A multi-feature library-based verification method, device and equipment

Abstract

This application provides a verification method, apparatus, and device based on a multi-feature library. The method includes: acquiring an initial image of a target object; selecting a reference base image from the base feature library based on a first similarity between the initial image and multiple base images in the base feature library; if the first similarity between the initial image and the reference base image is less than a first similarity threshold but greater than a second similarity threshold, determining whether a reference dynamic image exists in a dynamic feature library, wherein the object identifier of the reference dynamic image is associated with the object identifier of the reference base image; if a reference dynamic image exists, and the second similarity between the initial image and the reference dynamic image is greater than a third similarity threshold, then the verification of the target object is determined to be successful; the base feature library is used to store base images of basic forms, and the dynamic feature library is used to store dynamic images after morphological changes to the basic forms. Through the technical solution of this application, erroneous verification results are corrected, and the recognition accuracy is enhanced.

Description

Technical Field

[0001] This application relates to the field of access control technology, and in particular to a verification method, apparatus and equipment based on a multi-feature library. Background Technology

[0002] With the rapid development of society, economy, science, and technology, access control systems are being used in an increasing number of scenarios. In residential communities, industrial parks, and other settings, access control systems can be used to verify the identity of a target (such as a user needing to enter the community). If verification is successful, the target is allowed to enter; if verification fails, entry is denied. Access control systems, also known as entry and exit management and control systems, are intelligent management systems for managing the entry and exit of target individuals. For facial recognition access control systems (a type of access control system), to control the target, the facial features of the target need to be stored in advance. When verifying the target, an image of the target is acquired, and the facial features in that image are compared with the stored facial features. If they meet a preset comparison threshold, verification is successful.

[0003] However, when the image of the target object changes due to actions such as changing glasses or wearing a hat, the facial features of the image may fail to meet the preset comparison threshold when compared with the stored facial features, resulting in the target object failing verification and obtaining an incorrect verification result, leading to a poor user experience. Summary of the Invention

[0004] This application provides a verification method based on a multi-feature library, the method comprising:

[0005] An initial image of the target object is obtained, and a reference base image is selected from the base feature library based on the first similarity between the initial image and multiple base images in the base feature library.

[0006] If the first similarity between the initial image and the reference base image is less than the first similarity threshold and greater than the second similarity threshold, then it is determined whether there is a reference dynamic image in the dynamic feature library, and the object identifier of the reference dynamic image is associated with the object identifier of the reference base image;

[0007] If the reference dynamic image exists, and the second similarity between the initial image and the reference dynamic image is greater than the third similarity threshold, then the verification of the target object is determined to be successful.

[0008] The basic feature library is used to store basic images of basic forms, and the dynamic feature library is used to store dynamic images of forms after changes have occurred in the basic forms.

[0009] This application provides a verification device based on a multi-feature library, the device comprising:

[0010] The acquisition module is used to acquire an initial image of the target object, and select a reference base image from the base feature library based on the first similarity between the initial image and multiple base images in the base feature library;

[0011] The judgment module is used to determine whether a reference dynamic image exists in the dynamic feature library if the first similarity between the initial image and the reference base image is less than a first similarity threshold and greater than a second similarity threshold. The object identifier of the reference dynamic image is associated with the object identifier of the reference base image. The base feature library is used to store base images of basic forms, and the dynamic feature library is used to store dynamic images after morphological changes to the basic forms.

[0012] The determination module is configured to determine that the verification of the target object is successful if the reference dynamic image exists and the second similarity between the initial image and the reference dynamic image is greater than a third similarity threshold.

[0013] This application provides an electronic device, including: a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions that can be executed by the processor; the processor is used to execute the machine-executable instructions to implement the above-described verification method based on a multi-feature library.

[0014] As can be seen from the above technical solutions, in this embodiment, a basic feature library and a dynamic feature library can be maintained. The basic feature library stores basic images of basic forms, and the dynamic feature library stores dynamic images after changes in the basic form. After obtaining the initial image of the target object, if the first similarity between the initial image and the reference basic image in the basic feature library is less than a threshold, the second similarity between the initial image and the reference dynamic image in the dynamic feature library can be compared. If the second similarity is greater than the threshold, the verification of the target object is determined to be successful. Since the dynamic image is the image after changes in the basic form, even if the image of the target object changes due to actions such as changing glasses or wearing a hat, the initial image and the reference dynamic image can still meet the similarity threshold, allowing the verification of the target object to pass, i.e., obtaining the correct verification result, resulting in a good user experience. Correcting erroneous verification results through a verification method based on multiple feature libraries has higher security and enhances recognition accuracy. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings of the embodiments of this application.

[0016] Figure 1 This is a flowchart illustrating a verification method based on a multi-feature library according to one embodiment of this application.

[0017] Figure 2 This is a flowchart illustrating a verification method based on a multi-feature library according to one embodiment of this application.

[0018] Figure 3 This is a flowchart illustrating a verification method based on a multi-feature library according to one embodiment of this application.

[0019] Figure 4 This is a flowchart illustrating a verification method based on a multi-feature library according to one embodiment of this application.

[0020] Figure 5 This is a schematic diagram of the basic feature library and dynamic feature library in one embodiment of this application;

[0021] Figure 6 This is a schematic diagram of the structure of a verification device based on a multi-feature library according to one embodiment of this application;

[0022] Figure 7 This is a hardware structure diagram of an electronic device according to one embodiment of this application. Detailed Implementation

[0023] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms “a,” “the,” and “the” as used in this application and claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to any and all possible combinations comprising one or more of the associated listed items.

