CN114978623A - Privacy protection-based face comparison method and device - Google Patents
Privacy protection-based face comparison method and device Download PDFInfo
- Publication number
- CN114978623A CN114978623A CN202210496540.8A CN202210496540A CN114978623A CN 114978623 A CN114978623 A CN 114978623A CN 202210496540 A CN202210496540 A CN 202210496540A CN 114978623 A CN114978623 A CN 114978623A
- Authority
- CN
- China
- Prior art keywords
- matrix
- vector
- preset
- feature vector
- face
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 86
- 239000013598 vector Substances 0.000 claims abstract description 266
- 239000011159 matrix material Substances 0.000 claims abstract description 175
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 78
- 238000012546 transfer Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 description 33
- 230000000875 corresponding effect Effects 0.000 description 12
- 238000000605 extraction Methods 0.000 description 10
- 239000000284 extract Substances 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/08—Network architectures or network communication protocols for network security for authentication of entities
- H04L63/0861—Network architectures or network communication protocols for network security for authentication of entities using biometrical features, e.g. fingerprint, retina-scan
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/04—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks
- H04L63/0428—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks wherein the data content is protected, e.g. by encrypting or encapsulating the payload
- H04L63/0442—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks wherein the data content is protected, e.g. by encrypting or encapsulating the payload wherein the sending and receiving network entities apply asymmetric encryption, i.e. different keys for encryption and decryption
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
- H04L9/008—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols involving homomorphic encryption
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Security & Cryptography (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Mathematical Physics (AREA)
- Computer Hardware Design (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Optimization (AREA)
- General Health & Medical Sciences (AREA)
- Algebra (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Collating Specific Patterns (AREA)
- Health & Medical Sciences (AREA)
Abstract
The embodiment of the specification provides a face comparison method and a face comparison device based on privacy protection, and the method comprises the following steps: extracting a current face feature vector from a currently acquired first face image; encrypting the current face characteristic vector by using a preset encryption algorithm to obtain a current ciphertext vector, wherein the preset encryption algorithm relates to an encryption matrix and a preset large integer; performing inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector by using an intermediate matrix to obtain an intermediate result, wherein the registration ciphertext vector is obtained by encrypting the registration face feature vector obtained in the registration stage by using a preset encryption algorithm, and the intermediate matrix is determined based on an inverse matrix of the encryption matrix; and determining a plaintext comparison result based on the intermediate result and a preset large integer, wherein the plaintext comparison result represents the similarity between the current face feature vector and the registered face feature vector.
Description
Technical Field
The present disclosure relates to the technical field of privacy protection, and in particular, to a method and an apparatus for comparing faces based on privacy protection.
Background
In recent years, artificial intelligence technology represented by face comparison and recognition is rapidly developed, and face comparison and recognition is widely applied to a plurality of scenes, such as face payment, face verification, face login and the like.
However, with the importance of security of private data, the supervision of face data is more stringent at present. For example: at present, a supervision party requires that a server side cannot store the face image and only stores the face features. However, the face features may be reversely derived under certain conditions, and accordingly, the risk of personal privacy leakage still exists when the face features are stored in the server. Therefore, how to provide a face comparison method based on privacy (e.g. face data) protection becomes an urgent problem to be solved.
Disclosure of Invention
One or more embodiments of the present disclosure provide a face comparison method and apparatus based on privacy protection, so as to implement face comparison and recognition under the premise of protecting privacy data (e.g., face data).
According to a first aspect, a face comparison method based on privacy protection is provided, and is applied to a terminal device, and includes:
extracting a current face feature vector from a currently acquired first face image;
encrypting the current face characteristic vector by using a preset encryption algorithm to obtain a current ciphertext vector, wherein the preset encryption algorithm relates to an encryption matrix and a preset large integer;
performing inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector by using an intermediate matrix to obtain an intermediate result, wherein the registration ciphertext vector is obtained by encrypting the registration face feature vector obtained in the registration stage by using the preset encryption algorithm, and the intermediate matrix is determined based on an inverse matrix of the encryption matrix;
and determining a plain comparison result based on the intermediate result and the preset large integer, wherein the plain comparison result represents the similarity of the current face feature vector and the registered face feature vector.
In an optional implementation, the obtaining the current ciphertext vector includes:
encrypting the current face feature vector by using the encryption matrix and a preset large integer to obtain an initial face ciphertext;
generating a random error vector;
and adding the random error vector to the initial face ciphertext to obtain the current ciphertext vector.
In an alternative embodiment, the method further comprises:
in the registration stage, acquiring a registered face image, and extracting a registered face feature vector from the registered face image;
and encrypting the registered face feature vector based on the preset encryption algorithm to obtain the registered ciphertext vector.
In an alternative embodiment, the method further comprises:
determining the encryption matrix in the preset encryption algorithm;
determining the intermediate matrix using an inverse of the encryption matrix.
In an alternative embodiment, the determining the intermediate matrix includes:
determining a product of a transpose of an inverse of the encryption matrix and an inverse of the encryption matrix as the intermediate matrix.
In an optional embodiment, the preset encryption algorithm is a vector homomorphic encryption algorithm; the determining the encryption matrix in the preset encryption algorithm includes:
acquiring a public key matrix and a corresponding private key matrix in a vector homomorphic encryption algorithm;
determining the public key matrix as the encryption matrix; discarding the private key matrix.
In an alternative embodiment, the determining the plaintext alignment result comprises:
determining a first product representing a multiplication result of the current face feature vector and the registered face feature vector based on a ratio of the intermediate result to a specified power of the preset large integer, wherein the specified power is related to the preset encryption algorithm;
and determining the plaintext comparison result by using a preset distance formula and the first product.
In an alternative embodiment, the method further comprises:
and comparing the plaintext comparison result with a preset comparison threshold, and executing specified business operation based on the obtained comparison result.
In an alternative embodiment, the specified business operation comprises one of the following operations: starting up operation, transaction execution operation, transfer operation and login operation.
