Privacy-computing-based risk identification method and device, and electronic device

Abstract

One or more embodiments of the specification disclose a privacy computing-based risk identification method and device and electronic equipment. The method first acquires a target face feature to be queried and target certificate information corresponding to the target face feature, performs privacy protection processing and homomorphic encryption processing on the target face feature based on a garbled circuit, and obtains homomorphic encrypted target face features. Secondly, a query request is generated based on the homomorphic encrypted target face features, and the query request is sent to a server. Then, a matching result and user certificate information corresponding to each face feature, which are encrypted based on a shamir secret sharing algorithm and homomorphic encryption, are received and sent by the server, and the user certificate information is used to determine a first face feature and its corresponding homomorphic encrypted user certificate information. Finally, it is determined whether the target certificate information corresponding to the target face feature is consistent with the homomorphic encrypted user certificate information corresponding to the first face feature.

Description

Technical Field

[0001] This document relates to the field of risk identification technology, and in particular to a risk identification method, device and electronic device based on privacy computing. Background Technology

[0002] As users become increasingly concerned about their privacy, identifying potential risks associated with users' illegal financial activities requires ensuring user privacy while simultaneously assessing their potential for fraudulent activity. Taking eKYC (electronic Know Your Customer) as an example, users typically authenticate themselves by taking photos of their identification documents and faces. However, this method is vulnerable to being exploited by forging user documents to pass eKYC verification. This risk can be addressed by using methods that detect multiple documents with the same face. Therefore, a technical solution is needed that ensures privacy compliance for information exchange between different merchants while also enabling cross-merchant eKYC risk identification using multiple documents with the same face. Summary of the Invention

[0003] On one hand, one or more embodiments of this specification provide a risk identification method based on privacy computing, applied to a terminal, comprising: acquiring a target facial feature to be queried and target document information corresponding to the target facial feature; performing privacy protection processing on the target facial feature based on an obfuscation circuit, and performing homomorphic encryption processing on the privacy-protected target facial feature to obtain a homomorphically encrypted target facial feature; generating a query request based on the homomorphically encrypted target facial feature, and sending the query request to a server, wherein the query request is used to trigger the server to compare the homomorphically encrypted target facial feature with a plurality of pre-stored privacy-protected facial features. Each facial feature is matched to obtain a corresponding matching result; the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are received from the server; based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature is determined, as well as the homomorphically encrypted user identification information corresponding to the first facial feature; it is then determined whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0004] On the other hand, one or more embodiments of this specification provide a risk identification method based on privacy computing, applied to a server, comprising: performing privacy protection processing on multiple pre-stored facial features, and performing homomorphic encryption processing on user identification information corresponding to each facial feature, followed by encryption processing based on the Shamir secret sharing algorithm, to obtain user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted; obtaining a query request sent by a terminal, and matching the homomorphically encrypted target facial feature with each of the multiple pre-stored privacy-protected facial features according to the query request, to obtain corresponding matching results; and pairing the matching results with each facial feature... The corresponding user identification information, encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, is sent to the terminal. The matching result and the user identification information corresponding to each facial feature, encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, are used to trigger the terminal to determine, based on the matching result and the user identification information corresponding to each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, and to determine whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0005] Furthermore, one or more embodiments of this specification provide a risk identification device based on privacy computing, comprising: a first privacy processing module, which acquires a target facial feature to be queried and target document information corresponding to the target facial feature, performs privacy protection processing on the target facial feature based on an obfuscation circuit, and performs homomorphic encryption processing on the privacy-protected target facial feature to obtain a homomorphically encrypted target facial feature; and a query request generation module, which generates a query request based on the homomorphically encrypted target facial feature and sends the query request to a server, wherein the query request is used to trigger the server to compare the homomorphically encrypted target facial feature with a plurality of pre-stored privacy-protected facial features. The system matches each facial feature to obtain a corresponding matching result; the information determination module receives the matching result and the user ID information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each facial feature sent by the server, and determines the first facial feature that matches the homomorphically encrypted target facial feature and the homomorphically encrypted user ID information corresponding to the first facial feature based on the matching result and the user ID information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each facial feature; the judgment module judges whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

[0006] Furthermore, one or more embodiments of this specification provide a risk identification device based on privacy computing, comprising: a second privacy processing module, which performs privacy protection processing on multiple pre-stored facial features, and performs homomorphic encryption processing on user identification information corresponding to each facial feature, and then performs encryption processing based on the Shamir secret sharing algorithm, to obtain user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted; a matching module, which obtains a query request sent by a terminal, and matches the homomorphically encrypted target facial feature with each of the multiple pre-stored privacy-protected facial features according to the query request, to obtain a corresponding matching result; and an information sending module, which transmits the matching result and the... Each facial feature is associated with user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted. This information is then sent to the terminal. The matching result and the user identification information associated with each facial feature are used to trigger the terminal to determine, based on the matching result and the user identification information associated with each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature. The terminal then determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0007] Furthermore, one or more embodiments of this specification provide a risk identification system based on privacy computing, comprising: a terminal and a server, wherein the server performs privacy protection processing on multiple pre-stored facial features, and performs homomorphic encryption processing on user identification information corresponding to each facial feature, and then performs encryption processing based on the Shamir secret sharing algorithm to obtain user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted; the terminal acquires the target facial feature to be queried and the target identification information corresponding to the target facial feature, performs privacy protection processing on the target facial feature based on a confusion circuit, and performs homomorphic encryption processing on the privacy-protected target facial feature to obtain homomorphically encrypted target facial feature, and generates a query request based on the homomorphically encrypted target facial feature, and sends the query request to... The query request is sent to the server; the server matches the homomorphically encrypted target facial feature with each of the pre-stored privacy-protected facial features to obtain a corresponding matching result, and sends the matching result and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each facial feature to the terminal; the terminal, based on the matching result and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each facial feature, determines the first facial feature that matches the homomorphically encrypted target facial feature and the homomorphically encrypted user identification information corresponding to the first facial feature, and determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0008] Furthermore, one or more embodiments of this specification provide an electronic device, including: a processor; and a memory arranged to store computer-executable instructions, wherein when the executable instructions are executed, the processor is able to: acquire a target facial feature to be queried and target identification information corresponding to the target facial feature; perform privacy protection processing on the target facial feature based on an obfuscation circuit, and perform homomorphic encryption processing on the privacy-protected target facial feature to obtain a homomorphically encrypted target facial feature; generate a query request based on the homomorphically encrypted target facial feature, and send the query request to a server, wherein the query request is used to trigger the server to compare the homomorphically encrypted target facial feature with pre-stored... The system matches each facial feature among multiple privacy-protected facial features to obtain a corresponding matching result; it receives the matching result and the user identification information corresponding to each facial feature, which is encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, sent by the server; and determines, based on the matching result and the user identification information corresponding to each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature; and determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0009] Furthermore, one or more embodiments of this specification provide an electronic device, including: a processor; and a memory arranged to store computer-executable instructions, wherein when the executable instructions are executed, the processor is able to: perform privacy protection processing on a pre-stored plurality of facial features, and perform homomorphic encryption processing on user identification information corresponding to each facial feature, and then perform encryption processing based on the Shamir secret sharing algorithm, to obtain user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted; obtain a query request sent by a terminal, and match the homomorphically encrypted target facial feature with each of the pre-stored privacy-protected facial features according to the query request, to obtain a corresponding matching result; and... The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine, based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, and to determine whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0010] Furthermore, one or more embodiments of this specification provide a storage medium for storing computer-executable instructions. When executed by a processor, these instructions implement the following process: acquiring target facial features to be queried and target identification information corresponding to the target facial features; performing privacy protection processing on the target facial features based on an obfuscation circuit; and performing homomorphic encryption processing on the privacy-protected target facial features to obtain homomorphically encrypted target facial features; generating a query request based on the homomorphically encrypted target facial features; and sending the query request to a server. The query request triggers the server to compare the homomorphically encrypted target facial features with pre-stored privacy protection information. The system matches each facial feature among multiple processed facial features to obtain a corresponding matching result; it receives the matching result and the user identification information corresponding to each facial feature, which is encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, sent by the server; and determines, based on the matching result and the user identification information corresponding to each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature; and determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0011] Furthermore, one or more embodiments of this specification provide a storage medium for storing computer-executable instructions. When executed by a processor, these instructions implement the following process: performing privacy protection processing on multiple pre-stored facial features, and homomorphically encrypting the user identification information corresponding to each facial feature before further encrypting it using the Shamir secret sharing algorithm, resulting in user identification information encrypted using both the Shamir secret sharing algorithm and homomorphically; acquiring a query request sent by a terminal, and matching the homomorphically encrypted target facial feature with each of the pre-stored privacy-protected facial features according to the query request, obtaining corresponding matching results; and storing the matching results... The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine, based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, and to determine whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in one or more embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic flowchart of a risk identification method based on privacy computing according to an embodiment of this specification;

[0014] Figure 2 This is a schematic flowchart of a risk identification method based on privacy computing according to another embodiment of this specification;

[0015] Figure 3 This is a schematic diagram illustrating the implementation principle of risk identification based on privacy computing according to an embodiment of this specification;

[0016] Figure 4 This is a schematic block diagram of a privacy-based risk identification device according to an embodiment of this specification;

[0017] Figure 5 This is a schematic block diagram of another privacy-based risk identification device according to an embodiment of this specification;

[0018] Figure 6 This is a schematic block diagram of a privacy-based risk identification system according to an embodiment of this specification;

[0019] Figure 7 This is a schematic block diagram of an electronic device according to an embodiment of this specification. Detailed Implementation

[0020] This specification provides one or more embodiments of a risk identification method and apparatus based on privacy computing to address the current problems.

