sample feature vector of the first user.

[0083] The above steps S201 to S204 may be repeated until the users corresponding to the sample feature vectors in the second feature vector database are complete or approaching complete, that is, the update of the feature vector database is completed, and the feature vector database is migrated from the first feature vector database to the second feature vector database.

[0084] In an embodiment of the present application, in the process of two different versions of algorithm models running in parallel, the first feature vector can be generated in real time by the first algorithm model according to the biometric features in the user's business request, and the second feature vector can be generated in real time by the second algorithm model. The feature vector obtained by the algorithm model that is not fully online in the two algorithm models, that is, the second algorithm model, can be verified for validity based on at least part of the first feature vector, the second feature vector, and the historical feature vector. If the validity verification passes, the feature vector is stored as the user's sample feature vector in the unfinished feature vector database, that is, the second feature vector database. In the process of obtaining the sample feature vector in the second feature vector database, it is not necessary to store the original biometric features registered by the user, thereby ensuring that the user's privacy data will not be at risk in the process of obtaining the sample feature vector required by the feature vector database, thereby improving the security of the user's privacy data.

[0085] In the above process, the user is unaware of the update of the feature vector database. The update of the feature vector database can be completed without disturbing the user or increasing the user operation, thereby realizing the migration of sample feature vectors and improving the user experience.

[0086] Through validity verification, such as reliability verification, first difference verification, second difference verification, etc., the accuracy of the sample feature vectors in the updated feature vector database, ie, the second feature vector database, can be guaranteed.

[0087] Moreover, as time goes by, some of the user's biometric features may change. In the related technology of storing the original biometric features registered by the user and obtaining the sample feature vectors used for biometric identification based on the original biometric features, misidentification may occur. For example, the user's biometric identification uses facial features, but the user is a teenager. Due to the developmental changes of teenagers, the facial features of the teenagers may be quite different from the original facial features used for registration, which may cause problems in biometric identification. In an embodiment of the present application, a sample feature vector in a feature vector database used for biometric identification is obtained based on the feature vector of the biometric in the user's business request, and the user's business request is the most recent business request, thereby avoiding misidentification caused by the data in the feature vector database used for biometric identification being too old, thereby improving the accuracy of biometric identification.

[0088] In some embodiments, in addition to service requests where the user is an unknown user and 1:N biometric identification is required, service requests where the user is a known user and 1:1 biometric identification is required, such as registration service requests, login service requests, etc. In the case where the user of the service request is a known user, if the first feature vector successfully matches the sample feature vector of the known user in the first feature vector database, the second feature vector is stored in the second feature vector database as the sample feature vector of the known user in the second feature vector database.

[0089] Since the user who makes the business request for 1:1 biometric identification is a known user, the biometrics in the business request can be regarded as a reliable and valid biometric. Similarly, in this case, if the first feature vector obtained based on the biometrics in the business request is successfully matched, it can be guaranteed that the second feature vector obtained based on the biometrics in the business request is reliable and valid, and can be directly added to the second feature vector database as a sample feature vector of the known user.

[0090] In some examples, in order to further ensure the security of biometric identification of known users, liveness detection may be performed on the known users. If the liveness detection passes and the first feature vector successfully matches the sample feature vector of the known user in the first feature vector database, the second feature vector may be stored in the second feature vector database.

[0091] In some examples, if the above validity verification fails, an auxiliary operation may be performed to determine whether the second sample feature vector can be used as a sample feature vector in the second feature vector database. Figure 4 This is a flowchart of another embodiment of the method for updating the feature vector database provided by the present application. Figure 4 and Figure 2 The difference is that Figure 4 The method for updating the feature vector database shown may further include steps S205 to S208.

[0092] In step S205, if the validity verification of the second feature vector fails, a first auxiliary verification request is initiated.

[0093] The first auxiliary verification request may be used to prompt the user to enter auxiliary information.

[0094] In step S206, auxiliary information input in response to the first auxiliary authentication request is received.

[0095] The auxiliary information may be used to assist in determining the validity of the second feature vector. The auxiliary information may specifically include information that enables identity authentication. In some examples, the auxiliary information includes one or more of the following: a text message verification code, a user terminal token, user identity information, and a second biometric feature.

[0096] The user terminal token is a dynamic password used for identity authentication. User identity information may include user mobile phone number, user email address, etc., which are not limited here. The type of the second biometric feature is different from the type of the first biometric feature. For example, the first biometric feature is a face, and the second biometric feature is a palm print. For another example, the first biometric feature is a face, and the second biometric feature is a voice print.

[0097] In step S207, identity authentication is performed using the auxiliary information to obtain an identity authentication result.