According to a second aspect, a face comparison apparatus based on privacy protection is provided, and is deployed in a terminal device, and includes:
the first acquisition module is configured to extract a current face feature vector from a currently acquired first face image;
the first encryption module is configured to encrypt the current face feature vector by using a preset encryption algorithm to obtain a current ciphertext vector, wherein the preset encryption algorithm relates to an encryption matrix and a preset large integer;
the operation module is configured to perform inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector by using an intermediate matrix to obtain an intermediate result, wherein the registration ciphertext vector is obtained by encrypting the registration face feature vector obtained in the registration stage by using the preset encryption algorithm, and the intermediate matrix is determined based on an inverse matrix of the encryption matrix;
a first determining module configured to determine a plaintext comparison result, which represents a similarity between the current face feature vector and the registered face feature vector, based on the intermediate result and the preset large integer.
In an optional implementation manner, the first encryption module is specifically configured to encrypt the current face feature vector by using the encryption matrix and a preset large integer to obtain an initial face ciphertext;
generating a random error vector;
and adding the random error vector to the initial face ciphertext to obtain the current ciphertext vector.
In an alternative embodiment, the method further comprises:
the second acquisition module is configured to acquire a registered face image in a registration stage and extract a registered face feature vector from the registered face image;
and the second encryption module is configured to encrypt the registered face feature vector based on the preset encryption algorithm to obtain the registered ciphertext vector.
In an alternative embodiment, the method further comprises:
a second determining module configured to determine the encryption matrix in the preset encryption algorithm;
a third determination module configured to determine the intermediate matrix using an inverse of the encryption matrix.
In an alternative embodiment, the third determining module is specifically configured to determine a product of a transpose of the inverse of the encryption matrix and the inverse of the encryption matrix as the intermediate matrix.
In an optional embodiment, the preset encryption algorithm is a vector homomorphic encryption algorithm; the second determining module is specifically configured to obtain a public key matrix and a corresponding private key matrix in a vector homomorphic encryption algorithm;
determining the public key matrix as the encryption matrix; discarding the private key matrix.
In an optional implementation manner, the first determining module is specifically configured to determine a first product representing a multiplication result of the current face feature vector and the registered face feature vector based on a ratio of the intermediate result to a specified power of the preset large integer, where the specified power is related to the preset encryption algorithm;
and determining the plaintext comparison result by using a preset distance formula and the first product.
In an alternative embodiment, the method further comprises:
and comparing the plaintext comparison result with a preset comparison threshold, and executing specified business operation based on the obtained comparison result.
In an alternative embodiment, the specified business operation comprises one of the following operations: starting up operation, transaction execution operation, transfer operation and login operation.
According to a third aspect, there is provided a computer readable storage medium having stored thereon a computer program which, when executed in a computer, causes the computer to perform the method of the first aspect.
According to a fourth aspect, there is provided a computing device comprising a memory and a processor, wherein the memory has stored therein executable code, and the processor, when executing the executable code, implements the method of the first aspect.
According to the method and the device provided by the embodiment of the specification, in the face comparison stage, the current face feature vector is encrypted by using a preset encryption algorithm related to an encryption matrix and a preset large integer, namely, at least covering the current face characteristic vector through an encryption matrix and a preset large integer to obtain a current ciphertext vector, then utilizing an intermediate matrix determined based on an inverse matrix of the encryption matrix, performing inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector (obtained by encrypting the registration face feature vector obtained in the registration stage by using a preset encryption algorithm), the intermediate matrix can assist in eliminating the influence of the encryption matrix on the ciphertext vector in the inner product operation process to obtain an intermediate result (not influenced by the encryption matrix), and then a plaintext comparison result capable of representing the similarity of the current face feature vector and the registered face feature vector is determined based on the intermediate result and a preset large integer. In the process, the intermediate matrix is arranged, so that a decryption step is not needed in the process of comparing the face in the ciphertext domain, and a plaintext comparison result which is equal to a plaintext face feature vector comparison result (namely, the similarity of the current face feature vector and the registered face feature vector is represented) can be directly obtained, so that the efficient comparison of the face is realized, namely, the terminal device independently realizes the efficient and rapid comparison and identification of the face on the premise of protecting privacy data (namely, the face features).
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.
FIG. 1 is a schematic diagram of a framework for implementing one embodiment disclosed herein;
fig. 2 is a schematic flowchart of a face comparison method based on privacy protection according to an embodiment;
fig. 3 is a schematic flowchart of a face comparison method based on privacy protection according to an embodiment;
fig. 4 is a schematic block diagram of a face comparison apparatus based on privacy protection according to an embodiment.
Detailed Description
The technical solutions of the embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.
The embodiment of the specification discloses a face comparison method and a face comparison device based on privacy protection, and firstly introduces an application scene and a technical concept of the method, specifically as follows:
as described above, with importance placed on security of private data, face data is currently monitored more strictly. Because the human face features can be reversely pushed out under certain conditions, the human face features are stored at the server side, and the risk of personal privacy leakage still exists. Therefore, how to provide a face comparison method based on privacy (e.g. face data) protection becomes an urgent problem to be solved.
In view of the above, the inventor proposes a face comparison method based on privacy protection, which is applied to a terminal device, according to the method, in order to protect face data, after a terminal device obtains a registered face image of a user, a registered face feature vector is extracted from the registered face image, but the registered face feature vector is not uploaded, but is encrypted and stored locally. And then the terminal equipment obtains a face image to be compared, extracts a face feature vector to be compared from the face image to be compared, encrypts the face feature vector to be compared, and directly compares the encrypted registered face feature vector with the encrypted face feature vector to be compared locally by the terminal equipment so as to complete a face comparison process in a ciphertext domain to obtain a comparison result. That is to say, in order to protect user's face data, this people's face comparison process is accomplished by terminal equipment alone, need not upload people's face data to the server, confirms people's face comparison result through the server to, carry out the comparison of people's face in the ciphertext territory, avoid revealing of people's face data, realize the better protection to people's face data.