[0024] It should be understood that although the terms first, second, third, etc., may be used to describe various information in embodiments of this application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" may also be interpreted as "when," "when," or "in response to a determination."

[0025] This application proposes a verification method based on a multi-feature library. The multi-feature library may include, but is not limited to, a basic feature library and a dynamic feature library. The basic feature library stores a basic image of a basic form, and the dynamic feature library stores dynamic images of the basic form after morphological changes. See also... Figure 1 The diagram shown is a flowchart of the verification method based on a multi-feature library. This method may include:

[0026] Step 101: Obtain an initial image of the target object. Based on the first similarity between the initial image and multiple base images in the base feature library, select a reference base image from the base feature library.

[0027] Step 102: If the first similarity between the initial image and the reference base image is less than the first similarity threshold and greater than the second similarity threshold, then determine whether there is a reference dynamic image in the dynamic feature library, and the object identifier of the reference dynamic image is related to the object identifier of the reference base image.

[0028] Step 103: If the reference dynamic image exists, and the second similarity between the initial image and the reference dynamic image is greater than the third similarity threshold, then the verification of the target object is successful.

[0029] For example, after determining whether a reference dynamic image exists in the dynamic feature library, if the reference dynamic image exists and the second similarity between the initial image and the reference dynamic image is not greater than the third similarity threshold, then the verification count of the target object is incremented by 1; or, after determining whether a reference dynamic image exists in the dynamic feature library, if the reference dynamic image does not exist, then the verification count of the target object is incremented by 1.

[0030] After incrementing the verification count of the target object by 1, it can also be determined whether the verification count of the target object has reached the threshold for the first time (which can be configured based on experience); if yes, it is determined that the verification of the target object has failed; if no, it returns to the operation of obtaining the initial image of the target object, that is, returning to step 101.

[0031] For example, after selecting a reference base image from the base feature library based on the first similarity between the initial image and multiple base images in the base feature library, if the first similarity between the initial image and the reference base image is greater than or equal to a first similarity threshold, the verification of the target object can be determined to be successful, and the verification count of the target object can be queried. If the verification count of the target object is greater than a second threshold, and the second threshold can be less than the first threshold, the initial image can be determined as a candidate dynamic image of the target object, and the candidate dynamic image can be stored in the dynamic feature library. The object identifier of the candidate dynamic image and the object identifier of the reference base image can be associated.

[0032] For example, storing candidate dynamic images in a dynamic feature library may include, but is not limited to: counting the number of dynamic images of the target object already stored in the dynamic feature library; if the number is less than a number threshold, storing candidate dynamic images in the dynamic feature library; if the number is greater than or equal to the number threshold, deleting the dynamic images of the target object from the dynamic feature library and storing candidate dynamic images in the dynamic feature library.

[0033] Among them, the dynamic images deleted from the dynamic feature library can be the dynamic image of the target object that has been stored in the dynamic feature library for the longest time, or any dynamic image, without any restrictions.

[0034] For example, storing candidate dynamic images in the dynamic feature library may include, but is not limited to: if the total verification time of the target object is less than the failure time threshold, then candidate dynamic images can be stored in the dynamic feature library; wherein, the total verification time may be the difference between the current time and the verification start time of the target object, and the verification start time may be the time when the initial image of the target object is first obtained.

[0035] For example, after storing candidate dynamic images in the dynamic feature library, the object identifier of the candidate dynamic image can be set as the object identifier of the reference base image, so that the object identifier of the candidate dynamic image and the object identifier of the reference base image can be associated. Alternatively, after storing candidate dynamic images in the dynamic feature library, an object identifier can be set for the candidate dynamic image, and the association between the object identifier of the candidate dynamic image and the object identifier of the reference base image can be recorded in the object relationship table.

[0036] As can be seen from the above technical solutions, in this embodiment, a basic feature library and a dynamic feature library can be maintained. The basic feature library stores basic images of basic forms, and the dynamic feature library stores dynamic images after changes in the basic form. After obtaining the initial image of the target object, if the first similarity between the initial image and the reference basic image in the basic feature library is less than a threshold, the second similarity between the initial image and the reference dynamic image in the dynamic feature library can be compared. If the second similarity is greater than the threshold, the verification of the target object is determined to be successful. Since the dynamic image is the image after changes in the basic form, even if the image of the target object changes due to actions such as changing glasses or wearing a hat, the initial image and the reference dynamic image can still meet the similarity threshold, allowing the verification of the target object to pass, i.e., obtaining the correct verification result, resulting in a good user experience. Correcting erroneous verification results through a verification method based on multiple feature libraries has higher security and enhances recognition accuracy.

[0037] The technical solutions described above in this application will be explained below with reference to specific embodiments.

[0038] To control the target object, its facial features need to be stored in advance. When verifying the target object, an image of the target object can be obtained, and the facial features of the image can be compared with the stored facial features. If the comparison meets the preset threshold, the verification of the target object is successful.

[0039] However, when the image of the target object changes due to actions such as changing glasses or wearing a hat, the facial features of the image may fail to meet the preset comparison threshold when compared with the stored facial features, resulting in the target object failing verification and obtaining an incorrect verification result, leading to a poor user experience.

[0040] To address the above findings, this application proposes a verification method based on a multi-feature database. This method corrects erroneous verification results and enhances recognition accuracy. See also... Figure 2 The diagram shows a flowchart of a verification method based on a multi-feature library. This multi-feature library can include a basic feature library and a dynamic feature library. The basic feature library stores a base image of the basic form, while the dynamic feature library stores dynamic images of the form after changes have occurred.