[0098] The identity authentication result may represent whether the user indicated by the auxiliary information is the first user.

[0099] For example, the auxiliary information includes a text message verification code and / or a user terminal token, and the identity verification result can be generated based on whether the received text message verification code and / or user terminal token is consistent with the text message verification code and / or user terminal token sent to the first user. If they are consistent, the identity verification result indicates that the user indicated by the auxiliary information is the first user; if they are inconsistent, the identity verification result indicates that the user indicated by the auxiliary information is not the first user.

[0100] For another example, the auxiliary information includes user identity information, and the identity authentication result can be generated based on whether the user identity information is consistent with the registration information of the first user. If they are consistent, the identity authentication result indicates that the user indicated by the auxiliary information is the first user; if they are inconsistent, the identity authentication result indicates that the user indicated by the auxiliary information is not the first user.

[0101] For another example, if the auxiliary information includes a second biometric feature, an identity verification result may be generated based on whether the user obtained through identity verification using the second biometric feature is the first user. If the user obtained through identity verification using the second biometric feature is the first user, the identity verification result indicates that the user indicated by the auxiliary information is the first user; if the user obtained through identity verification using the second biometric feature is not the first user, the identity verification result indicates that the user indicated by the auxiliary information is not the first user.

[0102] In step S208, if the identity authentication result indicates that the user indicated by the auxiliary information is the first user, a second sample feature vector of the first user is obtained based on the second feature vector, and the second sample feature vector is stored in a second feature vector database configured by the second algorithm model.

[0103] The identity authentication result indicates that the user indicated by the auxiliary information is the first user, indicating that the second feature vector is valid, and based on the second feature vector, a characteristic vector can be obtained and stored in the second feature vector database as a sample feature vector of the first user.

[0104] In some embodiments, after the second sample feature vector obtained based on the second feature vector is stored in the second feature vector database, the validity of the second sample feature vector may be further verified and updated in a timely manner when a service request is subsequently obtained. Figure 5 This is a flowchart of another embodiment of the method for updating the feature vector database provided by the present application. Figure 5 and Figure 2 The difference is that Figure 5 The method for updating the feature vector database shown may further include steps S209 to S214.

[0105] In step S209, the third biometric feature in the newly received service request is obtained.

[0106] The third biometric feature is a biometric feature in a newly received service request, and the type of the third biometric feature is the same as the type of the first biometric feature.

[0107] In step S210, based on the third biological feature, a first feature vector and a second feature vector corresponding to the third biological feature are obtained through a first algorithm model and a second algorithm model respectively.

[0108] According to the third biological feature, the first feature vector corresponding to the third biological feature is obtained through the first algorithm model. According to the third biological feature, the second feature vector corresponding to the third biological feature is obtained through the second algorithm model.

[0109] In step S211, a third sample feature vector matching the first feature vector corresponding to the third biological feature is searched in the first feature vector database, and a second user corresponding to the third sample feature vector is determined.

[0110] The similarity between the third sample feature vector and the first feature vector corresponding to the third biometric feature is greater than a preset determination threshold. Based on the first feature vector corresponding to the third biometric feature, biometric identification is performed using the first feature vector database, and the second user is a user identified by biometric identification in the first feature vector database.

[0111] In step S212, when the sample feature vector of the second user exists in the second feature vector database, the second feature vector corresponding to the third biometric feature is used for matching in the second feature vector database.

[0112] If the second feature vector database contains a sample feature vector of the second user, a 1:N biometric identification can be performed in the second feature vector database using the second feature vector corresponding to the third biometric, i.e., matching is performed in the second feature vector database using the second feature vector corresponding to the third biometric to verify the validity of the sample feature vector of the second user that has been stored in the second feature vector database.

[0113] In step S213, if the second feature vector corresponding to the third biometric feature in the second feature vector database successfully matches the sample feature vector of the second user, it is determined that the sample feature vector of the second user is valid.

[0114] The second feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature is the sample feature vector of the second user, indicating that the 1:N biometric identification using the second feature vector corresponding to the third biometric feature is successful and correct, and it can be determined that the sample feature vector of the second user in the second feature vector database is valid.

[0115] In step S214, if there is no sample feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature, or if the second feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature is not the sample feature vector of the second user, the sample feature vector of the second user is deleted from the second feature vector database, and business requests including the biometric features of the second user continue to be obtained to obtain the sample feature vector of the second user stored in the second feature vector database again.

[0116] There is no sample feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature, indicating that the 1:N biometric identification using the second feature vector corresponding to the third biometric feature has failed. It can be determined that the sample feature vector of the second user in the second feature vector database is invalid.

