Method, server and medium for face recognition without network

Abstract

The application provides a network-free face recognition method, a server and a medium, which can be used in the field of big data. The method comprises the following steps: determining the functional relationship between the network-free face recognition threshold and the network-free face risk entropy according to the network-free face recognition data; determining the network face risk entropy of a target user according to the target face recognition threshold of the target user pre-stored in the server; determining the network-free face recognition threshold of the target user according to the functional relationship between the network-free face recognition threshold and the network-free face risk entropy and the network face risk entropy of the target user; and downloading the network-free face recognition threshold of the target user and the face image of the target user pre-stored in the server to the target mobile terminal of the target user, so that the target mobile terminal determines the face matching value when the target mobile terminal is in a network-free state, and performs face recognition when the face matching value is greater than the network-free face recognition threshold. The method of the application can accurately perform face recognition in a network-free environment.

Description

Technical Field

[0001] This application relates to the field of big data, and in particular to a method, server, and medium for offline facial recognition. Background Technology

[0002] With the development of internet technology and the progress of society, facial recognition technology has been applied to all aspects of people's lives and work.

[0003] In the facial recognition process, the terminal device first captures the target user's current facial image and then sends it to the server. The server matches the current facial image with the target user's pre-stored facial images. If the match is successful, the server sends a facial recognition success notification to the terminal device, indicating that the terminal device has passed the facial recognition for the target user and can proceed to the next step. Current facial recognition technology relies on a network to enable interaction between the terminal device and the server. However, users sometimes encounter situations where facial recognition is required without a network connection. For example, a user might experience a sudden network outage while performing facial recognition for certain transactions; or a user might need to conduct certain banking transactions in an offline environment for security reasons, but facial recognition is still required to verify their identity.

[0004] However, existing facial recognition technology cannot meet users' needs for facial recognition in environments without a network connection. Summary of the Invention

[0005] This application provides a method, server, and medium for offline face recognition, in order to solve the technical problem that existing face recognition technologies cannot perform face recognition in offline environments.

[0006] Firstly, this application provides a network-free face recognition method, comprising:

[0007] Based on offline face recognition data, the functional relationship between offline face recognition threshold and offline face risk entropy is determined, whereby offline face risk entropy is used to characterize the amount of information about risk when performing offline face recognition.

[0008] Based on the target user's pre-stored target face recognition threshold on the server, the network face risk entropy of the target user is determined. The network face risk entropy is used to characterize the amount of information related to risk when performing network face recognition.

[0009] Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, the offline face recognition threshold of the target user is determined.

[0010] The offline face recognition threshold of the target user and the face image of the target user pre-stored on the server are sent to the target user's target mobile terminal, so that the target mobile terminal can determine the face matching value of the target user based on the face image when it is in an offline state, and pass the face recognition of the target user when the face matching value is greater than the offline face recognition threshold.

[0011] In one possible implementation, determining the functional relationship between the offline face recognition threshold and the offline face risk entropy based on offline face recognition data specifically includes:

[0012] Based on the face matching values ​​corresponding to each offline face recognition data, multiple offline face recognition thresholds are set;

[0013] For each offline face recognition threshold, offline face recognition data with a face matching value greater than the offline face recognition threshold is used as the offline face recognition data corresponding to the offline face recognition threshold.

[0014] Based on the offline face recognition data corresponding to the offline face recognition threshold, determine the offline face risk entropy corresponding to the offline face recognition threshold;

[0015] Based on the aforementioned offline face recognition thresholds and the corresponding offline face risk entropy, the functional relationship between the offline face recognition thresholds and the offline face risk entropy is determined.

[0016] In one possible implementation, determining the offline face risk entropy corresponding to the offline face recognition threshold based on the offline face recognition data corresponding to the offline face recognition threshold specifically includes:

[0017] Multiple network-free face dimensions are selected, whereby the network-free face dimensions are used to divide the network-free face recognition data, such that the intersection between the network-free face recognition data sets corresponding to each network-free face dimension is empty;

[0018] Based on the multiple offline face dimensions, the offline face recognition data is divided to obtain offline face recognition data sets for each offline face dimension;

[0019] The number of risky face recognitions in the offline face recognition dataset of each offline face dimension is taken as the number of face recognitions in the offline face dimension.

[0020] The ratio of the number of face recognitions in the offline face dimension to the number of face recognitions contained in the offline face recognition data is taken as the offline face risk probability in the offline face dimension.

[0021] Determine the sum of the risk probabilities of non-network face in all non-network face dimensions, and use the difference between the sum threshold and the sum as the non-network face security probability;

[0022] Based on the offline face security probability and the offline face risk probability of each offline face dimension, the offline face risk entropy corresponding to the offline face recognition threshold is determined:

[0023]

[0024] Where P is the offline face risk entropy corresponding to the offline face recognition threshold, p is the offline face security probability, and p i It is the probability of face risk without network coverage in the i-th face dimension without network coverage.

[0025] In one possible implementation, determining the network face risk entropy of the target user based on the target user's pre-stored target face recognition threshold on the server specifically includes:

[0026] Obtain potential online facial recognition data of target users;

[0027] Potentially networked facial recognition data of the target user whose corresponding face matching value is greater than the target face recognition threshold are used as the target user's networked facial recognition data.

[0028] Based on the target user's online facial recognition data, determine the target user's online facial risk entropy.

[0029] In one possible implementation, acquiring the potential online facial recognition data of the target user specifically includes:

[0030] Determine the similarity entropy between the target user and all other users;

[0031] Other users whose similarity entropy to the target user is greater than the similarity threshold are considered as similar users of the target user.