[0021] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this document.

[0022] like Figure 1 As shown in the embodiments of this specification, a risk identification method based on privacy computing is provided. The execution subject of this method is a terminal device, which can be a mobile phone, tablet computer, or a computer device such as a laptop or desktop computer. The method specifically includes the following steps:

[0023] In step S102, the target face features to be queried and the target document information corresponding to the target face features are obtained. The target face features are subjected to privacy protection processing based on the obfuscation circuit, and the target face features after privacy protection processing are subjected to homomorphic encryption processing to obtain the homomorphically encrypted target face features.

[0024] Taking the eKYC process as an example, users typically complete identity verification by taking pictures of their ID cards and faces. However, this method can often be exploited by forging user ID cards to pass eKYC verification. This risk can be addressed through a "same face, multiple IDs" retrieval method. This method identifies facial features that match the target user's facial features from a database pre-stored on the server, comparing the matching facial features with the target user's ID card information. It then determines whether the corresponding ID card information matches the target user's ID card information and counts the number of different ID card information corresponding to the target facial features. In the embodiments of this specification, the user ID card information can be information from an ID card that proves the user's identity, such as the ID card number or social security card number.

[0025] In the server database, each facial feature and its corresponding user identification information constitute a data entry. Users can send a query request to the server via their terminal. This query request searches the server's database for facial features highly similar to the target facial feature. The user corresponding to the target facial feature and the user corresponding to a highly similar facial feature are identified as the same user. Then, the identification information corresponding to the highly similar facial feature is checked to see if it matches the target identification information.

[0026] When performing risk identification using multiple facial recognition credentials, the terminal typically initiates the query request, while the server pre-stores multiple facial features and their corresponding user identification information. The privacy-based computation-based risk identification method described in this specification is applied to the terminal and is based on the query request perspective.

[0027] Garbled circuits (GC) are a type of cryptographic protocol that enables secure multi-party computation. They are typically used in scenarios where multiple communicating parties need to share input data and compute a single result using the same function, and the input data between parties is not publicly disclosed. Garbled circuits are implemented based on oblivious transfer (OT) protocols and logic circuits. They mask the true input data by encrypting and scrambling circuit values. This encryption and scrambling is done on a per-logic-gate basis, with each gate corresponding to a truth table. Common examples include adders, comparators, and multipliers. A circuit can consist of one or more logic gates, such as AND gates, NOT gates, OR gates, and NAND gates.

[0028] The obfuscation circuit described in this specification can be applied to the above-mentioned retrieval process. In this embodiment, after comprehensively considering various influencing factors, the obfuscation circuit can adopt a 2PC obfuscation circuit (i.e., a Boolean circuit for secure computation between two parties). A detailed explanation is given using a 2PC obfuscation circuit as an example. In practical applications, other forms of obfuscation circuits can also be used, and the specific configuration can be determined according to the actual situation. This specification embodiment performs privacy protection processing on the target facial features based on the obfuscation circuit, enabling encryption of the target facial features to be queried without the querying party being aware of the privacy processing. The method for privacy protection processing of the target facial features can be achieved using the AES (Advanced Encryption Standard) encryption algorithm, or the DES (Data Encryption Standard) encryption algorithm, etc.

[0029] Homomorphic encryption involves homomorphically encrypting the original data, performing specific calculations on the resulting ciphertext, and then homomorphically decrypting the result to obtain the plaintext. This is equivalent to performing the same calculations on the original plaintext data. Compared to general encryption methods, homomorphic encryption prioritizes data processing security, providing a function for processing encrypted data.

[0030] The method for homomorphic encryption of target facial features in the embodiments of this specification can be to perform fully homomorphic encryption (such as BFV (Brakerski / Fan-Vercauteren) fully homomorphic encryption) on the target facial features.

[0031] In step S104, a query request is generated based on the homomorphically encrypted target face features and sent to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each of the pre-stored privacy-protected face features to obtain the corresponding matching results.

[0032] This embodiment of the specification sends a query request to the server via a terminal. The query request includes homomorphically encrypted target facial features. Specifically, the target facial features to be queried are first processed for privacy protection using an obfuscated circuit, and then homomorphically encrypted to obtain homomorphically encrypted target facial features. A query request is then generated based on these homomorphically encrypted target facial features. This query request triggers the server to perform facial feature matching in an encrypted state, thereby retrieving facial features that match the target facial features from multiple privacy-protected facial features pre-stored by the server. Because the target facial features are homomorphically encrypted, the queried party only performs facial feature matching based on the query request and cannot obtain information about the target facial features. Although a matching result can be obtained, information similar to the target facial features among multiple facial features cannot be obtained.

[0033] In step S106, the matching result sent by the server and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature are received. Based on the matching result and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature, the first face feature that matches the homomorphically encrypted target face feature and the user identification information corresponding to the first face feature are determined respectively.

[0034] In the embodiments of this specification, the homomorphically encrypted user ID information corresponding to the first facial feature comes from the user ID information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm stored on the server.

[0035] After receiving the matching results from the server, the terminal, having stored information about the target facial features, can determine whether the server contains the first facial feature and the number of such features, based on the received matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphic encryption pre-stored on the server. However, the terminal does not contain the actual content of each homomorphically encrypted facial feature pre-stored on the server. Therefore, while the terminal can determine whether the first facial feature exists on the server and the number of such features, it cannot obtain the actual content of each facial feature; the server, on the other hand, only performs a query operation. However, it cannot determine the true content of the facial features in the query. Moreover, the server can match the target facial features with each of the pre-stored privacy-protected facial features while keeping the data private. The terminal can determine the first facial feature that matches the homomorphically encrypted target facial feature based on the matching results and the user's identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature while keeping the data private. It can be seen that the terminal and the server can achieve privacy intersection while keeping the data private, thereby realizing risk identification and effectively improving the security of data in the risk identification process.

[0036] It should be noted that the first face feature matching the homomorphically encrypted target face feature in the embodiments of this specification is a face feature that is highly similar to the target face feature and is pre-stored in the server. These two face features can be considered to belong to the same user. Specifically, a similarity threshold can be preset, and homomorphically encrypted face features whose similarity to the homomorphically encrypted target face feature exceeds the similarity threshold can be defined as the first face feature. Alternatively, a homomorphically encrypted face feature that is completely identical to the homomorphically encrypted target face feature can be defined as the first face feature.

[0037] In step S108, it is determined whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

[0038] After determining the first facial feature and the homomorphically encrypted user ID information corresponding to the first facial feature in step S106, the target ID information is compared with the homomorphically encrypted user ID information corresponding to the first facial feature. If the two are consistent, it is determined that the target facial feature corresponds to only one user ID information and there is no situation of multiple IDs for the same face. If the two are inconsistent, it is determined that the target facial feature corresponds to one or more user ID information and there is a situation of multiple IDs for the same face. The specific number of multiple IDs for the same face is determined according to the number of the first facial features.