[0117] The second feature vector in the second feature vector database that corresponds to the third biometric feature successfully matches but is not the sample feature vector of the second user, indicating that the 1:N biometric recognition performed using the second feature vector corresponding to the third biometric feature is wrong. It can be determined that the sample feature vector of the second user in the second feature vector database is invalid.

[0118] When the sample feature vector of the second user in the second feature vector database is invalid, the sample feature vector of the second user is deleted from the second feature vector database, and the relevant steps in steps S201 to S208 in the above embodiment can be performed again to obtain a new sample feature vector of the second user stored in the second feature vector database.

[0119] In some embodiments, whether to fully launch the second algorithm model can be determined based on the number of users of the sample feature vectors in the second feature vector database to end the migration from the first algorithm model to the second algorithm model. Figure 6 This is a flowchart of another embodiment of the method for updating the feature vector database provided by the present application. Figure 6 and Figure 2 The difference is that Figure 6 The method for updating the feature vector database shown may further include steps S215 to S217.

[0120] In step S215, when the proportion of users with sample feature vectors in the second feature vector database to users with sample feature vectors in the first feature vector database is less than a preset completion coverage rate, the first algorithm model and the first feature vector database are used to perform biometric identification of the business request.

[0121] The preset completion coverage rate is a threshold value for determining whether users corresponding to the first feature vector database have been basically migrated to the second feature vector database, and can be set according to scenarios, requirements, experience, etc., and is not limited here.

[0122] The ratio of users with sample feature vectors in the second feature vector database to users with sample feature vectors in the first feature vector database is the real-time coverage rate. The real-time coverage rate is less than the preset completion coverage rate, indicating that the number of users with sample feature vectors in the second feature vector database is not enough to support the biometric identification service. In this case, the biometric identification of the service request is still performed using the first algorithm model and the first feature vector database.

[0123] In some examples, in the above embodiments, when the validity needs to be verified based on multiple second feature vectors, if the number of second feature vectors does not meet the verification requirement, the sample feature vector of the third user corresponding to the second feature vector in the first feature vector database may be retained. When the third user initiates a service request, the first algorithm model and the first feature vector database are used to perform biometric identification of the service request to avoid the situation where the biometric identification of the third user's service request cannot be performed when the sample feature vector of the third user has not yet been stored in the second feature vector.

[0124] In step S216, when the proportion of users with sample feature vectors in the second feature vector database to users with sample feature vectors in the first feature vector database is greater than or equal to a preset completion coverage rate, the second algorithm model and the second feature vector database are used to perform biometric identification of the business request.

[0125] The real-time coverage is greater than or equal to the preset completion coverage, indicating that the number of users with sample feature vectors in the second feature vector database is sufficient to support the biometric identification service. In this case, the biometric identification of the service request can be performed using the second algorithm model and the second feature vector database.

[0126] For example, the preset completion coverage rate is 98%. When the real-time coverage rate is less than 98%, when a user's service request is received, the first algorithm model and the first feature vector database are still used for biometric identification; when the real-time coverage rate is greater than or equal to 98%, the second algorithm model and the second feature vector database can be used for biometric identification.

[0127] In step S217, if a sample feature vector matching the second feature vector corresponding to the fourth biometric feature in the business request is not found in the second feature vector database, biometric feature recognition of the business request is performed based on the fourth biometric feature using the first algorithm model and the first feature vector database.

[0128] The fourth biometric feature is a biometric feature in the service request, and a type of the fourth biometric feature is the same as a type of the first biometric feature.

[0129] In the process of biometric identification using the second algorithm model and the second feature vector database, it is also possible that the matching sample feature vector is not identified. This may be a service request initiated by a low-frequency service user. Due to the low activity frequency of the low-frequency service user, the second sample feature vector of the low-frequency service user is not obtained during the feature vector library migration process, that is, the sample feature vector of the low-frequency service user has not been stored in the second feature vector database. In this case, in order to realize the biometric identification of the low-frequency service user, the first algorithm model and the first feature vector database can be used to perform biometric identification of the service request of the low-frequency service user to ensure the comprehensiveness of biometric identification.

[0130] In some examples, if a sample feature vector matching the second feature vector corresponding to the fourth biometric feature in the service request is not found in the second feature vector database, a second auxiliary verification request may be initiated. Auxiliary information input in response to the second auxiliary verification request is received, and the auxiliary information is used to perform identity authentication of the third user obtained by biometric identification based on the fourth biometric feature to obtain an identity authentication result. If the identity authentication result indicates that the user indicated by the auxiliary information is the third user, the third sample feature vector obtained based on the second feature vector corresponding to the fourth biometric feature is stored in the second feature vector database as the sample feature vector of the third user in the second feature vector database.