[0032] The target user's historical online facial recognition data, as well as the target user's historical online facial recognition data of similar users, are used as the target user's potential online facial recognition data.

[0033] In one possible implementation, determining the online facial risk entropy of the target user based on the target user's online facial recognition data specifically includes:

[0034] Multiple network face dimensions are selected, whereby the network face dimensions are used to divide the network face recognition data, such that the intersection between the network face recognition data sets corresponding to each network face dimension is empty;

[0035] Based on the multiple network face dimensions, the network face recognition data is divided to obtain a set of network face recognition data for each network face dimension.

[0036] The number of risky face recognitions in each set of networked face recognition data with networked face dimension is taken as the number of face recognitions in the networked face dimension.

[0037] The ratio of the number of face recognitions in the network face dimension to the number of face recognitions contained in the network face recognition data is taken as the network face risk probability in the network face dimension.

[0038] Determine the sum of all network face risk probabilities for network face dimensions, and use the difference between the sum threshold and the sum as the network face security probability;

[0039] Based on the aforementioned network-enabled face security probability and the network-enabled face risk probability of each network-enabled face dimension, the network-enabled face risk entropy of the target user is determined:

[0040]

[0041] Where Q is the network-based face risk entropy of the target user, q is the network-based face security probability, and q j It is the probability of network face risk for the j-th network face dimension.

[0042] In one possible implementation, determining the offline face recognition threshold of the target user based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, specifically includes:

[0043] Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, multiple offline face recognition threshold intervals are determined, wherein the offline face risk entropy in the offline face recognition threshold interval is less than or equal to the online face risk entropy of the target user.

[0044] Based on the multiple offline face recognition threshold ranges, the offline face recognition threshold for the target user is determined.

[0045] In one possible implementation, determining the offline face recognition threshold for the target user based on the plurality of offline face recognition threshold intervals specifically includes:

[0046] The minimum point of the functional relationship in each of the network-free face recognition threshold intervals is taken as the candidate minimum point;

[0047] The threshold value of the face recognition without network corresponding to the candidate minimum point is used as the candidate face recognition threshold.

[0048] Based on the candidate face recognition thresholds, determine the offline face recognition threshold for the target user.

[0049] Secondly, this application provides a server, comprising:

[0050] The processing module is used to determine the functional relationship between the offline face recognition threshold and the offline face risk entropy based on offline face recognition data. The offline face risk entropy is used to characterize the amount of risk-related information when performing offline face recognition. Based on the target face recognition threshold pre-stored on the server, the module determines the online face risk entropy of the target user. The online face risk entropy is used to characterize the amount of risk-related information when performing online face recognition. Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, the module determines the offline face recognition threshold of the target user.

[0051] The sending module is used to send the target user's offline face recognition threshold and the target user's pre-stored face image on the server to the target user's target mobile terminal, so that the target mobile terminal can determine the target user's face matching value based on the face image when it is in an offline state, and pass the target user's face recognition when the face matching value is greater than the offline face recognition threshold.

[0052] Thirdly, this application provides another server, including: a processor, and a memory communicatively connected to the processor;

[0053] The memory stores computer-executed instructions;

[0054] The processor executes computer execution instructions stored in the memory to implement the above-described method.

[0055] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the above-described method.

[0056] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0057] The offline face recognition method, server, and medium provided in this application can determine the functional relationship between the offline face recognition threshold and the offline face risk entropy based on offline face recognition data. The offline face risk entropy characterizes the amount of risk-related information during offline face recognition. Based on the target user's pre-stored target face recognition threshold on the server, the network-connected face risk entropy of the target user is determined. The network-connected face risk entropy characterizes the amount of risk-related information during network-connected face recognition. Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the target user's network-connected face risk entropy, the offline face recognition threshold of the target user is determined. The offline face recognition threshold of the target user, along with the face image of the target user pre-stored on the server, is sent to the target user's target mobile terminal. This allows the target mobile terminal to determine the target user's face matching value based on the face image when in an offline state, and to pass the target user's face recognition if the face matching value is greater than the offline face recognition threshold. The method of this application allows the server to pre-determine the functional relationship between the offline face recognition threshold and the offline face risk entropy, as well as the online face risk entropy of the target user, in a network-enabled environment. Based on this functional relationship and the online face risk entropy, the server then determines the offline face recognition threshold for the target user. The server can then send the offline face recognition threshold and the target user's pre-stored face image to the target user's mobile terminal. When the target mobile terminal is offline, the server can directly determine the target user's face matching value based on the face image. If the face matching value is greater than the offline face recognition threshold, the target user's face recognition is successful. This setup ensures accurate face recognition even in offline environments, improving the user experience. Furthermore, by utilizing the functional relationship and the online face risk entropy to determine the offline face recognition threshold, the server introduces risk information for both offline and online face recognition, improving the accuracy of the offline face recognition threshold and ultimately enhancing the overall accuracy of face recognition. Attached Figure Description

[0058] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0059] Figure 1 This is a schematic diagram of a face recognition process in existing technology.

[0060] Figure 2 This is a system architecture diagram of an embodiment of this application;

[0061] Figure 3 This is a flowchart of a network-free face recognition method according to an embodiment of this application;

[0062] Figure 4 This is a schematic diagram of the structure of a server according to an embodiment of this application;

[0063] Figure 5 This is a schematic diagram of the server structure according to another embodiment of this application.