[0039] This specification provides a risk identification method based on privacy computing. Applied to a terminal, the method acquires the target facial features and their corresponding identification information. It then performs privacy protection processing on the target facial features using a scrambling circuit, followed by homomorphic encryption to obtain homomorphically encrypted target facial features. A query request is then generated based on these homomorphically encrypted features and sent to a server. The method receives matching results from the server and user identification information encrypted using the Shamir secret sharing algorithm and homomorphically, corresponding to each facial feature. Based on the matching results and the user identification information, a first facial feature matching the homomorphically encrypted target facial features is determined, along with the corresponding homomorphically encrypted user identification information. Finally, it determines whether the target identification information corresponding to the target facial feature matches the homomorphically encrypted user identification information corresponding to the first facial feature. The privacy protection processing of the target facial features using a scrambling circuit ensures the privacy and security of the data of both the querying and queried parties. When data is transmitted between the terminal and the server, the terminal performs homomorphic encryption on the target facial features after privacy protection processing to obtain homomorphically encrypted target facial features. The matching results and user identification information sent by the server to the terminal are also homomorphically encrypted. This allows the querying party to only determine whether there is a risk of multiple identifications for the same face and the number of multiple user identifications, but not to obtain the specific content of multiple user identification information. The querying party only performs the query operation, but cannot determine the real content of the facial features in the query. Thus, the method of privacy intersection and homomorphic encryption realizes the risk identification of multiple identifications for the same face across merchants' eKYC, which is conducive to improving the security of privacy data in the risk identification process.

[0040] Furthermore, there are multiple methods for performing privacy protection processing on the target facial features based on the obfuscation circuit in step S102. The following is one optional method:

[0041] In step S1022, the target face features are masked in the confusion circuit based on multiple pre-generated masks to obtain the masked target face features.

[0042] In implementation, the method for masking target facial features can be as follows: the server pre-generates multiple masks and applies them to the terminal through obfuscation circuitry. At the terminal, the target facial features are logically ANDed with each of the pre-generated masks to obtain the masked target facial features. The mask can be composed of an array of 0s and 1s. This fast and effective privacy-preserving method of masking target facial features further improves the privacy security and processing efficiency of target facial features during data transmission.

[0043] In step S1024, the target face features after masking are encrypted using the AES encryption algorithm based on the pre-generated key in the obfuscation circuit to obtain the target face features after privacy protection processing.

[0044] The key is pre-generated by the server and applied to the terminal through a scrambling circuit. The pre-generated key is matched with the encryption algorithm used. In step S1024, the AES encryption algorithm is used to encrypt the target facial features. Correspondingly, the pre-generated key is also an AES key. The AES key can be generated using the standard AES key generation process, which will not be elaborated here.

[0045] In the embodiments of this specification, the AES encryption algorithm is used to encrypt the target facial features after masking. The AES encryption algorithm is a two-way encryption algorithm (also known as symmetric encryption). The AES encryption algorithm has a fast encryption speed, which can not only effectively protect the privacy of the target facial features, but also improve the efficiency of risk identification.

[0046] Furthermore, the processing in step S106 above can be carried out in various ways. The following is one optional processing method, which can be found in the following steps S1062-S1066:

[0047] In step S1062, each matching result is homomorphically decrypted to obtain the corresponding decrypted matching result.

[0048] Since the matching results from the server are obtained by matching the homomorphically encrypted target face features with each of the pre-stored privacy-protected face features, and the homomorphically encrypted target face features come from the terminal, specifically by homomorphically encrypting the privacy-protected target face features, the result obtained by homomorphically decrypting each matching result in step S1062 is the result after privacy protection processing (i.e., encryption using the AES encryption algorithm) and homomorphic decryption, which is the decrypted matching result.

[0049] In step S1064, based on the decrypted matching result, a first face feature that matches the homomorphically encrypted target face feature is determined.

[0050] In step S1066, based on the first facial feature and the Shamir secret sharing algorithm, the user ID information corresponding to each facial feature, which is encrypted and homomorphically encrypted using the Shamir secret sharing algorithm, is decrypted to obtain the homomorphically encrypted user ID information corresponding to the first facial feature.

[0051] The decryption process here refers to the decryption process after encryption based on the Shamir secret sharing algorithm. Therefore, by step S1066, the user ID information corresponding to each face feature after encryption based on the Shamir secret sharing algorithm and homomorphic encryption is decrypted to obtain the user ID information corresponding to the first face feature after homomorphic encryption.

[0052] It is important to note that during steps S1064 and S1066, in order to determine the homomorphically encrypted user identification information corresponding to the first facial feature, the terminal first needs to determine the first facial feature that matches the homomorphically encrypted target facial feature. Therefore, the terminal does not obtain the actual content of the first facial feature; what it obtains is the user identification information corresponding to the first facial feature. Thus, for both the terminal and the server, the actual content of the first facial feature is protected as private data.

[0053] Furthermore, the processing in step S108 above can be carried out in various ways. The following is one optional processing method, which can be found in the following steps S1082-S1086:

[0054] In step S1082, the homomorphically encrypted user ID information corresponding to the first face feature is homomorphically encrypted again to obtain the user ID information corresponding to the first face feature after two homomorphic encryptions.

[0055] In practice, since the first homomorphic encryption process of the user ID information corresponding to the first facial feature is executed on the server, the terminal cannot decrypt it. Through step S1082, the user ID information is homomorphically encrypted again on the terminal, which can ensure that the user ID information processed by the terminal is in a homomorphic encrypted state and facilitates subsequent processing of encrypted data based on relevant calculations.

[0056] In step S1084, a similarity comparison is performed between the target ID information corresponding to the target facial feature and the user ID information after two homomorphic encryptions corresponding to the first facial feature. The similarity comparison result of the ID information is then homomorphically encrypted and transmitted to the server.

[0057] The target ID information corresponding to the target facial feature is the original information without encryption, while the user ID information corresponding to the first facial feature is the user ID information after two homomorphic encryptions. Based on the principle of homomorphic encryption (the calculation result on the ciphertext is equivalent to the calculation result on the plaintext), it can be known that the result of comparing the similarity of the two user ID information is consistent with the result of comparing the similarity of the original data of the two user ID information. Therefore, the similarity comparison based on the homomorphic encryption method in the embodiments of this specification can achieve risk identification while ensuring data privacy and security.

[0058] The first homomorphic encryption process of the user's ID information corresponding to the first facial feature is performed on the server. The terminal cannot decrypt it. In order to determine the similarity comparison result, the terminal homomorphically encrypts the similarity comparison result of the ID information before transmitting it to the server. This method ensures that the server can process the similarity comparison result, but cannot obtain the real content of the similarity comparison result, thereby ensuring the privacy and security of the similarity comparison result.

[0059] In step S1086, the receiving server performs a first decryption of the homomorphically encrypted similarity comparison result, performs a second decryption of the first decryption result, and determines the consistency between the target document information and the homomorphically encrypted user document information based on the second decryption result.

[0060] In implementation, assuming the homomorphically encrypted user ID information corresponding to the first facial feature is A, and the target ID information corresponding to the target facial feature is B, the terminal needs to calculate C = AB to determine whether A and B are the same. If the plaintext corresponding to C after decryption is 0, then A and B are determined to be the same; otherwise, A and B are determined to be different. The decryption process requires sending the similarity comparison result to the server. To prevent the server from obtaining the true content of the similarity comparison result C, C is encrypted in step S1084 and transmitted to the server. The server performs a first decryption. In step S1086, the terminal receives the result of the first decryption from the server and performs a second decryption on the result of the first decryption to determine whether C is 0, and thus determine whether the target ID information is the same as the homomorphically encrypted user ID information.

[0061] This specification provides a risk identification method based on privacy computing. The method is applied to a terminal acting as the querying party. After obtaining the target facial features and corresponding target identification information, the method performs privacy protection processing on the target facial features using a scrambling circuit, and then performs homomorphic encryption on the privacy-protected facial features to obtain homomorphically encrypted target facial features. A query request is then generated based on the homomorphically encrypted target facial features and sent to a server. The server sends matching results and user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature. Based on the matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, the method determines a first facial feature that matches the homomorphically encrypted target facial features, and the homomorphically encrypted user identification information corresponding to the first facial feature. Finally, it determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature. The privacy protection processing of the target facial features using a scrambling circuit ensures the privacy and security of the querying party's data. When data is transmitted between the terminal and the server, the terminal performs homomorphic encryption on the target facial features after privacy protection processing to obtain homomorphically encrypted target facial features. The matching results and user identification information sent by the server to the terminal are also homomorphically encrypted. This allows the querying party to only determine whether there is a risk of multiple identity documents for the same face and the number of multiple user identification documents, but not to obtain the specific content of multiple user identification documents. The querying party only performs the query operation, but cannot determine the real content of the facial features in the query. Thus, the method of privacy intersection and homomorphic encryption realizes the risk identification of multiple identity documents for the same face across merchants' eKYC, which is conducive to improving the security of privacy data in the risk identification process.