[0041] When a new target object needs to be added, that is, when the verification of the target object can pass the access control system, the basic image of the target object can be obtained. It can be a locally entered basic image or a remotely entered basic image, and the basic image of the target object is stored in the basic feature library.

[0042] When the form of the target object changes, that is, when the appearance of the target object changes, the update process is triggered.

[0043] After the update process is triggered, the dynamic image of the target object is obtained. This dynamic image is the image after the basic shape has changed. For example, the image of the target object in its current shape is obtained and used as the dynamic image. Then, the dynamic image of the target object is stored in the dynamic feature library.

[0044] When verifying a target object, the initial image of the target object is compared with the base image in the base feature library, or the initial image of the target object is compared with the base image in the base feature library and the initial image of the target object is compared with the dynamic image in the dynamic feature library.

[0045] This application proposes a verification method based on a multi-feature library. This method can be applied to access control devices, such as the access control device acquiring an initial image of a target object and implementing a verification method based on the initial image. This method can also be applied to management devices, such as the access control device acquiring an initial image of a target object and sending the initial image to the management device, which then implements a verification method based on the initial image. There are no limitations on this application; the following explanation will use a management device as an example.

[0046] For example, the management device can maintain a basic feature library, which stores basic images of basic forms. These basic images, also known as base images, are images acquired through information collection, such as ID card images. The basic images are relatively clear, and they already contain identity information when obtained. For instance, when the management device locally inputs a basic image of a target object, it collects the target object's basic image, fills in the identity information, binds the identity information to the basic image, and stores the basic image in the basic feature library. Alternatively, when remotely inputting a basic image of a target object, the remote device collects the target object's basic image, fills in the identity information, sends the identity information and basic image to the management device, which binds the identity information to the basic image and stores the basic image in the basic feature library.

[0047] The form of the target object in the base image is called the base form, which can be understood as the normal form of the target object. For example, if there is no behavior such as changing glasses or wearing a hat, there are no restrictions on this base form.

[0048] In the above application scenarios, see Figure 3 The diagram shows a flowchart of a verification method based on a multi-feature library. This method can be applied to device management and may include the following steps:

[0049] Step 301: Obtain the initial image for the target object.

[0050] For example, when the target object is in front of the access control device, the access control device can capture an initial image of the target object (such as a camera that can capture the initial image) and send the initial image to the management device, which can then obtain the initial image of the target object. The target object can be a user who needs to enter the community, etc., and there are no restrictions on this.

[0051] Step 302: Based on the first similarity between the initial image and multiple base images in the base feature library (such as some or all base images in the base feature library), select a reference base image from the base feature library and use the identity information corresponding to the reference base image as the identity information of the target object.

[0052] For example, after obtaining the initial image of the target object, the first similarity between the initial image and the base image (such as each base image in the base feature library) can be calculated. For instance, based on the facial features of the initial image and the facial features of the base image, the first similarity between the initial image and the base image can be calculated, such as distance similarity (such as Euclidean distance), cosine similarity, etc., without any limitation.

[0053] When storing the base image through the base feature library, the facial features of the base image can also be stored. In this way, after obtaining the initial image of the target object, the facial features of the initial image can be obtained first. Then, based on the facial features of the initial image and the facial features of the base image (which are already stored in the base feature library), the first similarity between the initial image and the base image can be calculated.

[0054] For example, after obtaining the first similarity between the initial image and each base image in the base feature library, the base image corresponding to the best similarity is used as the reference base image of the target object. Taking distance similarity as an example, the base image corresponding to the minimum distance similarity is used as the reference base image.

[0055] For example, since the base image already has identity information, i.e., the reference base image has identity information, the identity information corresponding to the reference base image can be used as the identity information of the target object. For instance, the identity information of the target object can be identity A1. After obtaining the identity information of the target object, the acquisition time of the initial image can also be used as the verification start time of the target object (identity A1). That is, the verification start time can be the time when the initial image of the target object is first acquired.

[0056] Step 303: Based on the first similarity between the initial image and the reference base image, determine whether the first similarity is not greater than the second similarity threshold. If yes, proceed to step 304; otherwise, proceed to step 305.

[0057] The second similarity threshold can be configured empirically, representing the minimum similarity threshold at which the target object can continue to be verified, such as 50%, 60%, etc., and there is no restriction on it. If the first similarity is less than or equal to the second similarity threshold, the target object will no longer be verified, and step 304 will be executed; if the first similarity is greater than the second similarity threshold, the target object will continue to be verified, and step 305 will be executed.

[0058] Step 304: The verification of the target object failed.

[0059] For example, after the verification of the target object fails, the management device can also send the verification failure information to the access control device, which will then notify the target object of the verification failure. This process will not be described in detail.

[0060] Step 305: Based on the first similarity between the initial image and the reference base image, determine whether the first similarity is less than the first similarity threshold. If not, proceed to step 306; if yes, proceed to step 307.

[0061] The first similarity threshold can be configured empirically and represents the similarity threshold for identity verification. The first similarity threshold can be greater than the second similarity threshold, such as 90% or 95%, without restriction. If the first similarity is greater than or equal to the first similarity threshold, it means that the identity of the target object has been confirmed based on the basic feature library, that is, the verification of the target object based on the basic feature library is successful, and step 306 is executed. If the first similarity is less than the first similarity threshold, the verification of the target object continues, and step 307 is executed.

[0062] Step 306: Verification of the target object was successful.

[0063] For example, after the target object is successfully verified, the management device can also send the verification success information to the access control device, which will then notify the target object that the verification was successful. This process will not be described in detail here.