[0064] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0065] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0066] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0067] It should also be noted that the network-free face recognition method, server and medium of this application can be used in the field of big data, or in any field other than big data, such as the financial field, face recognition field, image processing field, etc. The application field of the network-free face recognition method, server and medium of this application is not limited.

[0068] Figure 1 This is a schematic diagram of a face recognition process in existing technology, such as... Figure 1 As shown, during the face recognition process, terminal device 1 first captures the current face image of the target user and then sends the captured current face image to server 2. Server 2 performs face matching between the current face image and the target user's pre-stored face image. If the face matching is successful, server 2 sends a face recognition success indication message to terminal device 1, indicating that terminal device 1 has passed the face recognition of the target user and can proceed to the next step.

[0069] Current facial recognition technology relies on a network to enable interaction between terminal devices and servers. However, users sometimes encounter situations where facial recognition is required without a network connection. For example, a user might experience a sudden network outage while performing facial recognition for certain transactions; or, for security reasons, certain banking transactions might need to be conducted offline, yet facial recognition is still required to verify the user's identity. Clearly, existing facial recognition technology cannot meet users' needs for facial recognition in offline environments.

[0070] Based on this technical problem, the inventive concept of this application is: how to provide a method for face recognition without a network.

[0071] Specifically, based on offline face recognition data, a functional relationship between the offline face recognition threshold and the offline face risk entropy can be determined. The offline face risk entropy characterizes the amount of risk-related information during offline face recognition. Based on the target user's pre-stored face recognition threshold on the server, the target user's online face risk entropy is determined. The online face risk entropy characterizes the amount of risk-related information during online face recognition. Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the target user's online face risk entropy, the target user's offline face recognition threshold is determined. The target user's offline face recognition threshold and the target user's pre-stored face image on the server are then sent to the target user's target mobile terminal. This allows the target mobile terminal to determine the target user's face matching value based on the face image when in an offline state, and to pass the target user's face recognition if the face matching value is greater than the offline face recognition threshold. The method of this application allows the server to pre-determine the functional relationship between the offline face recognition threshold and the offline face risk entropy, as well as the online face risk entropy of the target user, in a network-enabled environment. Based on this functional relationship and the online face risk entropy, the server then determines the offline face recognition threshold for the target user. The server can then send the offline face recognition threshold and the target user's pre-stored face image to the target user's mobile terminal. When the target mobile terminal is offline, the server can directly determine the target user's face matching value based on the face image. If the face matching value is greater than the offline face recognition threshold, the target user's face recognition is successful. This setup ensures accurate face recognition even in offline environments, improving the user experience. Furthermore, by utilizing the functional relationship and the online face risk entropy to determine the offline face recognition threshold, the server introduces risk information for both offline and online face recognition, improving the accuracy of the offline face recognition threshold and ultimately enhancing the overall accuracy of face recognition.

[0072] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0073] Figure 2 This is a system architecture diagram of an embodiment of this application, such as... Figure 2 As shown, the target mobile terminal 3 can pre-send the target user's offline face recognition data to server 2 in a network-connected environment. Server 2 determines the functional relationship between the offline face recognition threshold and the offline face risk entropy based on the offline face recognition data; it also determines the target user's network-connected face risk entropy based on the target user's pre-stored target face recognition threshold on server 2; and finally, based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the target user's network-connected face risk entropy, it determines the target user's offline face recognition threshold. The target user's offline face recognition threshold and the target user's pre-stored face image on server 2 are then sent to the target user's target mobile terminal 3. When the target mobile terminal 3 is in a network-connected state, it interacts with server 2 and performs face recognition according to existing network-connected face recognition technology. When the target mobile terminal 3 is in a no-network state, the target mobile terminal 3 collects the current face image of the target user, determines the face matching value between the current face image and the face image sent by the server 2, and if the face matching value is greater than the no-network face recognition threshold, the target user's face recognition is passed; if the face matching value is not greater than the no-network face recognition threshold, the target user's face recognition is not passed.

[0074] Example 1

[0075] Figure 3 This is a flowchart illustrating a method for offline face recognition according to an embodiment of this application. This embodiment describes the offline face recognition method with a server as the executing entity. Figure 3 As shown, the network-free face recognition method may include the following steps:

[0076] S101: Based on offline face recognition data, determine the functional relationship between the offline face recognition threshold and the offline face risk entropy.

[0077] In this embodiment, the risk entropy of a face without a network can be used to characterize the amount of information about the risk when performing face recognition without a network.

[0078] In one possible implementation, step S101, which determines the functional relationship between the offline face recognition threshold and the offline face risk entropy based on offline face recognition data, may include:

[0079] S1011: Based on the face matching value corresponding to each offline face recognition data, set multiple offline face recognition thresholds.

[0080] S1012: For each offline face recognition threshold, the offline face recognition data whose corresponding face matching value is greater than the offline face recognition threshold is taken as the offline face recognition data corresponding to the offline face recognition threshold.

[0081] S1013: Based on the offline face recognition data corresponding to the offline face recognition threshold, determine the offline face risk entropy corresponding to the offline face recognition threshold.

[0082] S1014: Based on multiple offline face recognition thresholds and the corresponding offline face risk entropy, determine the functional relationship between the offline face recognition thresholds and the offline face risk entropy.

[0083] In this embodiment, the offline face recognition data can be the data generated by the target user when performing offline face recognition; the face matching value corresponding to the offline face recognition data can be the matching value between the face image captured by the target mobile terminal and the face image pre-stored in the target mobile terminal when the target user performs offline face recognition. The face matching value corresponding to the offline face recognition data can be obtained by the server from the target mobile terminal when there is a network connection.