[0062] like Figure 2 As shown in the embodiments of this specification, another risk identification method based on privacy computing is provided. The execution subject of this method is a server, which can be a single independent server or a server cluster composed of multiple servers. The server can be a backend server for financial transactions or online shopping transactions, or a backend server for an application. The method specifically includes the following steps:

[0063] In step S202, privacy protection processing is performed on the pre-stored multiple facial features, and the user ID information corresponding to each facial feature is homomorphically encrypted and then encrypted based on the Shamir secret sharing algorithm to obtain the user ID information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted.

[0064] In implementation, multiple facial features and their corresponding user identification information are first pre-stored in a database on the server. In the server database, each facial feature and its corresponding user identification information constitutes a single data entry. Data storage can be performed using Hamming space or Euclidean space.

[0065] After pre-storing multiple facial features and their corresponding user identification information, appropriate privacy protection processing is required for subsequent secure data transmission. Specifically, methods for privacy protection processing of multiple facial features include using masks, encryption, or sequentially applying masks and encryption algorithms to obtain the privacy-protected facial features. The user identification information corresponding to each facial feature is then homomorphically encrypted and further encrypted using the Shamir secret sharing algorithm. This process ensures that the user identification information remains privacy-protected during data transmission and retrieval.

[0066] In step S204, the query request sent by the terminal is obtained, and the homomorphically encrypted target face feature is matched with each face feature in the multiple face features that have been pre-stored after privacy protection processing, according to the query request, to obtain the corresponding matching result.

[0067] The method for matching the homomorphically encrypted target facial features with each of the pre-stored privacy-preserving facial features can employ either a similarity comparison method or a method that determines whether the two are identical. If a similarity comparison method is used, the Euclidean distance between the homomorphically encrypted target facial features and each of the pre-stored privacy-preserving facial features can be calculated, or the Hamming distance can be calculated. If a similarity comparison method is used, the specific distance calculation method can be determined based on the data storage space specified in step S202.

[0068] In step S206, the matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine the first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, based on the matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature. The terminal also determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0069] Other parts of the embodiments in this specification that are not described in detail can be found in [the relevant documentation]. Figure 1 The embodiments shown will not be described in detail here.

[0070] This specification provides a privacy-preserving computation-based risk identification method. The method is applied to a server, where the data of the queried party is pre-stored. First, multiple pre-stored facial features undergo privacy protection processing. Then, the user identification information corresponding to each facial feature is homomorphically encrypted and further encrypted using the Shamir secret sharing algorithm. This yields the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically. Next, a query request sent by the terminal is obtained. Based on the query request, the homomorphically encrypted target facial feature is matched against each of the pre-stored privacy-preserving facial features to obtain the corresponding matching results. Finally, the matching results are... The matching result and the user identification information corresponding to each facial feature, encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, are sent to the terminal. This matching result and the corresponding user identification information are used to trigger the terminal to determine, based on the matching result and the corresponding user identification information, the first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature. The terminal then determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature. By performing privacy protection processing on multiple pre-stored facial features and homomorphically encrypting the user identification information corresponding to each facial feature, the privacy and security of the queried data can be ensured. By homomorphically encrypting user identification information and matching the homomorphically encrypted target facial features with each of the pre-stored privacy-protected facial features, it is possible to ensure that the data is in a privacy-protected state during processing. The queried party only performs matching and query processing according to the query request, but cannot determine the true content of the facial features in the query. Thus, the method of privacy intersection and homomorphic encryption enables risk identification of multiple identity documents with the same face across merchants' eKYC, which helps to improve the security of privacy data during risk identification.

[0071] Furthermore, in step S202, multiple facial features and their corresponding user identification information can be stored using various methods. This specification embodiment provides a method for pre-storing these features using Hamming space. Correspondingly, the method for privacy protection processing of the pre-stored multiple facial features in step S202 may include the following steps:

[0072] In step S2022, multiple masks are randomly generated, and each facial feature is masked based on the generated multiple masks to obtain the masked facial features.

[0073] For example: if 64 masks are randomly generated and a facial features are pre-stored, then performing a logical AND operation between each mask and each facial feature requires 64*a logical AND operations. Each facial feature corresponds to a facial feature processed by 64 masks.

[0074] It is important to note that the number of randomly generated masks on the server is related to... Figure 1 The number of masks in step S1022 of the illustrated embodiment is consistent, thereby ensuring that the terminal and the server perform relevant matching and processing operations on the data.

[0075] In step S2024, a key is generated in advance, and based on the generated key, the AES encryption algorithm is used to encrypt each facial feature after masking to obtain the privacy-protected facial features.

[0076] The pre-generated key in step S2024 is also used Figure 1 In the AES encryption algorithm of step S1024 of the embodiment shown, it is ensured that the target face feature of the terminal and each face feature after privacy protection processing in the server are based on the same key and the same encryption algorithm, so that the two are comparable and the first face feature can be determined.

[0077] Furthermore, in step S202, the user ID information corresponding to each facial feature is homomorphically encrypted and then encrypted using the Shamir secret sharing algorithm to obtain the user ID information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted. Various processing methods can be used. The following provides an optional processing method. For details, please refer to the processing steps S2026-S2028.

[0078] S2026: Homomorphically encrypt the user ID information corresponding to each facial feature to obtain the homomorphically encrypted user ID information corresponding to each facial feature.

[0079] S2028: Based on the Shamir secret sharing algorithm, the homomorphically encrypted user ID information corresponding to each facial feature is divided into multiple user ID information subkeys. Based on the preset decryption rules, user ID information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted is obtained. The decryption rules include: if the number of user ID information subkeys held exceeds the preset subkey number threshold, the information obtained after encrypting the homomorphically encrypted user ID information corresponding to each facial feature based on the Shamir secret sharing algorithm is decrypted to obtain the homomorphically encrypted user ID information corresponding to each facial feature.

[0080] As can be seen from the above steps S2026-S2028, after homomorphically encrypting the user ID information corresponding to each facial feature, the homomorphically encrypted user ID information is further encrypted based on the Shamir secret sharing algorithm, and corresponding decryption rules are set, thereby further improving the data security in the risk identification process.

[0081] The Shamir Secret Sharing algorithm, also known as the Shamir Secret Sharing algorithm, is a threshold secret partitioning scheme based on the Lagrange interpolation formula. Specifically, it divides the secret *s* into *n* unrelated parts, each called a subkey. Each subkey is held by a participant, and the secret *s* can be recovered when at least *t* subkeys are possessed. This method is also known as the (t, n) secret partitioning threshold algorithm, where *t* is called the threshold value of the algorithm.

[0082] It should be noted that the number of subkeys in the Shamir secret sharing algorithm in this embodiment is the same as the number of masks in step S2022, that is: the result of masking each facial feature corresponds to one subkey.

[0083] In one embodiment, the threshold for the number of subkeys in the decryption rule can be set according to the actual situation, such as 2 or 5. Taking a threshold of 2 for the number of subkeys in the decryption rule as an example, when the number of subkeys for the user's ID information exceeds 2, the user's ID information corresponding to each face feature, which has been homomorphically encrypted and encrypted based on the Shamir secret sharing algorithm, can be decrypted to recover the homomorphically encrypted user's ID information corresponding to each face feature.

[0084] Furthermore, the processing in step S204 above can be performed in multiple ways. The following provides one optional processing method, which can be found in the following steps S2042 and S2044:

[0085] In step S2042, the Hamming distance between the homomorphically encrypted target face feature and each pre-stored privacy-processed face feature is calculated according to the preset feature algorithm.

[0086] Hamming distance is commonly used in data transmission error control coding. The Hamming distance between two strings of equal length is the number of different characters at corresponding positions in the two strings; that is, the number of characters that need to be replaced to transform one string into the other. For example, the Hamming distance between 1011101 and 1001001 is 2, the Hamming distance between 2143896 and 2233796 is 3, and the Hamming distance between "toned" and "roses" is 3.

[0087] The preset feature algorithm can be feature subtraction. Based on feature subtraction, the target face feature is subtracted from the corresponding bits of each face feature after privacy processing, and the Hamming distance is obtained. Then, the similarity of face features is calculated based on the Hamming distance, and face features similar to the target face feature are found.

[0088] In step S2044, the corresponding matching result is determined based on the Hamming distance.