For clarity of layout, a face comparison process of the plaintext domain is first described below, where the process generally includes obtaining two face feature vectors to be compared (e.g., the aforementioned registered face feature vector and the face feature vector to be compared), and then calculating a distance between the two face feature vectors, where the distance may represent a similarity between the two face feature vectors. In the case of the euclidean distance, the distance is inversely related to the similarity. Under the condition of adopting the cosine distance, the cosine distance is positively correlated with the similarity, and the cosine distance is the comparison result. The following description is made in conjunction with the cosine distance. Subsequently, the obtained comparison result, namely the cosine distance, can be compared with a preset threshold value to obtain a comparison result, wherein if the comparison result represents that the obtained cosine distance is not less than the threshold value, the result indicates that the two face feature vectors correspond to the same face, and if the comparison result represents that the obtained cosine distance is less than the threshold value, the result indicates that the two face feature vectors correspond to different faces. Any of the execution tasks may then be performed based on the comparison. It is understood that the face comparison can be used in any authentication scenario, such as a boot scenario, a transfer scenario, a payment scenario, a commute scenario, an information (e.g., health code, account information, etc.) query scenario, and the like.
It should be noted that the cosine distance can be expressed by the following formula (1):
wherein, X and Y respectively represent two face feature vectors to be compared.
As can be seen from the above formula, the core of the aforementioned process of calculating the cosine distance between two face feature vectors (i.e., the process of comparing two face feature vectors) is that the process of calculating the inner product of two face feature vectors, i.e., the process of comparing two face feature vectors, can be converted into the process of calculating the inner product of two face feature vectors.
In view of this, the process of face comparison in the ciphertext domain is essentially to calculate the inner product of two ciphertext vectors. In view of the security of the local face data of the terminal device, it is desirable to find a solution, so that the inner product of two ciphertext vectors is calculated, i.e., the two vectors are compared in a ciphertext domain, and the obtained result does not need to be decrypted, so that a calculation result equal to the inner product of a plaintext vector can be directly obtained, and privacy protection of the registered face data stored locally in the terminal device is realized.
In view of this, fig. 1 shows a schematic diagram of an implementation framework according to one embodiment disclosed herein. As shown in fig. 1, the implementation process of the method is divided into two stages, namely a registration stage and a face comparison stage (which may also be referred to as an authentication stage). In the registration stage, the terminal device collects a face image of the user a through an image collecting device arranged in the terminal device, and the face image is called a registered face image 1. The terminal device extracts a face feature vector, namely a registered face feature vector a, from the registered face image 1 through a feature extraction module of the terminal device. In order to ensure the safety of the face feature vector a registered in the terminal equipment, the terminal equipment determines a preset encryption algorithm ENC through a parameter determination module of the terminal equipment, namely, an encryption matrix M and a preset large integer w related to the preset encryption algorithm ENC, and determines an intermediate matrix H based on an inverse matrix B of the encryption matrix M, wherein the intermediate matrix H is used for assisting in eliminating the influence of the encryption matrix in a ciphertext vector in the inner product operation process of the ciphertext vector. Then, the terminal equipment encrypts the registered face characteristic vector a by using the encryption matrix M and a preset large integer w through an encryption module of the terminal equipment to obtain a registered ciphertext vector [ a ]; the terminal device stores the registration ciphertext vector [ a ], the preset encryption algorithm (the encryption matrix M and the preset large integer w), and the intermediate matrix H in a local preset storage space. And then, the terminal equipment enters a face comparison stage after acquiring the face image to be compared.
In the face comparison stage, the terminal device collects a face image through the image collection device thereof to serve as a face image to be compared, and extracts a face feature vector, namely a current face feature vector b, from the face image to be compared through the feature extraction module thereof. And then, the terminal equipment reads the encryption matrix M and the preset large integer w related to the preset encryption algorithm from the preset storage space through the encryption module of the terminal equipment, and encrypts the current face characteristic vector b by using the encryption matrix M and the preset large integer w to obtain a current ciphertext vector [ b ]. Then, the terminal device reads the intermediate matrix H and the registered ciphertext vector [ a ] from a preset storage space through an operation module of the terminal device, and performs inner product operation on the current ciphertext vector [ b ] and the registered ciphertext vector [ a ] by using the intermediate matrix H to obtain an intermediate result U, wherein in the inner product operation process, the intermediate matrix H assists in eliminating the influence of an encryption matrix M in the two ciphertext vectors; then, the terminal device reads the preset large integer w from the preset storage space through the determining module, and determines a plaintext comparison result Z based on the intermediate result U and the preset large integer w, that is, the intermediate result U is processed by using the preset large integer w (that is, the influence of the preset large integer w in the intermediate result U is eliminated), and the plaintext comparison result Z can represent the similarity between the current face feature vector b and the registered face feature vector a, that is, the similarity is equal to the similarity between two plaintext face feature vectors.
In the process, the intermediate matrix is arranged, so that a plain text comparison result which is equal to a plain text face feature vector comparison result (namely, the similarity between the current face feature vector and the registered face feature vector is represented) can be directly obtained without a decryption step in the face comparison process in a ciphertext domain, the offline efficient comparison of the face is realized, and the comparison and the identification of the face are efficiently and quickly completed on the premise of protecting privacy data (namely, the face features) by the terminal device alone.
The following describes in detail a face comparison method and apparatus based on privacy protection provided in this specification, with reference to specific embodiments.
Fig. 2 is a flowchart illustrating a face comparison method based on privacy protection in an embodiment of the present disclosure. The method is applied to a terminal device, and the terminal device can be implemented by any device, equipment, platform, equipment cluster and the like with computing and processing capabilities. The terminal equipment can be intelligent mobile phones, computers, all-in-one machines, IOT equipment and other equipment. In the face comparison stage, as shown in fig. 2, the method includes the following steps S210-S240:
in step S210, a current face feature vector b is extracted from the currently acquired first face image. In this step, the terminal device may acquire a face image through the set image acquisition device, which is called a first face image, and then extract a current face feature vector b from the first face image by using a preset feature extraction algorithm, where the preset feature extraction algorithm may be any algorithm that can realize face feature extraction, such as a pre-trained feature extraction algorithm based on a neural network or a conventional feature extraction algorithm, where the conventional feature extraction algorithm may be, for example, a feature extraction algorithm based on Scale Invariant Feature Transform (SIFT), a feature extraction algorithm based on a Histogram of Oriented Gradients (HOG), and so on.
After the terminal device obtains the current face feature vector b, in step S220, the current face feature vector b is encrypted by using a preset encryption algorithm ENC to obtain a current ciphertext vector [ b ]. The preset encryption algorithm ENC relates to an encryption matrix M and a preset large integer w.