[0084] In this embodiment, the multiple offline face recognition thresholds can be set randomly or arbitrarily within a certain range. Those skilled in the art can flexibly set the offline face recognition thresholds.

[0085] In this embodiment, face recognition data without a network connection whose face matching value is greater than the face recognition threshold without a network connection can be considered as data that can be recognized by face recognition. Therefore, it can be used as face recognition data without a network connection corresponding to the face recognition threshold without a network connection.

[0086] In this embodiment, after determining the offline face risk entropy corresponding to each offline face recognition threshold, the functional relationship between the offline face recognition threshold and the offline face risk entropy can be determined by curve fitting based on multiple offline face recognition thresholds and their corresponding offline face risk entropies.

[0087] In this embodiment, offline face recognition data with a face matching value greater than the offline face recognition threshold is selected as the offline face recognition data corresponding to the offline face recognition threshold. This allows offline face recognition data to be considered as data that has passed face recognition. The offline face recognition data corresponding to the offline face recognition threshold is used to determine the corresponding offline face risk entropy, which improves the accuracy of the offline face risk entropy. This allows the offline face risk entropy to accurately represent the amount of risk information during offline face recognition, thereby improving the accuracy of subsequent offline face recognition threshold determination.

[0088] In one possible implementation, step S1013, which determines the functional relationship between the offline face recognition thresholds and the offline face risk entropy based on multiple offline face recognition thresholds and their corresponding offline face risk entropies, may include:

[0089] S11: Select multiple network-free face dimensions, where the network-free face dimensions are used to divide the network-free face recognition data so that the intersection between the network-free face recognition data sets corresponding to each network-free face dimension is empty.

[0090] S12: Based on multiple offline face dimensions, divide the offline face recognition data to obtain offline face recognition data sets for each offline face dimension.

[0091] S13: The number of risky face recognitions in the offline face recognition dataset of each offline face dimension is taken as the number of face recognitions in the offline face dimension.

[0092] S14: The ratio of the number of face recognitions in the offline face dimension to the number of face recognitions contained in the offline face recognition data is used as the offline face risk probability in the offline face dimension.

[0093] S15: Determine the sum of the risk probabilities of non-network faces for all non-network face dimensions, and use the difference between the sum threshold and the sum as the non-network face security probability.

[0094] S16: Based on the security probability of face recognition without network and the risk probability of face recognition without network in each dimension, determine the risk entropy of face recognition without network corresponding to the face recognition without network threshold, i.e., the following formula (1):

[0095]

[0096] Where P is the offline face risk entropy corresponding to the offline face recognition threshold, p is the offline face security probability, and p i It is the probability of face risk without network coverage in the i-th face dimension without network coverage.

[0097] In this embodiment, the number of network-free face dimensions can be flexibly set by those skilled in the art, as long as the intersection between the network-free face recognition data sets divided according to the network-free face dimensions is empty.

[0098] In this embodiment, the sum threshold can be set to 1.

[0099] In this embodiment, the network-free face risk probability of each network-free face dimension is determined based on the ratio of the number of face recognitions corresponding to each network-free face dimension to the number of face recognitions contained in the network-free face recognition data. This ensures that the network-free face risk probability accurately represents the risk information of the corresponding network-free face dimension. The network-free face security probability can be accurately determined by summing the network-free face risk probabilities of all network-free face dimensions. Furthermore, using formula (1), the network-free face security probability, and the network-free face risk probability of each network-free face dimension, the network-free face risk entropy corresponding to the network-free face recognition threshold can be determined simply and accurately.

[0100] S102: Determine the network face risk entropy of the target user based on the target user's pre-stored target face recognition threshold on the server.

[0101] In this embodiment, the network face risk entropy can be used to characterize the amount of information about risk when performing network face recognition.

[0102] In one possible implementation, step S102 above, which determines the network face risk entropy of the target user based on the target user's pre-stored target face recognition threshold on the server, may include:

[0103] S1021: Obtain potential networked facial recognition data of the target user.

[0104] S1022: The potential network face recognition data of the target user whose corresponding face matching value is greater than the target face recognition threshold is taken as the network face recognition data of the target user.

[0105] S1023: Based on the target user's online facial recognition data, determine the target user's online facial risk entropy.

[0106] In this embodiment, the face matching value corresponding to the potential network face recognition data of the target user can be the face matching value between the face image of other users and the pre-stored image of the target user during network face recognition.

[0107] In this embodiment, based on the potential online face recognition data of the target user whose face matching value is greater than the target face recognition threshold, the online face recognition data of the target user can be determined simply and accurately, thereby accurately determining the online face risk entropy of the target user.

[0108] In one possible implementation, step S1021, which involves acquiring potential online face recognition data of the target user, may include: determining the similarity entropy between the target user and each other user; identifying other users whose similarity entropy is greater than the similarity threshold as similar users of the target user; and using the target user's historical online face recognition data and the historical online face recognition data of the target user's similar users as potential online face recognition data of the target user.

[0109] In this embodiment, the similar threshold can be flexibly set by those skilled in the art according to actual conditions, and no restrictions are imposed here.

[0110] In this embodiment, the method for determining the similarity entropy between the target user and each other user can refer to existing similarity entropy calculation methods, and will not be elaborated here.