[0089] Since each facial feature is masked in step S2022, combined with the Hamming distance in step S2042, the process of comparing the distance of facial features in Hamming space through the mask is transformed into a process of comparing whether corresponding bits are equal, thereby greatly improving the efficiency of facial feature matching and the accuracy of matching results, and thus improving the efficiency of risk identification.

[0090] See Figure 3 , Figure 3 This is a schematic diagram illustrating the implementation principle of privacy-based risk identification according to an embodiment of this specification. Figure 3 As can be seen, the risk identification method based on privacy computing in the embodiments of this specification mainly involves two parties: the client (terminal) and the server. The homomorphic encryption method uses BFV encryption, and takes the random generation of 64 masks as an example. Figure 2 The risk identification process and principles for both the Client and Server sides are as follows:

[0091] 1) The server stores multiple facial features and their corresponding user identification information in Hamming space. Before each query, 64 masks are randomly generated, defined as m1, m2...m64. Each facial feature is ANDed with each mask to obtain each masked facial feature. Each masked facial feature is then encrypted using a pre-generated AES key (AES key_s) to obtain a privacy-protected facial feature, defined as: F_ij = AES(AES key_s, F_i and mask_j). The more similar two facial features F1 and F2 are in the original space, the larger the intersection of the resulting F_ij. It can be defined that if the number of intersections between two facial features in Hamming space exceeds a preset threshold, then the two facial features F1 and F2 correspond to the same user. Here, the preset threshold can be set to 2.

[0092] Meanwhile, after the server encrypts the user's identification information using BFV, it then uses the Shamir Secret Sharing algorithm to divide the user's identification information into T shares (subkeys). With a prerequisite of holding at least 2 shares, the Shamir Secret Sharing algorithm can be used to decrypt the information, thereby recovering the homomorphically encrypted user's identification information.

[0093] 2) The client initiates the query. First, a 2PC obfuscation circuit is used to mask and encrypt the target face feature to be queried by the client. Then, the client performs BFV encryption on the AES-encrypted target face feature to obtain the homomorphically encrypted target face feature BFV_en(F), and transmits it to the server.

[0094] Depend on Figure 3 It can be seen that when using the obfuscation circuit, the client inputs the target face feature to be queried into the obfuscation circuit, and the server inputs the randomly generated mask and the pre-generated key into the obfuscation circuit. The obfuscation circuit outputs 64 feature sub-items F_1, F_2...F_64 corresponding to the target face feature. By performing mask processing and AES encryption processing on the target face feature to be queried in the obfuscation circuit, the privacy and security protection of the data can be achieved.

[0095] 3) After performing feature subtraction, the server adds the user's identification information to obtain the share, and returns it to the client. Taking a single comparison as an example, the server returns BFV_en(F-Fi)*r+si to the client, where r is a random factor and si is the subkey corresponding to the feature item Fi.

[0096] The server returns BFV_en(F-Fi)*r+si to the client, ensuring that the client cannot directly decrypt BFV_en(F-Fi). Furthermore, the client cannot determine whether the value of F-Fi is 0 (i.e., whether F and Fi are the same), nor can it determine how many of the 64 possible F-Fi values ​​are 0 (i.e., how many Fi values ​​are similar to F). The client can only determine whether the number of 0 F-Fi values ​​exceeds a set threshold, which is based on the pre-defined decryption rules in the Shamir secret sharing algorithm. Therefore, this homomorphic encryption method allows the terminal and server to transmit and compute data while maintaining data privacy, thus greatly improving data security during risk identification.

[0097] Setting the random factor r ensures that the Cl ient method cannot determine the specific value of BFV_en(F-Fi).

[0098] 4) The Cl ient decrypts the returned result. If the feature sub-item F_i is equal, the corresponding share can be obtained. According to the preset decryption rules, if at least 2 shares are obtained, the user's identity information encrypted by the Server using the Shamir Secret Sharing algorithm can be decrypted. At this time, the user's identity information is homomorphically encrypted.

[0099] The client subtracts the decrypted user ID information from the target ID information corresponding to the current target face feature, then performs BFV encryption to obtain [[ID_i-ID]], which is then transmitted to the server.

[0100] 5) The server performs a decryption to obtain the information [ID_i-ID] obtained by subtracting the target ID information from the homomorphically encrypted user ID information corresponding to the first facial feature, and transmits it to the client.

[0101] 6) Cl ient performs another decryption to obtain ID_i-ID. If the result is 0, it means that the target ID information is the same as the homomorphically encrypted user ID information corresponding to the first face feature; otherwise, they are different.

[0102] This specification provides a risk identification method based on privacy computing. The method is applied to a server acting as the query target. First, multiple pre-stored facial features undergo privacy protection processing. Then, the user identification information corresponding to each facial feature is homomorphically encrypted and further encrypted using the Shamir secret sharing algorithm. This yields user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted. Next, a query request from a terminal is received. Based on the query request, the homomorphically encrypted target facial feature is matched against each of the pre-stored privacy-protected facial features to obtain matching results. Finally, the matching results and the homomorphically encrypted user identification information corresponding to each facial feature are sent to the terminal. These matching results and the homomorphically encrypted user identification information trigger the terminal to determine, based on the matching results and the homomorphically encrypted user identification information corresponding to each facial feature, a first facial feature matching the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature. The terminal then determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature. By performing privacy protection processing on multiple pre-stored facial features and homomorphically encrypting the user identification information corresponding to each facial feature, the privacy and security of the queried party's data can be ensured. By homomorphically encrypting the user identification information and matching the homomorphically encrypted target facial feature with each of the pre-stored privacy-protected facial features, the data remains in a privacy-protected state during processing. The queried party only performs matching and query processing according to the query request and cannot determine the true content of the facial features in the query. Therefore, based on privacy intersection and homomorphic encryption, risk identification of multiple identity documents with the same face across merchants' eKYC is achieved, which helps improve the security of privacy data during risk identification.

[0103] In summary, specific embodiments of this subject matter have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing can be advantageous.

[0104] The above describes a risk identification method based on privacy computing provided by one or more embodiments of this specification. Based on the same idea, one or more embodiments of this specification also provide a risk identification device based on privacy computing, such as... Figure 4 As shown.

[0105] The privacy-based computing-based risk identification device includes: a first privacy processing module 310, a query request generation module 320, an information determination module 330, and a judgment module 340, wherein:

[0106] The first privacy processing module 310 obtains the target face features to be queried and the target document information corresponding to the target face features, performs privacy protection processing on the target face features based on the obfuscation circuit, and performs homomorphic encryption processing on the target face features after privacy protection processing to obtain the homomorphically encrypted target face features.

[0107] The query request generation module 320 generates a query request based on the homomorphically encrypted target face features and sends the query request to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain the corresponding matching results.

[0108] The information determination module 330 receives the matching result sent by the server and the user ID information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature. Based on the matching result and the user ID information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature, it determines the first face feature that matches the homomorphically encrypted target face feature and the homomorphically encrypted user ID information corresponding to the first face feature.

[0109] The judgment module 340 judges whether the target document information corresponding to the target face feature is consistent with the homomorphically encrypted user document information corresponding to the first face feature.

[0110] In one embodiment, the information determination module 330 includes:

[0111] The homomorphic decryption unit performs homomorphic decryption on each matching result to obtain the corresponding decrypted matching result.

[0112] The first face feature determination unit determines the first face feature that matches the homomorphically encrypted target face feature based on the decrypted matching result.

[0113] The user identification information determination unit, based on the first facial feature and the Shamir secret sharing algorithm, decrypts the user identification information corresponding to each facial feature after encryption and homomorphic encryption based on the Shamir secret sharing algorithm, to obtain the homomorphically encrypted user identification information corresponding to the first facial feature.

[0114] In one embodiment, the determination module 340 includes:

[0115] The homomorphic encryption processing unit performs homomorphic encryption processing again on the homomorphically encrypted user ID information corresponding to the first face feature to obtain the user ID information corresponding to the first face feature after two homomorphic encryptions.

[0116] The similarity comparison unit performs a similarity comparison between the target ID information corresponding to the target facial feature and the user ID information after two homomorphic encryptions corresponding to the first facial feature, and transmits the similarity comparison result of the ID information to the server after homomorphic encryption.

[0117] The consistency determination unit receives the decryption result of the homomorphically encrypted similarity comparison result from the server after the first decryption, performs a second decryption on the first decryption result, and determines the consistency between the target document information and the homomorphically encrypted user document information based on the second decryption result.

[0118] In one embodiment, the first privacy processing module 310 includes:

[0119] The target facial feature and target document information acquisition unit acquires the target facial features to be queried and the target document information corresponding to the target facial features;

[0120] The mask processing unit performs mask processing on the target face features based on multiple pre-generated masks in the confusion circuit to obtain the masked target face features.

[0121] The target face feature privacy processing unit encrypts the masked target face features using the AES encryption algorithm based on a pre-generated key in the obfuscation circuit, thereby obtaining the privacy-protected target face features.

[0122] The target face feature homomorphic encryption unit performs homomorphic encryption on the target face features after privacy protection processing to obtain the homomorphically encrypted target face features.

[0123] This specification provides a risk identification device based on privacy computing. This device can be a terminal acting as a querying party. After obtaining the target facial features and corresponding identification information through a first privacy processing module, it performs privacy protection processing on the target facial features using an obfuscation circuit. Then, it performs homomorphic encryption on the privacy-protected target facial features and further encryption using the Shamir secret sharing algorithm, resulting in a target facial feature encrypted with both Shamir secret sharing and homomorphic encryption. Finally, a query request generation module generates a query request based on the homomorphically encrypted target facial features and sends the query request to the relevant authority. The request is sent to the server. The information determination module receives the matching results and the user identification information (encrypted using the Shamir secret sharing algorithm and homomorphically encrypted) corresponding to each facial feature. Based on the matching results and the user identification information (encrypted using the Shamir secret sharing algorithm and homomorphically encrypted) corresponding to each facial feature, it determines the first facial feature that matches the homomorphically encrypted target facial feature, and the corresponding homomorphically encrypted user identification information. Finally, the judgment module determines whether the target identification information corresponding to the target facial feature matches the homomorphically encrypted user identification information corresponding to the first facial feature. Using obfuscation circuits to perform privacy protection processing on the target facial features ensures the privacy and security of the queryer's data. When data is transmitted between the terminal and the server, the terminal performs homomorphic encryption on the target facial features after privacy protection processing to obtain homomorphically encrypted target facial features. The matching results and user identification information sent by the server to the terminal are also homomorphically encrypted. This allows the querying party to only determine whether there is a risk of multiple identity documents for the same face and the number of multiple user identification documents, but not to obtain the specific content of multiple user identification documents. The querying party only performs the query operation, but cannot determine the real content of the facial features in the query. Thus, the method of privacy intersection and homomorphic encryption realizes the risk identification of multiple identity documents for the same face across merchants' eKYC, which is conducive to improving the security of privacy data in the risk identification process.

[0124] Those skilled in the art will understand that the above-described privacy-based computation-based risk identification device can be used to achieve... Figure 1 The detailed description of the risk identification method based on privacy computing should be similar to that described in the previous method section, and will not be repeated here to avoid being cumbersome.

[0125] like Figure 5 As shown, one or more embodiments of this specification provide another privacy-based risk identification device, which includes: a second privacy processing module 410, a matching module 420, and an information sending module 430, wherein:

[0126] The second privacy processing module 410 performs privacy protection processing on multiple pre-stored facial features, and performs homomorphic encryption processing on the user ID information corresponding to each facial feature, and then performs encryption processing based on the Shamir secret sharing algorithm to obtain the user ID information encrypted based on the Shamir secret sharing algorithm and homomorphic encryption.

[0127] The matching module 420 obtains the query request sent by the terminal, and matches the homomorphically encrypted target face feature with each face feature in the multiple face features that have been pre-stored and processed for privacy protection, in accordance with the query request, to obtain the corresponding matching result.

[0128] The information sending module 430 sends the matching results and the user ID information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature to the terminal. The matching results and the user ID information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine the first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user ID information corresponding to the first facial feature, based on the matching results and the user ID information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature. The terminal also determines whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

[0129] In one embodiment, multiple facial features and their corresponding user identification information are pre-stored on a server in Hamming space format. The second privacy processing module 410 includes:

[0130] The mask generation unit randomly generates multiple masks and performs masking processing on each facial feature based on the generated multiple masks to obtain the masked facial features;

[0131] The face feature privacy processing unit uses a pre-generated key and, based on the pre-generated key, encrypts each face feature after masking using the AES encryption algorithm to obtain the face features after privacy protection processing.

[0132] The user identification information homomorphic encryption unit performs homomorphic encryption processing on the user identification information corresponding to each facial feature to obtain homomorphically encrypted user identification information.

[0133] In one embodiment, the second privacy processing module 410 includes:

[0134] The first encryption processing unit performs homomorphic encryption processing on the user ID information corresponding to each facial feature to obtain the homomorphically encrypted user ID information corresponding to each facial feature.

[0135] The second encryption processing unit, based on the Shamir secret sharing algorithm, divides the homomorphically encrypted user ID information corresponding to each facial feature into multiple user ID information subkeys. Based on preset decryption rules, it obtains user ID information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted. The decryption rules include: if the number of user ID information subkeys held exceeds a preset subkey number threshold, then the information obtained after encrypting the homomorphically encrypted user ID information corresponding to each facial feature using the Shamir secret sharing algorithm is decrypted to obtain the homomorphically encrypted user ID information corresponding to each facial feature.

[0136] In one embodiment, the matching module 420 includes:

[0137] The Hamming distance calculation unit calculates the Hamming distance between the homomorphically encrypted target face feature and each pre-stored privacy-processed face feature according to the preset feature algorithm.

[0138] The Hamming distance matching unit determines the corresponding matching result based on the Hamming distance.

[0139] This specification provides a risk identification device based on privacy computing. The device can be a server acting as the queried party. First, a second privacy processing module performs privacy protection processing on multiple pre-stored facial features. Then, it performs homomorphic encryption on the user identification information corresponding to each facial feature, followed by encryption using the Shamir secret sharing algorithm. This results in user identification information encrypted using both the Shamir secret sharing algorithm and homomorphically encrypted. Next, a matching module obtains the query request sent by the terminal and matches the homomorphically encrypted target facial feature with each of the pre-stored privacy-protected facial features according to the query request, obtaining the corresponding matching results. Finally, the information is sent... The sending module sends the matching results and the user identification information corresponding to each facial feature, encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, to the terminal. This matching result and the corresponding user identification information are used to trigger the terminal to determine, based on the matching result and the corresponding user identification information, the first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature. The terminal then determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature. By performing privacy protection processing on multiple pre-stored facial features and homomorphically encrypting the user identification information corresponding to each facial feature, the privacy and security of the queried data can be ensured. By homomorphically encrypting user identification information and matching the homomorphically encrypted target facial features with each of the pre-stored privacy-protected facial features, it is possible to ensure that the data is in a privacy-protected state during processing. The queried party only performs matching and query processing according to the query request of the querying party and has no knowledge of the query details. Thus, the risk identification of multiple identity documents with the same face across merchants is realized based on the privacy intersection and homomorphic encryption method, which helps to improve the security of privacy data in the risk identification process.

[0140] Those skilled in the art will understand that the above-described privacy-based computation-based risk identification device can be used to achieve... Figure 2 and Figure 3 The detailed description of the risk identification method based on privacy computing should be similar to that described in the previous method section, and will not be repeated here to avoid being cumbersome.

[0141] like Figure 6 As shown, one or more embodiments of this specification provide a privacy-based computation-based risk identification system, which includes: a terminal 510 and a server 520, wherein:

[0142] Server 520 performs privacy protection processing on multiple pre-stored facial features and homomorphically encrypts the user ID information corresponding to each facial feature to obtain homomorphically encrypted user ID information.

[0143] Terminal 510 acquires the target face features to be queried and the target document information corresponding to the target face features, performs privacy protection processing on the target face features based on the obfuscated circuit, performs homomorphic encryption processing on the privacy-protected target face features, and then performs encryption processing based on the Shamir secret sharing algorithm to obtain the target face features encrypted with the Shamir secret sharing algorithm and homomorphically encrypted, and generates a query request based on the homomorphically encrypted target face features and sends the query request to the server;

[0144] Server 520, according to the query request, matches the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain the corresponding matching results, and sends the matching results and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature to the terminal.

[0145] Terminal 510, based on the matching results and the user ID information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each face feature, determines the first face feature that matches the homomorphically encrypted target face feature and the homomorphically encrypted user ID information corresponding to the first face feature, and judges whether the target ID information corresponding to the target face feature is consistent with the homomorphically encrypted user ID information corresponding to the first face feature.

[0146] This specification provides a risk identification system based on privacy computing. The system involves a server performing privacy protection processing on pre-stored multiple facial features, homomorphically encrypting the user identification information corresponding to each facial feature, and then further encrypting it using the Shamir secret sharing algorithm to obtain the user identification information encrypted using both the Shamir secret sharing algorithm and homomorphically. A terminal obtains the target facial feature to be queried and the target identification information corresponding to the target facial feature. The target facial feature is then subjected to privacy protection processing using a confusion circuit, followed by homomorphic encryption to obtain the homomorphically encrypted target facial feature. A query request is generated based on the homomorphically encrypted target facial feature and sent to the server. The system then... The server, based on the query request, matches the homomorphically encrypted target facial feature with each of a pre-stored set of privacy-protected facial features to obtain the corresponding matching results. Then, it sends the matching results and the corresponding user identification information (encrypted using the Shamir secret sharing algorithm and homomorphically encrypted) to the terminal. Finally, the terminal, based on the matching results and the corresponding user identification information (encrypted using the Shamir secret sharing algorithm and homomorphically encrypted), determines the first facial feature that matches the homomorphically encrypted target facial feature and the corresponding homomorphically encrypted user identification information. It then checks whether the target identification information matches the first facial feature. During data transmission between the server and terminal, using obfuscation circuits to perform privacy protection processing on the target facial feature ensures the privacy and security of the querying party's data. The terminal performs homomorphic encryption on the target facial features after privacy protection processing, resulting in homomorphically encrypted target facial features. The user identification information on the server side and the matching results generated based on the terminal's query request are also homomorphically encrypted. This allows the querying party to only determine whether there is a risk of multiple identifications for the same face and the number of multiple user identifications, but not to obtain information about multiple user identifications. The queried party only performs the query operation but cannot determine the true content of the facial features in the query. Thus, the method of privacy intersection and homomorphic encryption enables risk identification of multiple identifications for the same face across merchants' eKYC, which helps improve the security of privacy data during the risk identification process.

[0147] Those skilled in the art will understand that the above-described privacy-based computation-based risk identification system can be used to implement... Figures 1-3 The detailed description of the risk identification method based on privacy computing should be similar to that described in the previous method section, and will not be repeated here to avoid being cumbersome.

[0148] Based on the same idea, one or more embodiments of this specification also provide an electronic device, such as... Figure 7As shown. Electronic devices can vary considerably due to differences in configuration or performance, and may include one or more processors 601 and memory 602. Memory 602 may store one or more application programs or data. Memory 602 may be temporary or persistent storage. The application programs stored in memory 602 may include one or more modules (not shown), each module may include a series of computer-executable instructions for the electronic device. Furthermore, processor 601 may be configured to communicate with memory 602 and execute the series of computer-executable instructions in memory 602 on the electronic device. The electronic device may also include one or more power supplies 603, one or more wired or wireless network interfaces 604, one or more input / output interfaces 605, and one or more keyboards 606.

[0149] Specifically, in this embodiment, the electronic device includes a memory and one or more programs, wherein one or more programs are stored in the memory, and one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for use in the electronic device, and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following:

[0150] The system obtains the target facial features to be queried and the target document information corresponding to the target facial features. It performs privacy protection processing on the target facial features based on the obfuscated circuit, and then performs homomorphic encryption processing on the privacy-protected target facial features to obtain the homomorphically encrypted target facial features.

[0151] A query request is generated based on the homomorphically encrypted target face features and sent to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain the corresponding matching results.

[0152] The system receives the matching results sent by the server and the user identification information encrypted with Shamir secret sharing algorithm and homomorphic encryption corresponding to each face feature. Based on the matching results and the user identification information encrypted with Shamir secret sharing algorithm and homomorphic encryption corresponding to each face feature, the system determines the first face feature that matches the homomorphically encrypted target face feature and the user identification information corresponding to the first face feature.

[0153] Determine whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

[0154] In another embodiment, the electronic device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for use in the electronic device, and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following:

[0155] Privacy protection processing is performed on multiple pre-stored facial features, and the user identification information corresponding to each facial feature is homomorphically encrypted and then encrypted again based on the Shamir secret sharing algorithm to obtain the user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted.

[0156] The system obtains the query request sent by the terminal and matches the homomorphically encrypted target face feature with each of the pre-stored privacy-protected face features according to the query request to obtain the corresponding matching result.

[0157] The matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine the first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, based on the matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature. The terminal also determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0158] This specification provides one or more embodiments of a storage medium for storing computer-executable instructions, which, when executed by a processor, implement the following process:

[0159] The system obtains the target facial features to be queried and the target document information corresponding to the target facial features. It performs privacy protection processing on the target facial features based on the obfuscated circuit, and then performs homomorphic encryption processing on the privacy-protected target facial features to obtain the homomorphically encrypted target facial features.

[0160] A query request is generated based on the homomorphically encrypted target face features and sent to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain the corresponding matching results.

[0161] The system receives the matching results sent by the server and the user identification information encrypted with Shamir secret sharing algorithm and homomorphic encryption corresponding to each face feature. Based on the matching results and the user identification information encrypted with Shamir secret sharing algorithm and homomorphic encryption corresponding to each face feature, the system determines the first face feature that matches the homomorphically encrypted target face feature and the user identification information corresponding to the first face feature.

[0162] Determine whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

[0163] One or more embodiments of this specification also provide a storage medium for storing computer-executable instructions that, when executed by a processor, implement the following process:

[0164] Privacy protection processing is performed on multiple pre-stored facial features, and the user identification information corresponding to each facial feature is homomorphically encrypted and then encrypted again based on the Shamir secret sharing algorithm to obtain the user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted.

[0165] The system obtains the query request sent by the terminal and matches the homomorphically encrypted target face feature with each of the pre-stored privacy-protected face features according to the query request to obtain the corresponding matching result.

[0166] The matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine the first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, based on the matching results and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature. The terminal also determines whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

[0167] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0168] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0169] Those skilled in the art will understand that one or more embodiments of this specification can be provided as a method, system, or computer program product. Therefore, one or more embodiments of this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, one or more embodiments of this specification may 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.

[0170] This specification describes one or more embodiments of methods, apparatus (systems), and computer program products according to embodiments of this specification with reference to flowchart illustrations and / or block diagrams. 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, produce a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0171] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0172] 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.

[0173] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0174] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0175] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0176] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0177] One or more embodiments of this specification can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. This specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0178] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0179] The above description is merely one or more embodiments of this specification and is not intended to limit this application. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A risk identification method based on privacy computing, applied to a terminal, the method comprising: The target facial features to be queried and the target document information corresponding to the target facial features are obtained. The target facial features are subjected to privacy protection processing based on the obfuscation circuit, and the target facial features after privacy protection processing are subjected to homomorphic encryption processing to obtain the homomorphically encrypted target facial features. A query request is generated based on the homomorphically encrypted target face features, and the query request is sent to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain the corresponding matching results. The matching results are determined based on the similarity between the homomorphically encrypted target face features and each face feature in the pre-stored set of privacy-protected face features, as well as the multiple subkeys obtained by encrypting the corresponding user ID information based on the Shamir secret sharing algorithm for each face feature. The system receives the matching result sent by the server and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature. Based on the matching result and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature, the system determines the first face feature that matches the homomorphically encrypted target face feature and the user identification information corresponding to the first face feature. Determine whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

2. The method according to claim 1, wherein determining the first facial feature matching the homomorphically encrypted target facial feature based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, comprises: Each matching result is homomorphically decrypted to obtain the corresponding decrypted matching result. Based on the decrypted matching result, a first face feature that matches the homomorphically encrypted target face feature is determined; Based on the first facial feature and the Shamir secret sharing algorithm, the user ID information corresponding to each facial feature, which is encrypted and homomorphically encrypted using the Shamir secret sharing algorithm, is decrypted to obtain the homomorphically encrypted user ID information corresponding to the first facial feature.

3. The method according to claim 1, wherein determining whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature includes: The user ID information corresponding to the first face feature is homomorphically encrypted again to obtain the user ID information corresponding to the first face feature after two homomorphic encryptions. The similarity of the target ID information corresponding to the target facial feature and the user ID information after two homomorphic encryptions corresponding to the first facial feature is compared, and the similarity comparison result of the ID information is homomorphically encrypted and then transmitted to the server. The receiving server performs a first decryption of the homomorphically encrypted similarity comparison result, performs a second decryption on the first decryption result, and determines the consistency between the target document information and the homomorphically encrypted user document information based on the second decryption result.