[0131] The second auxiliary authentication request may be used to prompt the user to input auxiliary information. The specific contents of the auxiliary information, identity authentication, identity authentication result, etc. may refer to the relevant descriptions in the above embodiments, which will not be repeated here.

[0132] The identity authentication result characterizes that the user indicated by the auxiliary information is the third user, indicating that the second feature vector corresponding to the fourth biometric feature is valid. Based on the second feature vector corresponding to the fourth biometric feature, a characteristic vector can be obtained and stored in the second feature vector database as a sample feature vector of the third user.

[0133] The present application also provides a device for updating a feature vector database, which can be implemented as a server and / or a front-end device, but is not limited here. Figure 7 This is a schematic diagram of the structure of an embodiment of the device for updating the feature vector database provided by the present application. Figure 7 As shown, the updating device 300 for the feature vector database may include an acquisition module 301 , a calculation module 302 , a validity verification module 303 and a processing module 304 .

[0134] The acquisition module 301 may be used to acquire a first biometric feature in a service request.

[0135] The calculation module 302 may be used to obtain a first feature vector and a second feature vector respectively according to the first biometric feature through a first algorithm model and a second algorithm model.

[0136] The first algorithm model and the second algorithm model have different versions. The first algorithm model is configured with a first feature vector database. The first feature vector database includes sample feature vectors obtained based on the first algorithm model.

[0137] The validity verification module 303 may be used to verify the validity of the second feature vector according to the associated feature vector of the first user corresponding to the first sample feature vector.

[0138] The first sample feature vector is a sample feature vector in the first feature vector database that successfully matches the first feature vector. The associated feature vector includes at least part of the following: the first feature vector, the second feature vector, and at least part of the historical feature vector corresponding to the first algorithm model.

[0139] In some embodiments, validity verification may include one or more of the following: reliability verification, first difference verification, and second difference verification.

[0140] The validity verification includes reliability verification, and correspondingly, the associated feature vector includes the first feature vector and the historical feature vector corresponding to the first algorithm model.

[0141] The validity verification module 303 can perform reliability verification. Specifically, the validity verification module 303 can be used to: respectively calculate the first similarity between the first feature vector and at least part of the historical feature vector corresponding to the first algorithm model; if there is no first similarity less than the first determination threshold, determine that the reliability verification is passed.

[0142] The validity verification includes a first difference verification, and correspondingly, the associated feature vector includes a first feature vector and a second feature vector.

[0143] The effective verification module 303 can perform the first difference verification. Specifically, the effective verification module 303 can be used to: calculate the maximum angle between the multiple second feature vectors using the multiple second feature vectors obtained based on the multiple service requests of the first user; calculate the maximum angle between the multiple first feature vectors using the multiple first feature vectors obtained based on the multiple service requests of the first user; if the maximum angle between the multiple second feature vectors is less than the maximum angle between the multiple first feature vectors, it is determined that the first difference verification is passed.

[0144] The validity verification includes a second difference verification, and correspondingly, the associated feature vector includes a second feature vector.

[0145] The valid verification module 303 can perform a second difference verification. Specifically, the valid verification module 303 can be used to: obtain multiple second feature vectors obtained based on multiple service requests of the first user; respectively calculate the second similarity of each two second feature vectors in the multiple second feature vectors; if there is no second similarity less than the second determination threshold, determine that the second difference verification is passed.

[0146] The processing module 304 may be used to obtain a second sample feature vector of the first user based on the second feature vector when the validity verification of the second feature vector is passed, and store the second sample feature vector into a second feature vector database configured by the second algorithm model.

[0147] Specifically, the processing module 304 can be used to determine the second feature vector as the second sample feature vector, or, calculate the mean vector of multiple second feature vectors obtained based on multiple service requests of the first user, and determine the mean vector as the second sample feature vector, or, determine the most recently obtained second feature vector among the multiple second feature vectors obtained based on multiple service requests of the first user as the second sample feature vector.

[0148] In an embodiment of the present application, in the process of two different versions of algorithm models running in parallel, the first feature vector can be generated in real time by the first algorithm model according to the biometric features in the user's business request, and the second feature vector can be generated in real time by the second algorithm model. The feature vector obtained by the algorithm model that is not fully online in the two algorithm models, that is, the second algorithm model, can be verified for validity based on at least part of the first feature vector, the second feature vector, and the historical feature vector. If the validity verification passes, the feature vector is stored as the user's sample feature vector in the unfinished feature vector database, that is, the second feature vector database. In the process of obtaining the sample feature vector in the second feature vector database, it is not necessary to store the original biometric features registered by the user, thereby ensuring that the user's privacy data will not be at risk in the process of obtaining the sample feature vector required by the feature vector database, thereby improving the security of the user's privacy data.