[0111] In one possible implementation, determining the similarity entropy between the target user and each other user may include:

[0112] Retrieve multiple consecutive user attributes;

[0113] Determine the values ​​of the target user and each other user in each consecutive user attribute;

[0114] For each consecutive user attribute, determine the minimum n and maximum m between the target user and the other users in that consecutive user attribute;

[0115] When the maximum value m is greater than twice the minimum value n, the similarity entropy between the target user and other users in this continuous user attribute is determined by the following formula (2):

[0116]

[0117] When the maximum value m is not greater than twice the minimum value n, the similarity entropy between the target user and other users in this continuous user attribute is determined by the following formula (3):

[0118]

[0119] Based on the similarity entropy between the target user and other users in each consecutive user attribute, determine the similarity entropy between the target user and other users.

[0120] In this embodiment, using similarity entropy and similarity threshold to determine similar users of the target user can improve the accuracy of similar user determination, thereby improving the accuracy of potential online face recognition data of the target user determined based on the historical online face recognition data of the target user and similar users.

[0121] In one possible implementation, step S1023, which determines the network face risk entropy of the target user based on the target user's network face recognition data, may include:

[0122] S21: Select multiple network face dimensions, where the network face dimensions are used to divide the network face recognition data so that the intersection between the network face recognition data sets corresponding to each network face dimension is empty.

[0123] S22: Based on multiple network face dimensions, divide the network face recognition data to obtain the network face recognition data set for each network face dimension.

[0124] S23: The number of risky face recognitions in each set of networked face recognition datasets with networked face dimensions is taken as the number of face recognitions with networked face dimensions.

[0125] S24: The ratio of the number of face recognitions with network face dimension to the number of face recognitions contained in the network face recognition data is used as the probability of network face risk with network face dimension.

[0126] S25: Determine the sum of the network face risk probabilities for all network face dimensions, and use the difference between the sum threshold and the sum as the network face security probability.

[0127] S26: Based on the network face security probability and the network face risk probability of each network face dimension, the network face risk entropy of the target user is determined by the following formula (4):

[0128]

[0129] Where Q is the network face risk entropy of the target user, q is the network face security probability, and q j It is the probability of network face risk for the j-th network face dimension.

[0130] In this embodiment, the number of network face dimensions can be flexibly set by those skilled in the art according to actual needs, as long as the intersection between the network face recognition data sets divided according to the network face dimensions is empty.

[0131] In this embodiment, the sum threshold can be set to 1.

[0132] In this embodiment, the network face risk probability of each network face dimension is determined based on the ratio of the number of face recognitions corresponding to each network face dimension to the number of face recognitions contained in the network face recognition data. This ensures that the network face risk probability accurately represents the risk information of the corresponding network face dimension. The network face security probability can be accurately determined by summing the network face risk probabilities of all network face dimensions. Furthermore, using formula (4) and the network face security probability and the network face risk probability of each network face dimension, the network face risk entropy corresponding to the network face recognition threshold can be determined simply and accurately.

[0133] S103: Determine the offline face recognition threshold for the target user based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user.

[0134] In one possible implementation, step S103, which determines the offline face recognition threshold of the target user based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, may include:

[0135] S1031: Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, determine multiple offline face recognition threshold intervals, wherein the offline face risk entropy in the offline face recognition threshold interval is less than or equal to the online face risk entropy of the target user.

[0136] S1032: Determine the offline face recognition threshold for the target user based on multiple offline face recognition threshold ranges.

[0137] In this embodiment, the offline face recognition threshold range can be easily and accurately determined by using an offline face risk entropy that is less than or equal to the online face risk entropy, thereby accurately determining the offline face recognition threshold for the target user. Furthermore, having an offline face risk entropy less than or equal to the online face risk entropy of the target user ensures that the risk of offline face recognition is less than or equal to that of online face recognition, minimizing the risk of offline face recognition and improving the accuracy of the offline face recognition threshold.

[0138] In one possible implementation, step S1032, which determines the offline face recognition threshold for the target user based on multiple offline face recognition threshold intervals, may include: taking the minimum point of the functional relationship in each offline face recognition threshold interval as a candidate minimum point; taking the offline face recognition threshold corresponding to the candidate minimum point as a candidate face recognition threshold; and determining the offline face recognition threshold for the target user based on the candidate face recognition threshold.

[0139] In this embodiment, when there are multiple candidate face recognition thresholds, the candidate face recognition threshold with the smallest distance from the target face recognition threshold can be used as the offline face recognition threshold for the target user.

[0140] In this embodiment, the minimum point of each offline face recognition threshold interval is used as the candidate minimum point, and the offline face recognition threshold is determined based on the candidate minimum point. This can further reduce the risk of offline face recognition and improve the accuracy of offline face recognition threshold.

[0141] S104: Send the target user's offline face recognition threshold and the target user's pre-stored face image on the server to the target user's target mobile terminal, so that the target mobile terminal can determine the target user's face matching value based on the face image when it is in an offline state, and pass the target user's face recognition when the face matching value is greater than the offline face recognition threshold.

[0142] In this embodiment, when the target mobile terminal is connected to a network, it can interact with the server and perform face recognition using existing network-connected face recognition technology. When the target mobile terminal is offline, it cannot interact with the server. In this case, the target mobile terminal can perform offline face recognition based on a pre-issued offline face recognition threshold from the server and the target user's pre-stored face image on the server. Specifically, the target mobile terminal acquires the target user's current face image and determines the face matching value between the current face image and the target user's pre-stored face image on the server. If the face matching value is greater than the offline face recognition threshold, the target user's face recognition passes; if the face matching value is not greater than the offline face recognition threshold, the target user's face recognition fails.