[0064] Step 307: Determine whether there is a reference dynamic image of the target object in the dynamic feature library, and the object identifier of the reference dynamic image is related to the object identifier of the reference base image.

[0065] For example, if the first similarity between the initial image and the reference base image is less than a first similarity threshold, and the first similarity between the initial image and the reference base image is greater than a second similarity threshold, it indicates that the identity of the target object cannot be confirmed based on the base feature library, and the identity of the target object needs to be confirmed based on the dynamic feature library. Therefore, it can be determined whether there is a reference dynamic image of the target object in the dynamic feature library. If not, step 308 can be executed; if yes, step 309 can be executed.

[0066] For example, when storing a base image through a base feature library, the object identifier of the base image (i.e., the unique identifier of the base image) can also be stored in the base feature library. Therefore, after determining the reference base image corresponding to the target object, the object identifier of the reference base image can be determined.

[0067] When storing dynamic images through a dynamic feature library, the object identifier of the dynamic image can also be stored in the dynamic feature library. Based on this, for each dynamic image in the dynamic feature library, if the object identifier of the dynamic image is associated with the object identifier of the reference base image (e.g., the two object identifiers are the same, or the association between the two object identifiers is recorded in the object relationship table), then the dynamic image is used as the reference dynamic image of the target object. If the object identifier of the dynamic image is not associated with the object identifier of the reference base image, then the dynamic image is not used as the reference dynamic image of the target object.

[0068] In summary, based on the object identifier of the reference base image, it can be determined whether there is an object identifier in the dynamic feature library that is associated with that object identifier. If it exists, the dynamic image corresponding to this object identifier is used as the reference dynamic image of the target object, meaning that a reference dynamic image of the target object exists in the dynamic feature library. If it does not exist, then a reference dynamic image of the target object does not exist in the dynamic feature library.

[0069] Step 308: Increment the number of verifications for the target object by 1, and then execute step 312.

[0070] For example, if the reference dynamic image of the target object does not exist in the dynamic feature library, it means that the identity of the target object cannot be successfully confirmed based on the dynamic feature library. Therefore, the verification count of the target object can be incremented by 1. Here, the verification count of the target object represents how many times the target object has been verified. The initial value of the verification count is a fixed value (such as 0). Each time the target object is verified, if the identity of the target object cannot be successfully confirmed, the verification count of the target object is incremented by 1.

[0071] Step 309: Based on the second similarity between the initial image and the reference dynamic image, determine whether the second similarity is greater than the third similarity threshold. If yes, proceed to step 310; otherwise, proceed to step 311.

[0072] For example, after obtaining a reference dynamic image of the target object, a second similarity between the initial image and the reference dynamic image can be calculated. For instance, based on the facial features of the initial image and the reference dynamic image, a second similarity can be calculated, such as distance similarity (e.g., Euclidean distance), cosine similarity, etc., without limitation. Furthermore, when storing dynamic images in a dynamic feature library, the facial features of the dynamic images can also be stored. Thus, after obtaining the reference dynamic image, the facial features of the reference dynamic image can be retrieved from the dynamic feature library, and then the second similarity can be calculated based on the facial features of the initial image and the facial features of the reference dynamic image.

[0073] For example, the third similarity threshold can be configured empirically to represent the similarity threshold for identity verification. The third similarity threshold can be the same as or different from the first similarity threshold, such as 90% or 95%, etc., without restriction. If the second similarity is greater than the third similarity threshold, it means that the identity of the target object has been confirmed based on the dynamic feature library, that is, the verification of the target object based on the dynamic feature library is successful, and step 310 is executed. If the second similarity is less than or equal to the third similarity threshold, it means that the identity of the target object cannot be confirmed based on the dynamic feature library, and step 311 is executed.

[0074] Step 310: Verification of the target object was successful.

[0075] For example, after the target object is successfully verified, the management device can also send the verification success information to the access control device, which will then notify the target object that the verification was successful. This process will not be described in detail here.

[0076] Step 311: Increment the number of verifications for the target object by 1, and then execute step 312.

[0077] For example, if a reference dynamic image of the target object exists in the dynamic feature library, but the second similarity is less than or equal to the third similarity threshold, it means that the identity of the target object cannot be successfully confirmed based on the dynamic feature library. Therefore, the verification count of the target object can be incremented by 1.

[0078] For example, the number of reference animated images can be at least one. If the number of reference animated images is one, that is, only one second similarity is calculated, then if the second similarity is greater than the third similarity threshold, step 310 is executed; if the second similarity is less than or equal to the third similarity threshold, step 311 is executed. If the number of reference animated images is multiple, then the second similarity between the initial image and each reference animated image is calculated, that is, there are multiple second similarities. If any second similarity is greater than the third similarity threshold, then step 310 is executed; if all second similarities are less than or equal to the third similarity threshold, then step 311 is executed.

[0079] Step 312: After incrementing the verification count of the target object by 1, determine whether the verification count of the target object has reached the first threshold. If yes, proceed to step 313; otherwise, return to step 301.

[0080] Step 313: The verification of the target object failed.

[0081] For example, if no reference dynamic image exists in the dynamic feature library, or if a reference dynamic image exists but the second similarity is less than or equal to the third similarity threshold, then the identity of the target object cannot be successfully confirmed. The verification count for the target object can be incremented by 1. Then, it is determined whether the verification count for the target object has reached the first threshold. The first threshold represents the maximum number of verifications for the target object, which can be configured empirically, such as 6, 8, or 10. If the verification count for the target object has reached the first threshold, it means that after multiple verifications, the identity of the target object has still not been successfully confirmed. Therefore, the verification of the target object can be determined as a failure, and the verification failure information is sent to the access control device. If the verification count for the target object has not reached the first threshold, although the identity of the target object has not been successfully confirmed, the verification of the target object can still continue, and the process returns to step 301 to re-execute the verification process.