[0149] In the above process, the user is unaware of the update of the feature vector database. The update of the feature vector database can be completed without disturbing the user or increasing the user operation, thereby realizing the migration of sample feature vectors and improving the user experience.

[0150] Through validity verification, such as reliability verification, first difference verification, second difference verification, etc., the accuracy of the sample feature vectors in the updated feature vector database, ie, the second feature vector database, can be guaranteed.

[0151] Moreover, as time goes by, some of the user's biometric features may change. In the related technology of storing the original biometric features registered by the user and obtaining the sample feature vectors used for biometric identification based on the original biometric features, misidentification may occur. For example, the user's biometric identification uses facial features, but the user is a teenager. Due to the developmental changes of teenagers, the facial features of the teenagers may be quite different from the original facial features used for registration, which may cause problems in biometric identification. In an embodiment of the present application, a sample feature vector in a feature vector database used for biometric identification is obtained based on the feature vector of the biometric in the user's business request, and the user's business request is the most recent business request, thereby avoiding misidentification caused by the data in the feature vector database used for biometric identification being too old, thereby improving the accuracy of biometric identification.

[0152] In some embodiments, when the user requesting the service is a known user, if the first feature vector successfully matches the sample feature vector of the known user in the first feature vector database, the second feature vector is stored in the second feature vector database as the sample feature vector of the known user in the second feature vector database.

[0153] In some embodiments, when the validity verification fails, the auxiliary information may be used to collaboratively confirm the validity of the second feature vector. Figure 8 A schematic structural diagram of another embodiment of a device for updating a feature vector database provided in the present application. Figure 8 and Figure 7 The difference is that Figure 8 The feature vector database updating device 300 shown may further include a request initiating module 305 , a receiving module 306 and an identity authentication module 307 .

[0154] The request initiating module 305 may be used to initiate a first auxiliary verification request when the validity verification of the second feature vector fails.

[0155] The receiving module 306 may be configured to receive auxiliary information input in response to the first auxiliary verification request.

[0156] In some examples, the auxiliary information includes one or more of the following: a text message verification code, a user terminal token, user identity information, and a second biometric feature.

[0157] The type of the second biometric feature is different from the type of the first biometric feature.

[0158] The identity authentication module 307 may be used to perform identity authentication using the auxiliary information to obtain an identity authentication result.

[0159] The processing module 308 can also be used to obtain a second sample feature vector of the first user based on the second feature vector if the identity authentication result indicates that the user indicated by the auxiliary information is the first user, and store the second sample feature vector in a second feature vector database configured by the second algorithm model.

[0160] In some embodiments, after the second sample feature vector obtained based on the second feature vector is stored in the second feature vector database, the validity of the second sample feature vector may be further verified and updated in a timely manner when a service request is subsequently obtained. Fig. 9 A schematic structural diagram of another embodiment of the device for updating the feature vector database provided in the present application. Fig. 9 and Figure 7 The difference is that Fig. 9 The device 300 for updating the feature vector database shown may further include a recognition module 309 .

[0161] The acquisition module 301 may also be used to acquire a third biometric feature in a newly received service request.

[0162] The calculation module 302 may also be used to obtain a first feature vector and a second feature vector corresponding to the third biometric feature through a first algorithm model and a second algorithm model, respectively, according to the third biometric feature.

[0163] The identification module 309 can also be used to search the first feature vector database for a third sample feature vector that matches the first feature vector corresponding to the third biometric feature, and determine the second user corresponding to the third sample feature vector; and, when the second feature vector database contains a sample feature vector of the second user, use the second feature vector corresponding to the third biometric feature to perform a match in the second feature vector database.

[0164] The above-mentioned processing module 304 can also be used to determine that the sample feature vector of the second user is valid if the second feature vector corresponding to the third biometric feature in the second feature vector database is successfully matched with the sample feature vector of the second user; and, if there is no sample feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature, or the second feature vector corresponding to the third biometric feature in the second feature vector database is successfully matched but is not the sample feature vector of the second user, delete the sample feature vector of the second user from the second feature vector database, and continue to obtain the business request including the biometric feature of the second user to obtain the sample feature vector of the second user stored in the second feature vector database again.