It will be appreciated that there is also a registration phase before the face comparison phase. The registration stage may refer to that a registered user activates a face comparison recognition function, and in the registration stage, the terminal device may obtain a registered face image and may initialize information such as an encryption matrix and an intermediate matrix for the registered user. In an embodiment, after the terminal device detects an opening instruction of the face comparison recognition function, on one hand, the terminal device collects a face image for a registered user through an image collection device arranged in the terminal device, that is, obtains a registered face image, and extracts a registered face feature vector a from the registered face image. On the other hand, the terminal device initializes the encryption matrix and the intermediate matrix and other information for the registered user, and correspondingly, the terminal device determines the encryption matrix M in the preset encryption algorithm ENC and determines the preset large integer w.
Then, in the registration phase, the terminal device determines an intermediate matrix H by using the inverse matrix B of the encryption matrix M. Considering that in the face comparison stage, the current ciphertext vector [ b ] needs to be calculated]And register the ciphertext vector [ a]The process of determining the intermediate matrix, in one embodiment, may be: the product of the transpose of the inverse B of the encryption matrix M and the inverse B of the encryption matrix is determined as the intermediate matrix H. Wherein the intermediate matrix may be represented as H ═ B T B。
In the registration stage, after the terminal device obtains a registered face feature vector a, an encryption matrix M and a preset large integer w, the registered face feature vector a is encrypted based on a preset encryption algorithm ENC, namely the registered face feature vector a is encrypted by using the encryption matrix M and the preset large integer w related to the preset encryption algorithm ENC to obtain a registered ciphertext vector [ a ]. And then, the terminal equipment stores the encryption matrix M, the preset large integer w, the middle matrix H and the registration ciphertext vector [ a ] in a preset storage space.
In one implementation, the terminal device may be provided with a trusted execution environment TEE, and for better protecting privacy of the face data, the preset storage space may be a storage space provided in the TEE, and the face comparison stage may be performed in the TEE.
In another implementation manner, the preset storage space may be a storage space in a common execution environment REE of the terminal device, and the face comparison stage may be executed in the REE of the terminal device. In one embodiment, in order to better realize privacy protection of the registered face feature vector a, in the registration stage, the terminal device may further generate a random errorVector e 1 Accordingly, a registration ciphertext vector [ a ] is obtained]The process of (1) may specifically be that, the registered face feature vector a is encrypted by using an encryption matrix M and a preset large integer w related to a preset encryption algorithm ENC to obtain an initial registration ciphertext a, and a random error vector e is used 1 Adding the initial registration ciphertext a to obtain a registration ciphertext vector [ a]。
Then, in the face comparison stage, after the terminal device obtains the current face feature vector b, in step S220, the encryption matrix M and the preset large integer w related to the preset encryption algorithm ENC may be directly obtained from the preset storage space, and the current face feature vector b is encrypted by using the encryption matrix M and the preset large integer w to obtain the current ciphertext vector [ b ].
In one embodiment, in order to better implement encryption of the current face feature vector b and better ensure that the current face feature vector b is not acquired by an attacker, the step S220 may be specifically configured to encrypt the current face feature vector b by using an encryption matrix M and a preset large integer w to obtain an initial face ciphertext b'; generating a random error vector e 2 (ii) a Will random error vector e 2 Adding the current face ciphertext b' to obtain a current ciphertext vector [ b]。
Then, after obtaining the current ciphertext vector [ b ], the terminal device may obtain the registration ciphertext vector [ a ] stored in the registration stage from the preset storage space, and then, in step S230, perform an inner product operation on the current ciphertext vector [ b ] and the pre-stored registration ciphertext vector [ a ] by using the intermediate matrix H to obtain an intermediate result U. The registration ciphertext vector [ a ] is obtained by encrypting the registration face feature vector obtained in the registration stage by using a preset encryption algorithm, and the intermediate matrix H is determined based on the inverse matrix B of the encryption matrix M.
It can be understood that, the way of encrypting the current face feature vector (and the registered face feature vector) by using the encryption matrix M and the preset large integer w involved in the preset encryption algorithm ENC is various. In one implementation, the current face feature vector b is encrypted by using the encryption matrix M and the preset large integer w, where the encryption matrix M is pre-multiplied by the product of the preset large integer w and the current face feature vector b, and the result is used as a current ciphertext vector [ b ], and specifically, the current ciphertext vector [ b ] may be represented as [ b ] ═ M (wb). Accordingly, the registration ciphertext vector [ a ] may be denoted as [ a ] ═ m (wa).
Accordingly, the inner product operation process, i.e. the calculation process of the intermediate result U, can be expressed by the following formula (2):
U＝[a] T H[b]＝(M(wa)) T B T B M(wb)＝(B M(wa)) T (B M(wb))＝(wa) T (wb)＝w 2 a T b；(2)
in another implementation mode, after the current face feature vector b is encrypted by using the encryption matrix M and the preset large integer w, a random error vector e is added to the obtained result 2 To obtain the current ciphertext vector [ b]Correspondingly, the current ciphertext vector [ b ]]Can be expressed as [ b ]]＝M(wb)+e 2 . Accordingly, the registration ciphertext vector may be represented as [ a ]]＝M(wa)+e 1 。
Accordingly, the inner product operation process, i.e. the calculation process of the intermediate result U, can be expressed by the following formula (3):
U＝[a] T H[b]＝(M(wa)+e 1 ) T B T B(M(wb)+e 2 )＝(B M(wa)+Be 1 ) T (BM(wb)+Be 2 )＝(wa+Be 1 ) T (wb+Be 2 )＝w 2 a T b+error
(3)
in yet another implementation, the current face feature vector b is encrypted by using the encryption matrix M and the preset large integer w, and a product of the preset large integer w and the current face feature vector b is divided by the encryption matrix M, and a result is used as a current ciphertext vector [ b ], specifically, the current ciphertext vector [ b ] may be represented by a formula, [ b ] ═ wb)/M, and accordingly, the registration ciphertext vector may be represented by [ a ] ═ wa)/M.