[0082] After returning to step 301, the initial image of the target object can be re-acquired, and a reference base image can be selected from the base feature library based on the initial image. If the identity information corresponding to the reference base image has not changed (i.e. the reference base image has not changed), steps 301-313 can be repeated.

[0083] If the identity information corresponding to the reference base image changes (i.e., the reference base image changes), it means that the identity of the target object has changed and verification is required for the new identity of the target object. Therefore, the verification count of the target object can be reset to zero, the verification count of the target object can be recounted, and then steps 301-313 can be repeated. That is, steps 301-313 are performed for the new identity.

[0084] In one possible implementation, the management device can maintain a dynamic feature library. This library stores dynamic images of objects that have undergone morphological changes from their base form. These dynamic images can be the initial images mentioned above, meaning they are images captured by the access control device. For example, if a target object is located in front of the access control device, the device can capture a dynamic image of the target object and send it to the management device, which then stores it in the dynamic feature library. The form of the target object in the dynamic image is called the changed form, i.e., the form after a change from the base form. This can be understood as an abnormal form of the target object, such as changing glasses or wearing a hat, and is not limited in scope.

[0085] For example, when a target object undergoes a shape change, if the target object is successfully verified multiple times, the dynamic feature library update process is triggered, allowing the dynamic image of the target object to be stored in the dynamic feature library. In this embodiment, to update the dynamic feature library, the update process is only triggered when the verification of the target object is confirmed successfully based on the basic feature library, allowing the dynamic image to be stored in the dynamic feature library. If the verification of the target object is confirmed successfully based on the dynamic feature library, the update process is not triggered, and the dynamic image is not stored in the dynamic feature library.

[0086] For example, referring to steps 305 and 306, if the first similarity is greater than or equal to the first similarity threshold, the verification of the target object is determined to be successful. In this case, the dynamic feature library update process will be triggered, allowing the dynamic image to be stored in the dynamic feature library. Referring to steps 309 and 310, if the second similarity is greater than the third similarity threshold, the verification of the target object is determined to be successful. In this case, the dynamic feature library update process will not be triggered, and the dynamic image will not be allowed to be stored in the dynamic feature library.

[0087] In one possible implementation, if the first similarity is greater than or equal to a first similarity threshold, the verification of the target object is determined to be successful. In this case, the dynamic feature library update process is triggered. For the dynamic feature library update process, see [link to relevant documentation]. Figure 4 The diagram shows a flowchart of a verification method based on a multi-feature library. This method can be applied to device management and may include the following steps:

[0088] Step 401: Query the verification count of the target object. For example, since the verification count of the target object is incremented by 1 each time the target object is verified and its identity cannot be successfully confirmed, the verification count of the target object can be queried when the dynamic feature library update process is triggered.

[0089] Step 402: Determine whether the number of verifications of the target object is greater than the threshold for the second verification.

[0090] For example, the second threshold represents the storage threshold for dynamic images (i.e., the update threshold for the dynamic feature library). This second threshold can be configured empirically and can be less than the first threshold, such as 3, 4, or 5. When the number of verifications of the target object is less than or equal to the second threshold, it means that the current form of the target object has been successfully verified after only a few verification processes, and there is no need to store the dynamic image of the current form in the dynamic feature library. When the number of verifications of the target object is greater than the second threshold, it means that the current form of the target object has undergone multiple verification processes before finally being successfully verified. In this case, the dynamic image of the current form can be stored in the dynamic feature library so that the number of verifications can be reduced in the next verification.

[0091] In summary, after confirming successful verification of the target object (step 306), the number of verifications of the target object can be queried. If the number of verifications is not greater than the second threshold, the process ends, and it is not necessary to store the dynamic image in the dynamic feature library. If the number of verifications is less than the second threshold, then step 403 is executed.

[0092] Step 403: Determine the initial image of the target object (i.e. the last acquired initial image) as the candidate dynamic image of the target object, and store the candidate dynamic image in the dynamic feature library.

[0093] For example, if the number of dynamic images of the target object in the dynamic feature library is not limited, then after obtaining the candidate dynamic image of the target object, the candidate dynamic image is directly stored in the dynamic feature library.

[0094] For example, if the number of dynamic images of a target object in the dynamic feature library is limited, such as by K (where K is a positive integer), it means that the number of dynamic images in the dynamic feature library cannot exceed K. Based on this, after obtaining candidate dynamic images of the target object, the number of dynamic images of the target object already stored in the dynamic feature library can be counted. If this number is less than the threshold K, the candidate dynamic image is directly stored in the dynamic feature library. If the number is greater than or equal to the threshold K, the dynamic image of the target object is first deleted from the dynamic feature library, and the candidate dynamic image is then stored in the dynamic feature library.

[0095] For example, when deleting a dynamic image of a target object from the dynamic feature library, one dynamic image of the target object can be deleted randomly, or the dynamic image of the target object that has been stored in the dynamic feature library for the longest time can be deleted. That is, a cyclic overwrite method is used to overwrite the dynamic image that has been stored for the longest time.