[0143] In one possible implementation, risky face recognition can be determined as follows, wherein face recognition includes both network-connected and offline face recognition:

[0144] Obtain multiple face recognition relationship models, where each face recognition relationship model contains a rule body and a rule head. The rule body includes the magnitude relationship of the values ​​of two face recognition variables in various face recognition attributes, and the rule head includes the magnitude relationship of the risks of two face recognition variables.

[0145] The risk identifiers in the facial recognition data that represent risks are classified as risky facial recognition.

[0146] The face recognition that does not involve risk is identified as the face recognition to be determined;

[0147] For each face to be identified, the risk relationship between the face to be identified and the risky face is determined based on multiple face recognition relationship models.

[0148] When it is determined that a risky face recognition exists, such that the risk of the face recognition to be identified is greater than the risk of the face recognition itself, the face recognition to be identified is identified as a risky face recognition.

[0149] In this embodiment, the server can pre-determine the functional relationship between the offline face recognition threshold and the offline face risk entropy, as well as the network-enabled face risk entropy of the target user, in a network-enabled environment. Based on this functional relationship and the network-enabled face risk entropy, the server determines the offline face recognition threshold for the target user. The server then sends the offline face recognition threshold and the target user's pre-stored face image to the target user's mobile terminal. When the target mobile terminal is offline, the server can directly determine the target user's face matching value based on the face image. If the face matching value is greater than the offline face recognition threshold, the target user's face recognition is successful. This setup ensures accurate face recognition even in an offline environment, improving the user experience. Furthermore, by utilizing the functional relationship and the network-enabled face risk entropy to determine the offline face recognition threshold, the server introduces risk information for both offline and network-enabled face recognition, improving the accuracy of the offline face recognition threshold and ultimately enhancing the overall accuracy of face recognition.

[0150] The method for network-free face recognition of this application is illustrated below with a specific embodiment.

[0151] Example 2

[0152] In one specific embodiment, the offline face recognition process for user A is as follows:

[0153] The first step involves the server determining the functional relationship between the offline face recognition threshold and the offline face risk entropy in a network environment, based on offline face recognition data.

[0154] The second step is for the server to determine the network face risk entropy of user A based on the target face recognition threshold pre-stored on the server.

[0155] The third step involves the server determining the offline face recognition threshold for user A based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, as well as the online face risk entropy for user A.

[0156] The fourth step is for the server to send the offline face recognition threshold of user A and the face image of user A pre-stored on the server to user A's target mobile terminal.

[0157] The fifth step involves the target mobile terminal interacting with the server when it is connected to the network, and performing facial recognition using existing network-connected facial recognition technology.

[0158] Step 6: When the target mobile terminal is in a network-free state, the target mobile terminal collects the current face image of user A, determines the face matching value between the current face image and the face image sent by the server, and determines that the face matching value is greater than the face recognition threshold without network, then the face recognition of user A is passed.

[0159] Figure 4 This is a schematic diagram of the structure of a server according to an embodiment of this application, as shown below. Figure 4 As shown, the server includes: a processing module 41, used to determine the functional relationship between the offline face recognition threshold and the offline face risk entropy based on offline face recognition data, wherein the offline face risk entropy is used to characterize the amount of information related to risk when performing offline face recognition; to determine the network-enabled face risk entropy of the target user based on the target face recognition threshold pre-stored on the server, the network-enabled face risk entropy being used to characterize the amount of information related to risk when performing network-enabled face recognition; and to determine the offline face recognition threshold of the target user based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the network-enabled face risk entropy of the target user; and a sending module 42, used to send the offline face recognition threshold of the target user and the face image of the target user pre-stored on the server to the target user's target mobile terminal, so that the target mobile terminal can determine the face matching value of the target user based on the face image when in an offline state, and pass the face recognition of the target user when the face matching value is greater than the offline face recognition threshold. In one implementation, the specific functions of the server can be described in steps S101-S104 of Embodiment 1, and will not be repeated here.

[0160] In one possible implementation, the processing module 41 can also be used for:

[0161] Based on the face matching values ​​corresponding to each offline face recognition data, multiple offline face recognition thresholds are set;

[0162] For each offline face recognition threshold, offline face recognition data with a face matching value greater than the offline face recognition threshold is taken as the offline face recognition data corresponding to the offline face recognition threshold.

[0163] Based on the offline face recognition data corresponding to the offline face recognition threshold, determine the offline face risk entropy corresponding to the offline face recognition threshold;

[0164] Based on multiple offline face recognition thresholds and their corresponding offline face risk entropies, the functional relationship between the offline face recognition thresholds and offline face risk entropies is determined.

[0165] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S101 of Embodiment 1, and will not be repeated here.

[0166] In one possible implementation, the processing module 41 can also be used for:

[0167] Multiple network-free face dimensions are selected, which are used to divide the network-free face recognition data so that the intersection between the network-free face recognition data sets corresponding to each network-free face dimension is empty;

[0168] Based on multiple offline face dimensions, offline face recognition data is divided to obtain offline face recognition data sets for each offline face dimension;

[0169] The number of risky face recognitions in the offline face recognition dataset of each offline face dimension is taken as the number of face recognitions in the offline face dimension.

[0170] The ratio of the number of face recognitions in the offline face dimension to the number of face recognitions contained in the offline face recognition data is used as the offline face risk probability in the offline face dimension.