[0165] In some embodiments, the above-mentioned identification module 309 can also be used for: when the proportion of users whose sample feature vectors exist in the second feature vector database to users whose sample feature vectors exist in the first feature vector database is less than a preset completion coverage rate, using the first algorithm model and the first feature vector database to perform biometric identification of business requests; when the proportion of users whose sample feature vectors exist in the second feature vector database to users whose sample feature vectors exist in the first feature vector database is greater than or equal to the preset completion coverage rate, using the second algorithm model and the second feature vector database to perform biometric identification of business requests.

[0166] In some examples, the processing module 304 may also be used to: if the number of second feature vectors does not reach the verification requirement, retain the sample feature vector of the third user corresponding to the second feature vector in the first feature vector database.

[0167] The identification module 309 may also be used to: when a third user initiates a service request, perform biometric identification of the service request using the first algorithm model and the first feature vector database.

[0168] In some examples, the above-mentioned identification module 309 is also used to perform biometric identification of the business request based on the fourth biometric feature using the first algorithm model and the first feature vector database if a sample feature vector matching the second feature vector corresponding to the fourth biometric feature in the business request is not found in the second feature vector database.

[0169] In some embodiments, Fig. 9 The structure of the updating device 300 of the feature vector database shown in FIG. Figure 8 Combined with the structure of the feature vector database updating device 300 shown in the figure, when a sample feature vector matching the second feature vector corresponding to the fourth biometric feature in the business request is not found in the second feature vector database, the user's sample feature vector is obtained by using auxiliary information and stored in the second feature vector database. Fig.10 This is a schematic structural diagram of yet another embodiment of the device for updating the feature vector database provided by the present application. Fig.10 and Fig. 9 The difference is that Fig.10 The updating device 300 of the feature vector database shown may also include Figure 8 The request initiating module 305, the receiving module 306 and the identity authentication module 307.

[0170] The request initiating module 305 may also be configured to initiate a second auxiliary verification request if no sample feature vector matching the second feature vector corresponding to the fourth biometric feature in the service request is found in the second feature vector database.

[0171] The receiving module 306 may also be configured to receive auxiliary information input in response to the second auxiliary verification request.

[0172] The identity authentication module 307 may also be configured to use the auxiliary information to perform identity authentication of the third user obtained by performing biometric identification based on the fourth biometric feature to obtain an identity authentication result.

[0173] The above-mentioned processing module 304 can also be used to store the third sample feature vector obtained based on the second feature vector corresponding to the fourth biometric feature into the second feature vector database as the sample feature vector of the third user in the second feature vector database if the identity authentication result indicates that the user indicated by the auxiliary information is a third user.

[0174] The present application provides a device for updating a feature vector database. Fig.11 This is a schematic diagram of the structure of an embodiment of a device for updating a feature vector database provided by the present application. Fig.11As shown, the updating device 400 for the feature vector database includes a memory 401 , a processor 402 , and a computer program stored in the memory 401 and executable on the processor 402 .

[0175] In an example, the processor 402 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0176] The memory 401 may include a read-only memory (ROM), a random access memory (RAM), a disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method for updating the feature vector database in the embodiment of the present application.

[0177] The processor 402 runs a computer program corresponding to the executable program code by reading the executable program code stored in the memory 401 , so as to implement the method for updating the feature vector database in the above embodiment.

[0178] In one example, the feature vector database updating device 400 may further include a communication interface 403 and a bus 404. Fig.11 As shown, the memory 401, the processor 402, and the communication interface 403 are connected via a bus 404 and communicate with each other.

[0179] The communication interface 403 is mainly used to realize the communication between the modules, devices, units and / or equipment in the embodiment of the present application. The input device and / or output device can also be accessed through the communication interface 403.

[0180] The bus 404 includes hardware, software or both, coupling the components of the updating device 400 of the feature vector database to each other. For example and not limitation, the bus 404 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnection, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnection, a Low Pin Count (LPC) bus, a Memory Bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-E) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus or other suitable buses or a combination of two or more of these. Where appropriate, the bus 404 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.

[0181] The present application also provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method for updating the feature vector database in the above embodiment can be implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. Among them, the above-mentioned computer-readable storage medium may include a non-transitory computer-readable storage medium, such as a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc., which is not limited here.

[0182] It should be clear that each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. For device embodiments, equipment embodiments, and computer-readable storage medium embodiments, the relevant parts can refer to the description part of the method embodiment. The present application is not limited to the specific steps and structures described above and shown in the figures. Those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application. In addition, for the sake of brevity, a detailed description of known method technologies is omitted here.