[0096] For example, when storing candidate dynamic images in the dynamic feature library, the total verification time of the target object can be compared with a failure time threshold (which can be configured empirically). If the total verification time of the target object is less than the failure time threshold, it means that the target object has undergone multiple verifications in a short period of time, and all verifications have failed. Therefore, the candidate dynamic image can be stored in the dynamic feature library. If the total verification time of the target object is greater than or equal to the failure time threshold, the candidate dynamic image is not stored in the dynamic feature library. The total verification time can be the difference between the current time and the verification start time of the target object, and the verification start time can be the time when the initial image of the target object is first obtained.

[0097] Step 404: After storing the candidate dynamic image in the dynamic feature library, set an object identifier for the candidate dynamic image (the object identifier is the unique identifier of the candidate dynamic image), and the object identifier of the candidate dynamic image is associated with the object identifier of the reference base image, that is, bind the object identifier of the candidate dynamic image to the object identifier of the reference base image.

[0098] In one possible implementation, the object identifier of the candidate moving image can be set as the object identifier of the reference base image, so that the object identifier of the candidate moving image is associated with the object identifier of the reference base image. For example, see... Figure 5 The example shown illustrates a basic feature library and a dynamic feature library. The object identifier of the base image in the basic feature library is the same as the object identifier of the dynamic image in the dynamic feature library, and one base image can correspond to multiple dynamic images. In this case, for step 307, when determining whether a reference dynamic image exists in the dynamic feature library, if the object identifier of the dynamic image is the same as the object identifier of the reference base image, then the dynamic image is used as the reference dynamic image for the target object.

[0099] In one possible implementation, an object identifier can be set for the candidate dynamic image, and the association between the object identifier of the candidate dynamic image and the object identifier of the reference base image can be recorded in an object relationship table. That is, the object identifier of the candidate dynamic image is associated with the object identifier of the reference base image through the object relationship table. In this case, for step 307, when determining whether a reference dynamic image exists in the dynamic feature library, the object identifier of the dynamic image that matches the object identifier of the reference base image can be queried through the object relationship table, and then the dynamic image corresponding to the object identifier is used as the reference dynamic image of the target object.

[0100] For example, after obtaining a candidate dynamic image of the target object, the candidate dynamic image is stored in the dynamic feature library instead of the basic feature library. This is because the accuracy of facial recognition algorithms cannot be guaranteed, and there is a certain probability of misjudgment. If the current result is incorrect after obtaining a candidate dynamic image of target object A, storing the candidate dynamic image in the basic feature library will lead to a decrease in the recognition rate of target object A or even failure to recognize it during subsequent verification using the basic feature library, potentially posing a significant security risk. In this embodiment, because the candidate dynamic image is stored in the dynamic feature library instead of the basic feature library, even if the current result is incorrect, since only the dynamic feature library is adjusted and not the basic feature library, it still needs to be compared with the basic feature library during the next recognition, thus maintaining the interception capability of the target object and not affecting the normal use of the updated target object, resulting in higher security.

[0101] As can be seen from the above technical solutions, in this embodiment, a basic feature library and a dynamic feature library can be maintained. The basic feature library stores basic images of basic forms, and the dynamic feature library stores dynamic images after changes in the basic form. After obtaining the initial image of the target object, if the first similarity between the initial image and the reference basic image in the basic feature library is less than a threshold, the second similarity between the initial image and the reference dynamic image in the dynamic feature library can be compared. If the second similarity is greater than the threshold, the verification of the target object is determined to be successful. Since the dynamic image is the image after changes in the basic form, even if the image of the target object changes due to actions such as changing glasses or wearing a hat, the initial image and the reference dynamic image can still meet the similarity threshold, allowing the verification of the target object to pass, i.e., obtaining the correct verification result, resulting in a good user experience. Correcting erroneous verification results through a verification method based on a multi-feature library has higher security and enhances recognition accuracy. By proposing a dynamic adjustment method based on a multi-feature library, facial recognition errors can be corrected. When a misjudgment occurs in the basic feature library, the dynamic feature library can correct the erroneous results of the basic feature library.

[0102] Based on the same application concept as the above method, this application proposes a verification device based on a multi-feature library, see [link to relevant documentation]. Figure 6 The diagram shown is a structural schematic of the device, which includes:

[0103] The acquisition module 61 is used to acquire an initial image of the target object, and select a reference base image from the base feature library based on the first similarity between the initial image and multiple base images in the base feature library;

[0104] The judgment module 62 is used to determine whether a reference dynamic image exists in the dynamic feature library if the first similarity between the initial image and the reference base image is less than a first similarity threshold and greater than a second similarity threshold. The object identifier of the reference dynamic image is associated with the object identifier of the reference base image. The base feature library is used to store base images of basic forms, and the dynamic feature library is used to store dynamic images after morphological changes to the basic forms.

[0105] The determination module 63 is used to determine that the verification of the target object is successful if the reference dynamic image exists and the second similarity between the initial image and the reference dynamic image is greater than the third similarity threshold.

[0106] For example, the device further includes: a processing module, configured to, after determining whether a reference dynamic image exists in the dynamic feature library, increment the verification count of the target object by 1 if the reference dynamic image exists and the second similarity between the initial image and the reference dynamic image is not greater than a third similarity threshold; or, increment the verification count of the target object by 1 if the reference dynamic image does not exist; the determining module 62 is further configured to determine whether the verification count of the target object has reached a first threshold; the determining module 63 is further configured to determine that the verification of the target object has failed if the first threshold has been reached; if the first threshold has not been reached, the obtaining module 61 obtains the initial image for the target object.