[0171] Determine the sum of the risk probabilities of non-network face in all non-network face dimensions, and use the difference between the sum threshold and the sum as the non-network face security probability;

[0172] Based on the offline face security probability and the offline face risk probability of each offline face dimension, the offline face risk entropy corresponding to the offline face recognition threshold is determined:

[0173]

[0174] Where P is the offline face risk entropy corresponding to the offline face recognition threshold, p is the offline face security probability, and p i It is the probability of face risk without network coverage in the i-th face dimension without network coverage.

[0175] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S101 of Embodiment 1, and will not be repeated here.

[0176] In one possible implementation, the processing module 41 can also be used for:

[0177] Obtain potential online facial recognition data of target users;

[0178] Potentially networked face recognition data of the target user whose face matching value is greater than the target face recognition threshold are used as the target user's networked face recognition data.

[0179] Based on the target user's online facial recognition data, determine the target user's online facial risk entropy.

[0180] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S102 of Embodiment 1, and will not be repeated here.

[0181] In one possible implementation, the processing module 41 can also be used for:

[0182] Determine the similarity entropy between the target user and all other users;

[0183] Other users whose similarity entropy to the target user is greater than the similarity threshold are considered as similar users of the target user.

[0184] The target user's historical online facial recognition data, as well as the target user's historical online facial recognition data of similar users, are used as the target user's potential online facial recognition data.

[0185] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S102 of Embodiment 1, and will not be repeated here.

[0186] In one possible implementation, the processing module 41 can also be used for:

[0187] Multiple network face dimensions are selected, which are used to divide the network face recognition data so that the intersection between the network face recognition data sets corresponding to each network face dimension is empty;

[0188] Based on multiple dimensions of networked faces, the networked face recognition data is divided to obtain the networked face recognition data set for each networked face dimension;

[0189] The number of risky face recognitions in each set of networked face recognition datasets with networked face dimensions is taken as the number of face recognitions with networked face dimensions.

[0190] The ratio of the number of face recognitions with network face dimension to the number of face recognitions contained in the network face recognition data is used as the probability of network face risk with network face dimension.

[0191] Determine the sum of the network face risk probabilities for all network face dimensions, and use the difference between the sum threshold and the sum as the network face security probability.

[0192] Based on the probability of network-based facial recognition security and the probability of network-based facial recognition risk for each dimension, the network-based facial recognition risk entropy of the target user is determined:

[0193]

[0194] Where Q is the network face risk entropy of the target user, q is the network face security probability, and q j It is the probability of network face risk for the j-th network face dimension.

[0195] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S102 of Embodiment 1, and will not be repeated here.

[0196] In one possible implementation, the processing module 41 can also be used for:

[0197] Based on the functional relationship between the offline face recognition threshold and the offline face risk entropy, and the online face risk entropy of the target user, multiple offline face recognition threshold intervals are determined. Among them, the offline face risk entropy in the offline face recognition threshold interval is less than or equal to the online face risk entropy of the target user.

[0198] Based on multiple offline face recognition threshold ranges, the offline face recognition threshold for the target user is determined.

[0199] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S103 of Embodiment 1, and will not be repeated here.

[0200] In one possible implementation, the processing module 41 can also be used for:

[0201] The minimum point of the functional relationship in each threshold interval of face recognition without network is taken as the candidate minimum point;

[0202] The threshold value of the face recognition without network corresponding to the candidate minimum point is used as the candidate face recognition threshold.

[0203] Based on the candidate face recognition thresholds, determine the offline face recognition threshold for the target user.

[0204] In this embodiment, the specific functions implemented by the processing module 41 can be described in step S103 of Embodiment 1, and will not be repeated here.

[0205] Figure 5 This is a schematic diagram of the server structure according to another embodiment of this application, as shown below. Figure 5As shown, the server includes: a processor 101, and a memory 102 communicatively connected to the processor 101; the memory 102 stores computer execution instructions; the processor 101 executes the computer execution instructions stored in the memory 102 to implement the steps of the network-free face recognition method in the above method embodiments.

[0206] In the aforementioned server, the memory 102 and the processor 101 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines, such as a bus connection. The memory 102 stores computer-executable instructions that implement data access control methods, including at least one software functional module that can be stored in the memory 102 in the form of software or firmware. The processor 101 executes various functional applications and data processing by running the software programs and modules stored in the memory 102.

[0207] The memory 102 may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 102 stores programs, which are executed by the processor 101 upon receiving execution instructions. Furthermore, the software programs and modules within the memory 102 may include an operating system, which may include various software components and / or drivers for managing system tasks (e.g., memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components.

[0208] Processor 101 can be an integrated circuit chip with signal processing capabilities. The aforementioned processor 101 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.

[0209] An embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the steps of the various method embodiments of this application.

[0210] An embodiment of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the various method embodiments of this application.

[0211] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the appended claims.

[0212] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for network-free face recognition, the method comprising: The method comprises the following steps: determining the function relationship between the threshold value of the face recognition without network and the risk entropy of the face recognition without network according to the data of the face recognition without network, wherein the risk entropy of the face recognition without network is used to represent the information amount of the risk when the face recognition without network is performed; determining the risk entropy of the face recognition with network of the target user according to the threshold value of the face recognition of the target user pre-stored in the server, wherein the risk entropy of the face recognition with network is used to represent the information amount of the risk when the face recognition with network is performed; determining the threshold value of the face recognition without network of the target user according to the function relationship between the threshold value of the face recognition without network and the risk entropy of the face recognition without network and the risk entropy of the face recognition with network of the target user; downloading the threshold value of the face recognition without network of the target user and the face image of the target user pre-stored in the server to the target mobile terminal of the target user, so that the target mobile terminal determines the face matching value of the target user according to the face image when the target mobile terminal is in the state without network, and performs the face recognition of the target user when the face matching value is greater than the threshold value of the face recognition without network.