Accordingly, the inner product operation process, i.e. the calculation process of the intermediate result U, can be expressed by the following formula (4):
in another implementation, after the current face feature vector b is encrypted by using the encryption matrix M and the preset large integer w, a random error vector e is added to the result of the current face feature vector b 2 To obtain the current ciphertext vector [ b]Correspondingly, the current ciphertext vector [ b ]]Can be expressed by the following formula [ b]＝(wb)/M+e 2 And the registration ciphertext vector may be represented as [ a ]]＝(wa)/M+e 1 。
Accordingly, the inner product operation process, i.e. the calculation process of the intermediate result U, can be expressed by the following formula (5):
it should be noted that the terminal device may not only obtain the current ciphertext vector [ b ] through the foregoing exemplary manner, and then obtain the intermediate result U, but also obtain the current ciphertext vector [ b ] through other manners by using the encryption matrix M and the preset large integer w, for example, encrypt the current face feature vector b by using the encryption matrix M and the preset power of the preset large integer w to obtain the current ciphertext vector [ b ], then perform an inner product operation by using a corresponding manner, and obtain the intermediate result U, and so on, which is not described in this specification by way of one example.
Then, after the terminal device obtains the intermediate result U, in step S240, a plaintext comparison result Z is determined based on the intermediate result U and the preset large integer w, and represents the similarity between the current face feature vector b and the registered face feature vector a.
It can be understood that, through the intermediate matrix H, the influence of the encryption matrix M in the inner product of the two ciphertext vectors (the registration ciphertext vector and the current ciphertext vector) can be eliminated in the inner product operation process, and an intermediate result U which is not influenced by the encryption matrix M is obtained. And then, processing the intermediate result U by using the preset large integer w, namely eliminating the influence of the preset large integer w in the intermediate result by using the preset large integer w to obtain a plaintext comparison result Z which is equal to the inner product of the current face feature vector b and the registered face feature vector a, wherein the plaintext comparison result Z can represent the similarity of the current face feature vector b and the registered face feature vector a.
In one embodiment, in step S240, it may be specifically set that, based on a ratio of the intermediate result U to a specified power of the preset large integer w, a first product characterizing a multiplication result of the current face feature vector b and the registered face feature vector a is determined, where the specified power is related to a preset encryption algorithm; the plaintext comparison result Z is determined by using a predetermined distance formula (e.g., the above cosine distance formula) and the first product.
In one implementation, if the face feature vector (current face feature vector or registered face feature vector) is encrypted by using a predetermined encryption algorithm, a product of a predetermined large integer w and the face feature vector is first calculated, and then the product is encrypted by using an encryption matrix M, for example, the encryption matrix M is multiplied by the product, or the product is divided by the encryption matrix M. Accordingly, the intermediate result U may be represented as w 2 a T b, correspondingly, in step S240, a ratio of the intermediate result U to a given power of the preset large integer w may be calculated, wherein the given power may be a power of 2, and then the first product a may be obtained T And b, representing the multiplication result of the current face feature vector b and the registered face feature vector a.
In another implementation, if the face feature vector (the current face feature vector or the registered face feature vector) is encrypted by using a preset encryption algorithm, the product of the preset power of the preset large integer w and the face feature vector is first calculated, and then the product is encrypted by using the encryption matrix M, and accordingly, the specified power can be 2 times of the preset power, and then, in step S240, the ratio of the intermediate result U to the specified power of the preset large integer w is calculated to obtain the first product a T b。
In another implementation, if the face feature vector (current face feature vector or registered face feature vector) is encrypted by using a predetermined encryption algorithm, the predetermined power of the predetermined large integer w and the face feature are first calculatedThe vector multiplication is carried out, then the product is encrypted by using an encryption matrix M to obtain an initial ciphertext (an initial face ciphertext or an initial registration ciphertext), then a random error vector is generated, the random error vector is added to the initial ciphertext to obtain a final ciphertext vector (a current ciphertext vector or a registration ciphertext vector), and correspondingly, an intermediate result U can be represented as w 2 a T b + error, then in step S240, the terminal device may calculate a ratio of the intermediate result U to a given power of the preset large integer w, where the given power may be a power of 2, and then a first product may be obtained
It will be understood that w is a predetermined large integer, w 2 The value of (A) will be very large, corresponding to
Will be very small compared to a T b，
Can be ignored, thus, the first product
The multiplication result of the current face feature vector b and the registered face feature vector a can be represented.
After the terminal equipment determines the first product, a plaintext comparison result Z is determined by using a preset distance formula and the first product, the plaintext comparison result Z is equal to an inner product result of the current face feature vector and the registered face feature vector, and the similarity of the current face feature vector and the registered face feature vector can be represented.
In this embodiment, through the intermediate matrix, a plaintext comparison result equivalent to a plaintext face feature vector comparison result (i.e., representing the similarity between the current face feature vector and the registered face feature vector) can be directly obtained without a decryption step in the face comparison process in the ciphertext domain, which realizes efficient comparison of the face, i.e., the terminal device alone realizes efficient and rapid face comparison and identification under the premise of protecting privacy data (i.e., face features).
In an embodiment, on the basis of the flow shown in fig. 2, as shown in fig. 3, the method may further include: in step S250, the plaintext comparison result is compared with a preset comparison threshold, and based on the obtained comparison result, a designated service operation is executed. In this step, the terminal device may pre-store a preset comparison threshold, and after determining the plaintext comparison result Z, compare the plaintext comparison result with the preset comparison threshold, and determine the size of the plaintext comparison result and the preset comparison threshold, so as to obtain a comparison result. If the comparison result of the representation plaintext of the comparison result is not less than the preset comparison threshold, the feature vector of the face before representation and the feature vector of the registered face correspond to the same face, and the comparison of the representation face is successful; if the comparison result represents that the plaintext comparison result is smaller than the preset comparison threshold, the face feature vector before the representation and the registered face feature vector correspond to different faces, and the representation face comparison fails. Subsequently, the terminal device may perform a specified service operation based on the comparison result.