[0107] For example, the determining module 63 is further configured to determine that the verification of the target object is successful if the first similarity between the initial image and the reference base image is greater than or equal to a first similarity threshold, and query the number of verifications of the target object; if the number of verifications of the target object is greater than a second number threshold, and the second number threshold is less than the first number threshold, then the initial image is determined as a candidate dynamic image of the target object; the device further includes: a storage module, configured to store the candidate dynamic images of the target object in the dynamic feature library; wherein, the object identifier of the candidate dynamic image of the target object has an association relationship with the object identifier of the reference base image.

[0108] For example, when the storage module stores candidate dynamic images of the target object in the dynamic feature library, it is specifically used to: count the number of dynamic images of the target object already stored in the dynamic feature library; if the number is less than a number threshold, then store the candidate dynamic image in the dynamic feature library; if the number is greater than or equal to the number threshold, then delete the dynamic image of the target object from the dynamic feature library and store the candidate dynamic image in the dynamic feature library; wherein, when deleting the dynamic image of the target object from the dynamic feature library, the dynamic image deleted from the dynamic feature library is the dynamic image of the target object that has been stored in the dynamic feature library for the longest time.

[0109] For example, when the storage module stores candidate dynamic images in the dynamic feature library, it is specifically used to: if the total verification time of the target object is less than the failure time threshold, then store the candidate dynamic image in the dynamic feature library; the total verification time is the difference between the current time and the verification start time of the target object, and the verification start time is the time when the initial image of the target object is first obtained.

[0110] For example, the processing module is further configured to, after storing the candidate dynamic image in the dynamic feature library, set the object identifier of the candidate dynamic image as the object identifier of the reference base image; or, set the object identifier for the candidate dynamic image and record the association between the object identifier of the candidate dynamic image and the object identifier of the reference base image in the object relationship table.

[0111] Based on the same application concept as the method described above, this application proposes an electronic device (such as a management device), see [link to relevant documentation]. Figure 7 As shown, it includes: a processor 71 and a machine-readable storage medium 72, the machine-readable storage medium 72 storing machine-executable instructions that can be executed by the processor 71; the processor 71 is used to execute the machine-executable instructions to implement the verification method based on a multi-feature library disclosed in the above example of this application.

[0112] Based on the same concept as the above method, this application embodiment also provides a machine-readable storage medium storing a plurality of computer instructions, which, when executed by a processor, can implement the verification method based on a multi-feature library disclosed in the above examples of this application.

[0113] The aforementioned machine-readable storage medium can be any electronic, magnetic, optical, or other physical storage device that can contain or store information, such as executable instructions, data, etc. For example, machine-readable storage media can be: RAM (Random Access Memory), volatile memory, non-volatile memory, flash memory, storage drives (such as hard disk drives), solid-state drives, any type of storage disk (such as optical discs, DVDs, etc.), or similar storage media, or combinations thereof.

[0114] The systems, devices, modules, or units described in the above embodiments can be implemented by a computer or entity, or by a product with a certain function. A typical implementation device is a computer, which can be a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email sending and receiving device, game console, tablet computer, wearable device, or any combination of these devices.

[0115] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0116] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, embodiments of this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0117] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0118] Furthermore, these computer program instructions can also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in the process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0119] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0120] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A multi-feature library based verification method, characterized in that, The method comprises: acquiring an initial image of a target object, and selecting a reference base image from a base feature library based on a first similarity of the initial image and a plurality of base images in the base feature library; if the first similarity of the initial image and the reference base image is not greater than a second similarity threshold, determining that the verification of the target object fails; if the first similarity is not less than a first similarity threshold, determining that the verification of the target object succeeds; if the first similarity is less than the first similarity threshold and greater than the second similarity threshold, judging whether a reference dynamic image exists in a dynamic feature library, and an object identifier of the reference dynamic image has a correlation with an object identifier of the reference base image; wherein the base feature library is used to store base images of a base form, and the dynamic feature library is used to store dynamic images of the base form after a form change occurs; if the reference dynamic image exists, and a second similarity of the initial image and the reference dynamic image is greater than a third similarity threshold, determining that the verification of the target object succeeds; if the reference dynamic image exists, and the second similarity of the initial image and the reference dynamic image is not greater than the third similarity threshold, adding 1 to a verification number of the target object; or if the reference dynamic image does not exist, adding 1 to the verification number of the target object; judging whether the verification number of the target object reaches a first number threshold; if yes, determining that the verification of the target object fails; if no, returning to acquire the initial image of the target object; wherein when the target object succeeds in verification after multiple times of verification, and the verification of the target object is confirmed based on the base feature library, a dynamic image of the target object is stored in the dynamic feature library; when the target object succeeds in verification after multiple times of verification, and the verification of the target object is confirmed based on the dynamic feature library, the dynamic image of the target object is prohibited from being stored in the dynamic feature library.

2. The method of claim 1, wherein after the reference base image is selected from the base feature library based on the first similarity of the initial image and the plurality of base images in the base feature library, the method further comprises: if the first similarity of the initial image and the reference base image is greater than or equal to the first similarity threshold, determining that the verification of the target object succeeds, and querying a verification number of the target object; if the verification number of the target object is greater than a second number threshold, and the second number threshold is less than a first number threshold, determining the initial image as a candidate dynamic image of the target object, and storing the candidate dynamic image in the dynamic feature library; wherein an object identifier of the candidate dynamic image has a correlation with an object identifier of the reference base image.

3. The method of claim 2, wherein the candidate dynamic image is stored in the dynamic feature library, comprising: counting a number of dynamic images of the target object that have been stored in the dynamic feature library. ​ ​ If the number is less than a number threshold, the candidate dynamic image is stored in the dynamic feature library; If the number is greater than or equal to the number threshold, a dynamic image of the target object is deleted from the dynamic feature library, and the candidate dynamic image is stored in the dynamic feature library.