[0183] The above reference is according to the method of the embodiment of the present application, the flow chart of the device (system) and the computer program product and / or the block diagram described various aspects of the present application.It should be understood that each square box in the flow chart and / or the block diagram and the combination of each square box in the flow chart and / or the block diagram can be realized by computer program instructions.These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the realization of the function / action specified in one or more square boxes of the flow chart and / or the block diagram.Such a processor can be but is not limited to a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit.It can also be understood that each square box in the block diagram and / or the flow chart and the combination of the square boxes in the block diagram and / or the flow chart can also be realized by the dedicated hardware that performs the specified function or action, or can be realized by the combination of dedicated hardware and computer instructions.

[0184] Those skilled in the art should understand that the above embodiments are exemplary rather than restrictive. Different technical features appearing in different embodiments can be combined to achieve beneficial effects. Based on the study of the drawings, the specification and the claims, those skilled in the art should be able to understand and implement other variations of the disclosed embodiments. In the claims, the term "comprising" does not exclude other devices or steps; the quantifier "one" does not exclude multiple; the terms "first" and "second" are used to indicate names rather than to indicate any specific order. Any figure marks in the claims should not be understood as limiting the scope of protection. The functions of multiple parts appearing in the claims can be implemented by a single hardware or software module. The appearance of certain technical features in different dependent claims does not mean that these technical features cannot be combined to achieve beneficial effects.

Claims

1. A method for updating a feature vector database, characterized in that: include: Obtaining the first biometric feature in the service request; According to the first biometric feature, a first feature vector and a second feature vector are obtained respectively through a first algorithm model and a second algorithm model, the first algorithm model and the second algorithm model are different in version, the first algorithm model is configured with a first feature vector database, and the first feature vector database includes sample feature vectors obtained based on the first algorithm model; Verify the validity of the second feature vector according to the associated feature vector of the first user corresponding to the first sample feature vector, wherein the first sample feature vector is a sample feature vector that successfully matches the first feature vector in the first feature vector database, and the associated feature vector includes at least part of the following: the first feature vector, the second feature vector, and at least part of the historical feature vectors corresponding to the first algorithm model; When the validity verification of the second feature vector is passed, a second sample feature vector of the first user is obtained based on the second feature vector, and the second sample feature vector is stored in a second feature vector database configured by the second algorithm model.

2. The method according to claim 1, characterized in that The validity verification includes one or more of the following: Reliability verification, first difference verification, second difference verification.

3. The method according to claim 1, characterized in that The associated feature vector includes the first feature vector and the historical feature vector corresponding to the first algorithm model; The reliability verification includes: respectively calculating a first similarity between the first feature vector and at least a portion of the historical feature vectors corresponding to the first algorithm model; If the first similarity less than the first determination threshold does not exist, it is determined that the reliability verification is passed.

4. The method according to claim 2, characterized in that: The associated feature vector includes the first feature vector and the second feature vector; The first difference verification includes: Using a plurality of the second feature vectors obtained based on a plurality of service requests of the first user, calculating a maximum angle between the plurality of the second feature vectors; Using the multiple first feature vectors obtained based on the multiple service requests of the first user, calculating the maximum angle between the multiple first feature vectors; If the maximum angle between the plurality of second feature vectors is smaller than the maximum angle between the plurality of first feature vectors, it is determined that the first difference verification is passed.

5. The method according to claim 2, characterized in that: The associated feature vector includes the second feature vector; The second difference verification includes: Acquire a plurality of the second feature vectors obtained based on multiple service requests of the first user; respectively calculating a second similarity between every two of the second feature vectors in a plurality of the second feature vectors; If the second similarity less than the second determination threshold does not exist, it is determined that the second difference verification is passed.

6. The method according to claim 1, characterized in that Also includes: In the case where the validity verification of the second feature vector fails, initiating a first auxiliary verification request; receiving auxiliary information input in response to the first auxiliary verification request; Perform identity authentication using the auxiliary information to obtain an identity authentication result; If the identity authentication result indicates that the user indicated by the auxiliary information is the first user, a second sample feature vector of the first user is obtained based on the second feature vector, and the second sample feature vector is stored in a second feature vector database configured by the second algorithm model.

7. The method according to claim 6, characterized in that The auxiliary information includes one or more of the following: SMS verification code, user terminal token, user identity information, and second biometric feature; The type of the second biometric feature is different from the type of the first biometric feature.

8. The method according to claim 1 or 6, characterized in that: The obtaining, based on the second feature vector, a second sample feature vector of the first user includes: Determine the second feature vector as the second sample feature vector; or, Calculating a mean vector of a plurality of second feature vectors obtained based on a plurality of service requests of the first user, and determining the mean vector as the second sample feature vector; or, The second feature vector obtained most recently among multiple second feature vectors obtained through multiple service requests of the first user is determined as the second sample feature vector.