2. The method of claim 1, wherein, The step of determining the function relationship between the threshold value of the face recognition without network and the risk entropy of the face recognition without network according to the data of the face recognition without network comprises the following steps: setting multiple threshold values of the face recognition without network according to the face matching value corresponding to each data of the face recognition without network; for each threshold value of the face recognition without network, taking the data of the face recognition without network whose face matching value is greater than the threshold value of the face recognition without network as the data of the face recognition without network corresponding to the threshold value of the face recognition without network; determining the risk entropy of the face recognition without network corresponding to the threshold value of the face recognition without network according to the data of the face recognition without network corresponding to the threshold value of the face recognition without network; determining the function relationship between the threshold value of the face recognition without network and the risk entropy of the face recognition without network according to multiple threshold values of the face recognition without network and the risk entropy corresponding to the threshold values.

3. The method of claim 2, wherein, The step of determining the risk entropy of the face recognition without network corresponding to the threshold value of the face recognition without network according to the data of the face recognition without network corresponding to the threshold value of the face recognition without network comprises the following steps: selecting multiple dimensions of the face recognition without network, wherein the dimension of the face recognition without network is used to divide the data of the face recognition without network, so that the intersection between the data set of each dimension of the face recognition without network is empty; dividing the data of the face recognition without network according to the multiple dimensions of the face recognition without network to obtain the data set of each dimension of the face recognition without network; taking the number of the risk face recognition in the data set of each dimension of the face recognition without network as the face recognition number of the dimension of the face recognition without network; taking the ratio between the face recognition number of the dimension of the face recognition without network and the face recognition number contained in the data of the face recognition without network as the risk probability of the dimension of the face recognition without network; determining the sum of the risk probability of all dimensions of the face recognition without network, and taking the difference between the sum threshold value and the sum as the safety probability of the face recognition without network; determining the risk entropy of the face recognition without network corresponding to the threshold value of the face recognition without network according to the safety probability of the face recognition without network and the risk probability of each dimension of the face recognition without network: Wherein, P is the no-network face risk entropy corresponding to the no-network face recognition threshold, p is the no-network face security probability, p i is the no-network face risk probability of the ith no-network face dimension.

4. The method of claim 1, wherein, The target face recognition threshold pre-existing in the server according to a target user is used to determine a network face risk entropy of the target user, and specifically includes the following steps: Potential network face recognition data of the target user is acquired; Potential network face recognition data corresponding to a face matching value greater than the target face recognition threshold in the potential network face recognition data of the target user is taken as network face recognition data of the target user; The network face risk entropy of the target user is determined according to the network face recognition data of the target user.

5. The method of claim 4, wherein, The potential network face recognition data of the target user is acquired, and specifically includes the following steps: The similarity entropy between the target user and each other user is determined; Other users with a similarity entropy greater than a similarity threshold are taken as similar users of the target user; The historical network face recognition data of the target user and the historical network face recognition data of the similar users of the target user are taken as the potential network face recognition data of the target user.

6. The method of claim 4, wherein, The network face risk entropy of the target user is determined according to the network face recognition data of the target user, and specifically includes the following steps: Multiple network face dimensions are selected, wherein the network face dimensions are used to divide the network face recognition data, so that the intersection between the network face recognition data sets corresponding to each network face dimension is empty; The network face recognition data is divided according to the multiple network face dimensions to obtain the network face recognition data set of each network face dimension; The number of risk face recognitions in the network face recognition data set of each network face dimension is taken as the face recognition number of the network face dimension; The ratio of the face recognition number of the network face dimension to the face recognition number contained in the network face recognition data is taken as the network face risk probability of the network face dimension; The sum of the network face risk probabilities of all network face dimensions is determined, and the difference between the sum value threshold and the sum value is taken as the network face security probability; The network face risk entropy of the target user is determined according to the network face security probability, the network face risk probability of each network face dimension: Wherein, Q is the network face risk entropy of the target user, q is the network face security probability of the target user, q j is the network face risk probability of the jth network face dimension.

7. The method of claim 1, wherein, The network face recognition threshold of the target user is determined according to the functional relationship between the network face risk entropy of the target user and the network face recognition threshold and the network face risk entropy, and specifically includes the following steps: The functional relationship between the network face risk entropy of the target user and the network face recognition threshold is used to determine multiple network face recognition threshold intervals, wherein the network face risk entropy of the functional relationship in the network face recognition threshold interval is less than or equal to the network face risk entropy of the target user; The network face recognition threshold of the target user is determined according to the multiple network face recognition threshold intervals.

8. The method of claim 7, wherein, The network face recognition threshold of the target user is determined according to the multiple network face recognition threshold intervals, and specifically includes the following steps: The minimum point of the functional relationship in each network face recognition threshold interval is taken as a candidate minimum point; The network-free face recognition threshold corresponding to the candidate minimum value point is taken as a candidate face recognition threshold; A network-free face recognition threshold of a target user is determined according to the candidate face recognition threshold.

9. A server, characterized by The device comprises a processor and a memory connected to the processor in communication; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are executed by the processor to implement the method according to any one of claims 1 to 8.

Images