4. The method of claim 3, wherein, The dynamic image deleted from the dynamic feature library is a dynamic image of the target object stored in the dynamic feature library for the longest time.

5. The method of claim 2, wherein the storing the candidate dynamic image in the dynamic feature library comprises: if a total verification duration of the target object is less than an invalid duration threshold, storing the candidate dynamic image in the dynamic feature library; wherein the total verification duration is a difference between a current time and a verification start time of the target object, and the verification start time is a time when an initial image of the target object is first acquired.

6. The method of claim 2, wherein the storing the candidate dynamic image in the dynamic feature library comprises: if a total verification duration of the target object is less than an invalid duration threshold, storing the candidate dynamic image in the dynamic feature library; wherein the total verification duration is a difference between a current time and a verification start time of the target object, and the verification start time is a time when an initial image of the target object is first acquired. After the storing the candidate dynamic image in the dynamic feature library, the method further comprises: setting an object identifier of the candidate dynamic image as an object identifier of the reference base image; or setting an object identifier for the candidate dynamic image, and recording an association relationship between the object identifier of the candidate dynamic image and the object identifier of the reference base image in an object relationship table. The apparatus comprises: an acquisition module configured to acquire an initial image of a target object, and select a reference base image from a base feature library based on a first similarity between the initial image and a plurality of base images in the base feature library; 7. A multi-feature library based verification apparatus, characterized by, a judgment module configured to determine that verification of the target object fails if the first similarity between the initial image and the reference base image is not greater than a second similarity threshold, determine that verification of the target object succeeds if the first similarity is not less than a first similarity threshold, and determine whether a reference dynamic image exists in a dynamic feature library if the first similarity is less than the first similarity threshold and greater than the second similarity threshold, the reference dynamic image having an association relationship with an object identifier of the reference base image, wherein the base feature library is configured to store base images of a base form, and the dynamic feature library is configured to store dynamic images of the base form after a form change; a determination module configured to determine that verification of the target object succeeds if the reference dynamic image exists and a second similarity between the initial image and the reference dynamic image is greater than a third similarity threshold. ​ ​ The device further includes a processing module configured to, after determining whether the reference dynamic image exists in the dynamic feature library, if the reference dynamic image exists and a second similarity between the initial image and the reference dynamic image is not greater than a third similarity threshold, then add 1 to a verification number of the target object; or, if the reference dynamic image does not exist, then add 1 to the verification number of the target object; the determining module is further configured to determine whether the verification number of the target object reaches a first number threshold; the determining module is further configured to, if the first number threshold is reached, determine that the verification of the target object fails; and if the first number threshold is not reached, then the initial image of the target object is acquired by the acquiring module. The dynamic image of the target object is stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library; and the dynamic image of the target object is prohibited from being stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library.

8. The apparatus of claim 7, The device further includes a processing module configured to, after determining whether the reference dynamic image exists in the dynamic feature library, if the reference dynamic image exists and a second similarity between the initial image and the reference dynamic image is not greater than a third similarity threshold, then add 1 to a verification number of the target object; or, if the reference dynamic image does not exist, then add 1 to the verification number of the target object; the determining module is further configured to determine whether the verification number of the target object reaches a first number threshold; the determining module is further configured to, if the first number threshold is reached, determine that the verification of the target object fails; and if the first number threshold is not reached, then the initial image of the target object is acquired by the acquiring module. The dynamic image of the target object is stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library; and the dynamic image of the target object is prohibited from being stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library. The device further includes a processing module configured to, after determining whether the reference dynamic image exists in the dynamic feature library, if the reference dynamic image exists and a second similarity between the initial image and the reference dynamic image is not greater than a third similarity threshold, then add 1 to a verification number of the target object; or, if the reference dynamic image does not exist, then add 1 to the verification number of the target object; the determining module is further configured to determine whether the verification number of the target object reaches a first number threshold; the determining module is further configured to, if the first number threshold is reached, determine that the verification of the target object fails; and if the first number threshold is not reached, then the initial image of the target object is acquired by the acquiring module. The dynamic image of the target object is stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library; and the dynamic image of the target object is prohibited from being stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library. The device further includes a processing module configured to, after determining whether the reference dynamic image exists in the dynamic feature library, if the reference dynamic image exists and a second similarity between the initial image and the reference dynamic image is not greater than a third similarity threshold, then add 1 to a verification number of the target object; or, if the reference dynamic image does not exist, then add 1 to the verification number of the target object; the determining module is further configured to determine whether the verification number of the target object reaches a first number threshold; the determining module is further configured to, if the first number threshold is reached, determine that the verification of the target object fails; and if the first number threshold is not reached, then the initial image of the target object is acquired by the acquiring module. The dynamic image of the target object is stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library; and the dynamic image of the target object is prohibited from being stored in the dynamic feature library when the target object is successfully verified after multiple verifications and the verification of the target object is confirmed to be successful based on the dynamic feature library. The processing module is further configured to, after storing the candidate dynamic image in the dynamic feature library, set an object identifier of the candidate dynamic image as an object identifier of the reference base image, or set an object identifier for the candidate dynamic image and record an association relationship between the object identifier of the candidate dynamic image and the object identifier of the reference base image in an object relationship table.

9. An electronic device, comprising: The method comprises: a processor and a machine readable storage medium storing machine executable instructions executable by the processor; the processor is configured to execute the machine executable instructions to implement the method of any one of claims 1-6.

Images