It can be understood that the specified business operation is related to a current scene for performing face comparison, where the current scene may include, but is not limited to, a login scene, a transfer scene, a payment scene, a transaction scene, a startup scene, an information query scene, and the like. Correspondingly, for example, if the current scene is a login scene, if the comparison result indicates that the plaintext comparison result is not less than the preset comparison threshold, the face comparison is successful, the designated service operation is a power-on operation, and if the comparison result indicates that the plaintext comparison result is less than the preset comparison threshold, the face comparison is successful, the designated service operation may be an operation of outputting and prompting the face authentication failure. If the comparison result represents that the plaintext comparison result is smaller than the preset comparison threshold, the designated service operation can be an operation for outputting and prompting face authentication failure, transfer failure and the like. Accordingly, in one embodiment, the specified business operation may include, but is not limited to, one of the following: starting up operation, transaction execution operation, transfer operation and login operation.
In one embodiment, the predetermined encryption algorithm may be a vector homomorphic encryption algorithm; the process of determining an encryption matrix in a preset encryption algorithm may specifically include: acquiring a public key matrix and a corresponding private key matrix in a vector homomorphic encryption algorithm; determining the public key matrix as an encryption matrix; the private key matrix is discarded.
In one implementation, the terminal device may generate a preset large integer w, an identity matrix I, an error matrix E, two random matrices a and T, and a pair of reversible matrices Ps and Pm, Ps × Pm ═ I, based on a vector homomorphic encryption algorithm.
Then, the terminal device generates a public key matrix M1 and a corresponding private key matrix S based on a public-private key matrix generation manner of a vector homomorphic encryption algorithm, where the private key matrix is generated based on a random matrix T, an identity matrix I, and a matrix Ps, and may be specifically represented as: s ═ I, T]P s The public key matrix is based on the random matrices a and T, the identity matrix I, and the matrix Pm, and may be specifically expressed as:
therefore, the terminal equipment obtains a public key matrix M1 and a corresponding private key matrix S in the vector homomorphic encryption algorithm, and then the terminal equipment determines the public key matrix M1 as an encryption matrix M; and considering the security of the locally stored registration ciphertext vector [ a ], the possibility of decryption by an attacker is avoided, and considering that the face comparison process provided by the specification does not need to utilize the private key matrix S for decryption, the terminal device discards the private key matrix S, that is, in the face comparison stage, the terminal device does not locally store the private key matrix S.
The foregoing describes certain embodiments of the present specification, and other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily have to be in the particular order shown or in sequential order to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
Corresponding to the foregoing method embodiment, an embodiment of the present specification provides a face comparison apparatus 400 based on privacy protection, where the apparatus is deployed in a terminal device, and a schematic block diagram of the apparatus is shown in fig. 4, and the apparatus includes:
a first obtaining module 410 configured to extract a current face feature vector from a currently obtained first face image;
a first encryption module 420, configured to encrypt the current face feature vector by using a preset encryption algorithm to obtain a current ciphertext vector, where the preset encryption algorithm relates to an encryption matrix and a preset large integer;
an operation module 430, configured to perform an inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector to obtain an intermediate result, where the registration ciphertext vector is obtained by encrypting the registration face feature vector obtained in the registration stage by using the preset encryption algorithm, and the intermediate matrix is determined based on an inverse matrix of the encryption matrix;
a first determining module 440 configured to determine a plaintext comparison result representing a similarity between the current face feature vector and the registered face feature vector based on the intermediate result and the preset large integer.
In an optional implementation manner, the first encryption module 410 is specifically configured to encrypt the current face feature vector by using the encryption matrix and a preset large integer to obtain an initial face ciphertext;
generating a random error vector;
and adding the random error vector to the initial face ciphertext to obtain the current ciphertext vector.
In an alternative embodiment, the method further comprises:
a second obtaining module (not shown in the figure) configured to, in a registration stage, obtain a registered face image, and extract a registered face feature vector from the registered face image;
and a second encryption module (not shown in the figure) configured to encrypt the registered face feature vector based on the preset encryption algorithm to obtain the registered ciphertext vector.
In an alternative embodiment, the method further comprises:
a second determining module (not shown in the figure) configured to determine the encryption matrix in the preset encryption algorithm;
a third determining module (not shown in the figures) configured to determine the intermediate matrix using an inverse of the encryption matrix.
In an alternative embodiment, the third determining module is specifically configured to determine, as the intermediate matrix, a product of a transpose of an inverse of the encryption matrix and an inverse of the encryption matrix.
In an optional embodiment, the preset encryption algorithm is a vector homomorphic encryption algorithm; the second determining module is specifically configured to obtain a public key matrix and a corresponding private key matrix in a vector homomorphic encryption algorithm;
determining the public key matrix as the encryption matrix; discarding the private key matrix.
In an optional implementation manner, the first determining module 440 is specifically configured to determine a first product representing a multiplication result of the current face feature vector and the registered face feature vector based on a ratio of the intermediate result to a specified power of the preset large integer, where the specified power is related to the preset encryption algorithm;
and determining the plaintext comparison result by using a preset distance formula and the first product.
In an alternative embodiment, the method further comprises:
and a comparison executing module (not shown in the figure) configured to compare the plaintext comparison result with a preset comparison threshold, and execute a specified service operation based on the obtained comparison result.
In an alternative embodiment, the specified business operation comprises one of the following operations: starting up operation, transaction execution operation, transfer operation and login operation.
The above device embodiments correspond to the method embodiments, and specific descriptions may refer to descriptions of the method embodiments, which are not repeated herein. The device embodiment is obtained based on the corresponding method embodiment, has the same technical effect as the corresponding method embodiment, and for the specific description, reference may be made to the corresponding method embodiment.
Embodiments of the present specification further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed in a computer, the computer program causes the computer to execute the privacy protection-based face comparison method provided in the present specification.
The embodiment of the present specification further provides a computing device, which includes a memory and a processor, where the memory stores executable codes, and when the processor executes the executable codes, the method for comparing faces based on privacy protection provided by the present specification is implemented.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the storage medium and the computing device embodiments, since they are substantially similar to the method embodiments, they are described relatively simply, and reference may be made to some descriptions of the method embodiments for relevant points.
Those skilled in the art will recognize that, in one or more of the examples described above, the functions described in connection with the embodiments of the invention may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium.
The above-mentioned embodiments further describe the objects, technical solutions and advantages of the embodiments of the present invention in detail. It should be understood that the above description is only exemplary of the embodiments of the present invention, and is not intended to limit the scope of the present invention, and any modification, equivalent replacement, or improvement made on the basis of the technical solutions of the present invention should be included in the scope of the present invention.