9. The method according to claim 1, characterized in that: After obtaining the first feature vector and the second feature vector respectively through the first algorithm model and the second algorithm model according to the first biometric feature, the method further includes: When the user making the service request is a known user, if the first feature vector successfully matches the sample feature vector of the known user in the first feature vector database, the second feature vector is stored in the second feature vector database as the sample feature vector of the known user in the second feature vector database.

10. The method according to claim 1, characterized in that After storing the second sample feature vector into the second feature vector database configured by the second algorithm model, the method further includes: Acquire a third biometric feature in a newly received service request; According to the third biometric feature, obtaining the first feature vector and the second feature vector corresponding to the third biometric feature through a first algorithm model and a second algorithm model respectively; Searching the first feature vector database for a third sample feature vector that matches the first feature vector corresponding to the third biometric feature, and determining a second user corresponding to the third sample feature vector; In the case where the second feature vector database has a sample feature vector of the second user, matching the sample feature vector of the second user in the second feature vector database using the second feature vector corresponding to the third biometric feature; If the second feature vector in the second feature vector database that is successfully matched with the third biometric feature is a sample feature vector of the second user, determining that the sample feature vector of the second user is valid; If there is no sample feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature, or if the second feature vector in the second feature vector database that successfully matches the second feature vector corresponding to the third biometric feature is not the sample feature vector of the second user, delete the sample feature vector of the second user from the second feature vector database, and continue to obtain business requests including the biometric features of the second user to obtain the sample feature vector of the second user stored in the second feature vector database again.

11. The method according to claim 1, characterized in that: Also includes: When the proportion of users with sample feature vectors in the second feature vector database to users with sample feature vectors in the first feature vector database is less than a preset completion coverage rate, using the first algorithm model and the first feature vector database to perform biometric recognition of the service request; When the proportion of users with sample feature vectors in the second feature vector database to users with sample feature vectors in the first feature vector database is greater than or equal to the preset completion coverage rate, the second algorithm model and the second feature vector database are used to perform biometric identification of business requests.

12. The method according to claim 11, characterized in that Also includes: If the number of the second feature vectors does not reach the verification requirement, retaining the sample feature vector of the third user corresponding to the second feature vector in the first feature vector database; When the third user initiates a service request, the first algorithm model and the first feature vector database are used to perform biometric identification of the service request.

13. The method according to claim 11, characterized in that Also includes: If a sample feature vector matching the second feature vector corresponding to the fourth biometric feature in the business request is not found in the second feature vector database, biometric feature identification of the business request is performed based on the fourth biometric feature using the first algorithm model and the first feature vector database.

14. The method according to claim 13, characterized in that After performing biometric identification of the service request according to the fourth biometric feature by using the first algorithm model and the first feature vector database, the method further includes: Initiating a second auxiliary verification request; receiving auxiliary information input in response to the second auxiliary verification request; Using the auxiliary information to perform identity authentication of a third user obtained by performing biometric identification based on the fourth biometric feature, to obtain an identity authentication result; If the identity authentication result indicates that the user indicated by the auxiliary information is the third user, the third sample feature vector obtained based on the second feature vector corresponding to the fourth biometric feature is stored in the second feature vector database as the sample feature vector of the third user in the second feature vector database.

15. A device for updating a feature vector database, characterized in that: include: An acquisition module, used for acquiring a first biometric feature in a service request; a calculation module, configured to obtain a first feature vector and a second feature vector respectively according to the first biometric feature through a first algorithm model and a second algorithm model, wherein the first algorithm model and the second algorithm model are different in version, the first algorithm model is configured with a first feature vector database, and the first feature vector database includes sample feature vectors obtained based on the first algorithm model; A validity verification module, configured to verify the validity of the second feature vector according to an associated feature vector of the first user corresponding to the first sample feature vector, wherein the first sample feature vector is a sample feature vector that successfully matches the first feature vector in the first feature vector database, and the associated feature vector includes at least part of the following: the first feature vector, the second feature vector, and at least part of the historical feature vectors corresponding to the first algorithm model; A processing module is used to obtain a second sample feature vector of the first user based on the second feature vector when the validity verification of the second feature vector is passed, and store the second sample feature vector into a second feature vector database configured by the second algorithm model.

16. A device for updating a feature vector database, characterized in that: include: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the method for updating the feature vector database according to any one of claims 1 to 14 is implemented.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the method for updating the feature vector database according to any one of claims 1 to 14 is implemented.

Images