4. The method according to claim 1, wherein the privacy protection processing of the target facial features based on the obfuscation circuit includes: In the obfuscation circuit, the target facial features are masked based on multiple pre-generated masks to obtain the masked target facial features. In the obfuscated circuit, the target face features after masking are encrypted using the AES encryption algorithm based on a pre-generated key to obtain the target face features after privacy protection processing.

5. A risk identification method based on privacy computing, applied to a server, the method comprising: Privacy protection processing is performed on multiple pre-stored facial features, and the user identification information corresponding to each facial feature is homomorphically encrypted and then encrypted based on the Shamir secret sharing algorithm to obtain the user identification information encrypted based on the Shamir secret sharing algorithm and homomorphically encrypted. The system obtains a query request sent by the terminal and matches the homomorphically encrypted target face feature with each of the pre-stored privacy-protected face features according to the query request, to obtain the corresponding matching result. The matching result is determined based on the similarity between the homomorphically encrypted target face feature and each of the pre-stored privacy-protected face features, as well as the multiple subkeys obtained by encrypting the corresponding user ID information based on the Shamir secret sharing algorithm for each face feature. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine, based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, and to determine whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

6. The method according to claim 5, wherein the plurality of facial features and their corresponding user identification information are pre-stored in a server in Hamming space, and the privacy protection processing of the pre-stored plurality of facial features includes: Multiple masks are randomly generated, and each facial feature is masked based on the generated masks to obtain the masked facial features. A key is pre-generated, and based on the pre-generated key, each facial feature after masking is encrypted using the AES encryption algorithm to obtain the privacy-protected facial features.

7. The method according to claim 5, wherein the step of performing homomorphic encryption on the user identification information corresponding to each facial feature and then encrypting it based on the Shamir secret sharing algorithm to obtain the user identification information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted includes: Homomorphic encryption is performed on the user ID information corresponding to each facial feature to obtain the homomorphically encrypted user ID information corresponding to each facial feature. Based on the Shamir secret sharing algorithm, the homomorphically encrypted user ID information corresponding to each facial feature is divided into multiple user ID information subkeys. Based on a preset decryption rule, user ID information encrypted with the Shamir secret sharing algorithm and homomorphically encrypted is obtained. The decryption rule includes: if the number of user ID information subkeys held exceeds a preset subkey number threshold, then the information obtained by encrypting the homomorphically encrypted user ID information corresponding to each facial feature using the Shamir secret sharing algorithm is decrypted to obtain the homomorphically encrypted user ID information corresponding to each facial feature.

8. The method according to claim 5, wherein matching the homomorphically encrypted target face feature with each of the pre-stored privacy-preserving face features according to the query request to obtain the corresponding matching result includes: Based on the preset feature algorithm, the Hamming distance between the homomorphically encrypted target face feature and each pre-stored privacy-processed face feature is calculated respectively. The corresponding matching result is determined based on the Hamming distance.

9. A risk identification device based on privacy computing, comprising: The first privacy processing module obtains the target facial features to be queried and the target document information corresponding to the target facial features, performs privacy protection processing on the target facial features based on the obfuscation circuit, and performs homomorphic encryption processing on the privacy-protected target facial features to obtain homomorphically encrypted target facial features. The query request generation module generates a query request based on the homomorphically encrypted target face features and sends the query request to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain corresponding matching results. The matching results are determined based on the similarity between the homomorphically encrypted target face features and each face feature in the pre-stored set of privacy-protected face features, as well as the multiple subkeys obtained by encrypting the corresponding user ID information based on the Shamir secret sharing algorithm for each face feature. The information determination module receives the matching result sent by the server and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each facial feature. Based on the matching result and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each facial feature, the module determines the first facial feature that matches the homomorphically encrypted target facial feature and the user identification information corresponding to the first facial feature. The judgment module determines whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

10. A risk identification device based on privacy computing, comprising: The second privacy processing module performs privacy protection processing on multiple pre-stored facial features, and performs homomorphic encryption processing on the user ID information corresponding to each facial feature, and then performs encryption processing based on the Shamir secret sharing algorithm to obtain the user ID information encrypted based on the Shamir secret sharing algorithm and homomorphic encryption. The matching module acquires the query request sent by the terminal, and matches the homomorphically encrypted target face feature with each of the pre-stored privacy-protected face features according to the query request, to obtain the corresponding matching result. The matching result is determined based on the similarity between the homomorphically encrypted target face feature and each of the pre-stored privacy-protected face features, and the multiple subkeys obtained by encrypting the corresponding user ID information based on the Shamir secret sharing algorithm for each face feature. The information sending module sends the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature to the terminal. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine, based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, and to determine whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

11. A risk identification system based on privacy computing, comprising: Terminals and servers, among which, The server performs privacy protection processing on multiple pre-stored facial features, and performs homomorphic encryption on the user ID information corresponding to each facial feature, and then performs encryption processing based on the Shamir secret sharing algorithm to obtain user ID information encrypted with the Shamir secret sharing algorithm and homomorphic encryption. The terminal acquires the target facial features to be queried and the target document information corresponding to the target facial features, performs privacy protection processing on the target facial features based on the obfuscation circuit, and performs homomorphic encryption processing on the privacy-protected target facial features to obtain homomorphically encrypted target facial features, and generates a query request based on the homomorphically encrypted target facial features and sends the query request to the server. The server, according to the query request, matches the homomorphically encrypted target facial features with each of the pre-stored privacy-protected facial features to obtain corresponding matching results. It then sends the matching results and the user identification information corresponding to each facial feature, encrypted using the Shamir secret sharing algorithm and homomorphically encrypted, to the terminal. The matching results are determined based on the similarity between the homomorphically encrypted target facial features and each of the pre-stored privacy-protected facial features, and the multiple subkeys obtained by encrypting the corresponding user identification information using the Shamir secret sharing algorithm for each facial feature. The terminal, based on the matching result and the user ID information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, determines the first facial feature that matches the homomorphically encrypted target facial feature and the homomorphically encrypted user ID information corresponding to the first facial feature, and determines whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

12. An electronic device, comprising: processor; as well as A memory configured to store computer-executable instructions, which, when executed, enable the processor to: The target facial features to be queried and the target document information corresponding to the target facial features are obtained. The target facial features are subjected to privacy protection processing based on the obfuscation circuit, and the target facial features after privacy protection processing are subjected to homomorphic encryption processing to obtain the homomorphically encrypted target facial features. A query request is generated based on the homomorphically encrypted target face features, and the query request is sent to the server. The query request is used to trigger the server to match the homomorphically encrypted target face features with each face feature in a pre-stored set of privacy-protected face features to obtain the corresponding matching results. The matching results are determined based on the similarity between the homomorphically encrypted target face features and each face feature in the pre-stored set of privacy-protected face features, as well as the multiple subkeys obtained by encrypting the corresponding user ID information based on the Shamir secret sharing algorithm for each face feature. The system receives the matching result sent by the server and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature. Based on the matching result and the user identification information encrypted and homomorphically encrypted based on the Shamir secret sharing algorithm corresponding to each face feature, the system determines the first face feature that matches the homomorphically encrypted target face feature and the user identification information corresponding to the first face feature. Determine whether the target ID information corresponding to the target facial feature is consistent with the homomorphically encrypted user ID information corresponding to the first facial feature.

13. An electronic device, comprising: processor; as well as A memory configured to store computer-executable instructions, which, when executed, enable the processor to: Privacy protection processing is performed on multiple pre-stored facial features, and the user identification information corresponding to each facial feature is homomorphically encrypted and then encrypted based on the Shamir secret sharing algorithm to obtain the user identification information encrypted based on the Shamir secret sharing algorithm and homomorphically encrypted. The system obtains a query request sent by the terminal and matches the homomorphically encrypted target face feature with each of the pre-stored privacy-protected face features according to the query request, to obtain the corresponding matching result. The matching result is determined based on the similarity between the homomorphically encrypted target face feature and each of the pre-stored privacy-protected face features, as well as the multiple subkeys obtained by encrypting the corresponding user ID information based on the Shamir secret sharing algorithm for each face feature. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are sent to the terminal. The matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature are used to trigger the terminal to determine, based on the matching result and the user identification information encrypted using the Shamir secret sharing algorithm and homomorphically encrypted for each facial feature, a first facial feature that matches the homomorphically encrypted target facial feature, and the homomorphically encrypted user identification information corresponding to the first facial feature, and to determine whether the target identification information corresponding to the target facial feature is consistent with the homomorphically encrypted user identification information corresponding to the first facial feature.

Images