Claims (11)
1. A face comparison method based on privacy protection is applied to terminal equipment and comprises the following steps:
extracting a current face feature vector from a currently acquired first face image;
encrypting the current face characteristic vector by using a preset encryption algorithm to obtain a current ciphertext vector, wherein the preset encryption algorithm relates to an encryption matrix and a preset large integer;
performing inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector by using an intermediate matrix to obtain an intermediate result, wherein the registration ciphertext vector is obtained by encrypting the registration face feature vector obtained in the registration stage by using the preset encryption algorithm, and the intermediate matrix is determined based on an inverse matrix of the encryption matrix;
and determining a plain comparison result based on the intermediate result and the preset large integer, wherein the plain comparison result represents the similarity of the current face feature vector and the registered face feature vector.
2. The method of claim 1, wherein the obtaining the current ciphertext vector comprises:
encrypting the current face feature vector by using the encryption matrix and a preset large integer to obtain an initial face ciphertext;
generating a random error vector;
and adding the random error vector to the initial face ciphertext to obtain the current ciphertext vector.
3. The method of claim 1, further comprising:
in the registration stage, acquiring a registered face image, and extracting a registered face feature vector from the registered face image;
and encrypting the registered face feature vector based on the preset encryption algorithm to obtain the registered ciphertext vector.
4. The method of claim 1, further comprising:
determining the encryption matrix in the preset encryption algorithm;
determining the intermediate matrix using an inverse of the encryption matrix.
5. The method of claim 4, wherein the determining the intermediate matrix comprises:
determining a product of the transpose of the inverse of the encryption matrix and the inverse of the encryption matrix as the intermediate matrix.
6. The method of claim 4, wherein the predetermined encryption algorithm is a vector homomorphic encryption algorithm; the determining the encryption matrix in the preset encryption algorithm includes:
acquiring a public key matrix and a corresponding private key matrix in a vector homomorphic encryption algorithm;
determining the public key matrix as the encryption matrix; discarding the private key matrix.
7. The method of any one of claims 1-6, wherein the determining the plaintext alignment results comprises:
determining a first product representing a multiplication result of the current face feature vector and the registered face feature vector based on a ratio of the intermediate result to a specified power of the preset large integer, wherein the specified power is related to the preset encryption algorithm;
and determining the plaintext comparison result by using a preset distance formula and the first product.
8. The method of any of claims 1-6, further comprising:
and comparing the plaintext comparison result with a preset comparison threshold, and executing specified business operation based on the obtained comparison result.
9. The method of claim 8, wherein the specified business operation comprises one of: starting up operation, transaction execution operation, transfer operation and login operation.
10. A face comparison device based on privacy protection is deployed in a terminal device and comprises:
the first acquisition module is configured to extract a current face feature vector from a currently acquired first face image;
the first encryption module is configured to encrypt the current face feature vector by using a preset encryption algorithm to obtain a current ciphertext vector, wherein the preset encryption algorithm relates to an encryption matrix and a preset large integer;
the operation module is configured to perform inner product operation on the current ciphertext vector and a pre-stored registration ciphertext vector by using an intermediate matrix to obtain an intermediate result, wherein the registration ciphertext vector is obtained by encrypting the registration face feature vector obtained in the registration stage by using the preset encryption algorithm, and the intermediate matrix is determined based on an inverse matrix of the encryption matrix;
a first determining module configured to determine a plaintext comparison result based on the intermediate result and the preset large integer, wherein the plaintext comparison result represents a similarity between the current face feature vector and the registered face feature vector.
11. A computing device comprising a memory and a processor, wherein the memory has stored therein executable code that when executed by the processor implements the method of any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210496540.8A CN114978623B (en) | 2022-05-06 | 2022-05-06 | Face comparison method and device based on privacy protection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210496540.8A CN114978623B (en) | 2022-05-06 | 2022-05-06 | Face comparison method and device based on privacy protection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114978623A true CN114978623A (en) | 2022-08-30 |
| CN114978623B CN114978623B (en) | 2023-11-17 |
Family
ID=82981743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210496540.8A Active CN114978623B (en) | 2022-05-06 | 2022-05-06 | Face comparison method and device based on privacy protection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114978623B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117201698A (en) * | 2023-11-07 | 2023-12-08 | 北京隐算科技有限公司 | Safe and efficient image recognition method |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104821942A (en) * | 2015-04-22 | 2015-08-05 | 广州大学 | Face recognition method and system |
| JP2016014936A (en) * | 2014-06-30 | 2016-01-28 | 日本信号株式会社 | Movement pathway identification device |
| CN106096548A (en) * | 2016-06-12 | 2016-11-09 | 北京电子科技学院 | A kind of many intelligent terminal based on cloud environment share face secret recognition methods |
| KR101755995B1 (en) * | 2016-03-16 | 2017-07-10 | 인하대학교 산학협력단 | Method and system for feature vector based remote biometric verification using homomorphic encryption |
| CN107045627A (en) * | 2017-03-22 | 2017-08-15 | 海南大学 | A kind of face encryption recognition methods based on ciphertext domain |
| CN108446680A (en) * | 2018-05-07 | 2018-08-24 | 西安电子科技大学 | A kind of method for secret protection in face authentication system based on edge calculations |
| KR20180112302A (en) * | 2017-04-03 | 2018-10-12 | 한국전자통신연구원 | User authentication method based on face recognition |
| CN108881204A (en) * | 2018-06-08 | 2018-11-23 | 浙江捷尚人工智能研究发展有限公司 | Secret protection cluster data mining method, electronic equipment, storage medium and system |
| CN109165581A (en) * | 2018-08-09 | 2019-01-08 | 广州洪荒智能科技有限公司 | A kind of secret protection face identification method based on homomorphic cryptography |
| CN109344588A (en) * | 2018-09-03 | 2019-02-15 | 平安科技（深圳）有限公司 | Safety certifying method and terminal device |
| CN109918881A (en) * | 2019-01-22 | 2019-06-21 | 西安电子科技大学 | The online face authentication method and system of secret protection, intelligent movable equipment |
| CN111401277A (en) * | 2020-03-20 | 2020-07-10 | 深圳前海微众银行股份有限公司 | Face recognition model updating method, device, equipment and medium |
| CN112200133A (en) * | 2020-10-28 | 2021-01-08 | 支付宝(杭州)信息技术有限公司 | Privacy-protecting face recognition method and device |
| CN112398838A (en) * | 2020-11-06 | 2021-02-23 | 支付宝(杭州)信息技术有限公司 | Authentication method, device and equipment based on privacy protection |
| WO2021227349A1 (en) * | 2020-05-11 | 2021-11-18 | 华南理工大学 | Front-end facial image encryption and recognition method for biometric privacy protection |
| CN114093001A (en) * | 2021-11-16 | 2022-02-25 | 中国电子科技集团公司第三十研究所 | Face recognition method for protecting privacy security |
-
2022
- 2022-05-06 CN CN202210496540.8A patent/CN114978623B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2016014936A (en) * | 2014-06-30 | 2016-01-28 | 日本信号株式会社 | Movement pathway identification device |
| CN104821942A (en) * | 2015-04-22 | 2015-08-05 | 广州大学 | Face recognition method and system |
| KR101755995B1 (en) * | 2016-03-16 | 2017-07-10 | 인하대학교 산학협력단 | Method and system for feature vector based remote biometric verification using homomorphic encryption |
| CN106096548A (en) * | 2016-06-12 | 2016-11-09 | 北京电子科技学院 | A kind of many intelligent terminal based on cloud environment share face secret recognition methods |
| CN107045627A (en) * | 2017-03-22 | 2017-08-15 | 海南大学 | A kind of face encryption recognition methods based on ciphertext domain |
| KR20180112302A (en) * | 2017-04-03 | 2018-10-12 | 한국전자통신연구원 | User authentication method based on face recognition |
| CN108446680A (en) * | 2018-05-07 | 2018-08-24 | 西安电子科技大学 | A kind of method for secret protection in face authentication system based on edge calculations |
| CN108881204A (en) * | 2018-06-08 | 2018-11-23 | 浙江捷尚人工智能研究发展有限公司 | Secret protection cluster data mining method, electronic equipment, storage medium and system |
| CN109165581A (en) * | 2018-08-09 | 2019-01-08 | 广州洪荒智能科技有限公司 | A kind of secret protection face identification method based on homomorphic cryptography |
| CN109344588A (en) * | 2018-09-03 | 2019-02-15 | 平安科技（深圳）有限公司 | Safety certifying method and terminal device |
| CN109918881A (en) * | 2019-01-22 | 2019-06-21 | 西安电子科技大学 | The online face authentication method and system of secret protection, intelligent movable equipment |
| CN111401277A (en) * | 2020-03-20 | 2020-07-10 | 深圳前海微众银行股份有限公司 | Face recognition model updating method, device, equipment and medium |
| WO2021227349A1 (en) * | 2020-05-11 | 2021-11-18 | 华南理工大学 | Front-end facial image encryption and recognition method for biometric privacy protection |
| CN112200133A (en) * | 2020-10-28 | 2021-01-08 | 支付宝(杭州)信息技术有限公司 | Privacy-protecting face recognition method and device |
| CN112398838A (en) * | 2020-11-06 | 2021-02-23 | 支付宝(杭州)信息技术有限公司 | Authentication method, device and equipment based on privacy protection |
| CN114093001A (en) * | 2021-11-16 | 2022-02-25 | 中国电子科技集团公司第三十研究所 | Face recognition method for protecting privacy security |
Non-Patent Citations (2)
| Title |
|---|
| 杨雄;张晓惠;: "基于全同态加密的人脸特征密文认证系统", 微电子学与计算机, no. 09 * |
| 王保仓;韦永壮;胡予濮;: "基于中国剩余定理的快速公钥加密算法", 西安电子科技大学学报, no. 03 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117201698A (en) * | 2023-11-07 | 2023-12-08 | 北京隐算科技有限公司 | Safe and efficient image recognition method |
| CN117201698B (en) * | 2023-11-07 | 2024-01-12 | 北京隐算科技有限公司 | Safe and efficient image recognition method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114978623B (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2634955B1 (en) | Information authentication method and information authentication system | |
| Erkin et al. | Privacy-preserving face recognition | |
| EP3038286B1 (en) | Information processing method, information processing program, and information processing apparatus | |
| Gomez-Barrero et al. | Privacy-preserving comparison of variable-length data with application to biometric template protection | |
| CN109802942B (en) | Voiceprint authentication method with privacy protection function | |
| Guo et al. | Towards efficient privacy-preserving face recognition in the cloud | |
| CN112948795B (en) | Identity authentication method and device for protecting privacy | |
| EP3189459A1 (en) | Encrypting and decrypting information | |
| CN112101531B (en) | Neural network model training method, device and system based on privacy protection | |
| US20170085379A1 (en) | Learning parity with noise-based relational encryption for proximity relations | |
| WO2021184974A1 (en) | Identity authentication method for privacy protection, and apparatus | |
| CN114596639B (en) | Biological feature recognition method and device, electronic equipment and storage medium | |
| Gomez-Barrero et al. | Variable-length template protection based on homomorphic encryption with application to signature biometrics | |
| CN114978623B (en) | Face comparison method and device based on privacy protection | |
| Im et al. | Privacy-preserving palm print authentication using homomorphic encryption | |
| Jasmine et al. | A privacy preserving based multi-biometric system for secure identification in cloud environment | |
| Bentafat et al. | Towards real-time privacy-preserving video surveillance | |
| Takahashi et al. | Unconditionally provably secure cancellable biometrics based on a quotient polynomial ring | |
| Natgunanathan et al. | An overview of protection of privacy in multibiometrics | |
| Aparna et al. | An iris biometric-based dual encryption technique for medical image in e-healthcare application | |
| CN115001652A (en) | Image comparison method and device based on privacy protection | |
| CN113592696A (en) | Encryption model training, image encryption and encrypted face image recognition method and device | |
| Furukawa et al. | A secure face recognition scheme using noisy images based on kernel sparse representation | |
| Sarkar et al. | Cancelable biometric based key generation for symmetric cryptography | |
| CN112187477A (en) | Iris privacy authentication method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |