CN109359210A - The face retrieval method and system of double blind secret protection - Google Patents
The face retrieval method and system of double blind secret protection Download PDFInfo
- Publication number
- CN109359210A CN109359210A CN201810903319.3A CN201810903319A CN109359210A CN 109359210 A CN109359210 A CN 109359210A CN 201810903319 A CN201810903319 A CN 201810903319A CN 109359210 A CN109359210 A CN 109359210A
- Authority
- CN
- China
- Prior art keywords
- face
- image
- client
- missing
- missing crew
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 230000001815 facial effect Effects 0.000 claims abstract description 53
- 239000013598 vector Substances 0.000 claims abstract description 40
- 238000000605 extraction Methods 0.000 claims abstract description 26
- 238000012512 characterization method Methods 0.000 claims abstract description 20
- 238000001514 detection method Methods 0.000 claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000012790 confirmation Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 2
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000002452 interceptive effect Effects 0.000 description 3
- 208000027418 Wounds and injury Diseases 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000006378 damage Effects 0.000 description 2
- 208000014674 injury Diseases 0.000 description 2
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/602—Providing cryptographic facilities or services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computer Security & Cryptography (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- Bioethics (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Collating Specific Patterns (AREA)
Abstract
The present invention relates to a kind of face retrieval method and systems of double blind secret protection.This method comprises: 1) client detection includes the image of suspicious missing crew or the image comprising real missing crew, facial image is obtained;2) client carries out feature extraction to the facial image detected, obtains face characterization vector;3) client obtains face characterization vector to facial image and extraction and is encrypted and upload to Cloud Server;Cloud Server matches the human face image information of suspicious missing crew with the human face image information of real missing crew using face characterization vector in encrypted domain, obtains matching result, and matching result is decrypted into facial image;4) client receives the facial image decrypted according to matching result that Cloud Server is sent, and is confirmed whether to be real missing crew.The present invention has good Privacy Safeguarding, higher precision and processing speed, and be capable of providing effective missing crew especially missing child gives scheme for change.
Description
Technical field
The invention belongs to children's faces under artificial intelligence and information security field more particularly to a kind of double blind secret protection
Search method and system are particularly suitable for children's face retrieval field.
Background technique
There is the youngster of about 460,000,11.3 ten thousand and 100,000 respectively in the U.S., Britain and Germany in daily worldwide according to statistics
Child is missing.An essential attribute between individual is to discriminate between based on face to have now been developed to give the children lost for change
Some effective means, such as help center (the Center of Missing&Exploited with battered child by missing
Children, CME), the parent of missing child can be by the photographic intelligence comprising missing time and location and real missing child
Publication, can inform where and when to find if someone sees the missing child, so as to help to give for change
Missing child.
In recent years, with the fast development of social networks and cell phone, suspicious mistake is shot in the street by warm-hearted people
Then the photo of track children is posted on social networks is becoming a kind of feasible pattern for giving missing child for change.Nevertheless,
This mode can bring hidden danger, and the exposure of suspicious missing child photo may cause privacy and be invaded, and real missing child is then
Even exist by the danger of secondary injury.Therefore, it is necessary to a kind of effective schemes, can effectively give missing child for change
Meanwhile guarantee two sides safety, be not only avoided that suspicious missing child photo privacy leakage but can prevent real missing child by
Secondary injury.
The method of early stage carries out security context using Encryption Tool such as multi-party computations, privacy sharing, homomorphic cryptography etc.
Under vision calculate, these methods achieve good safety, but usually require that cloud retains original photo, cause hidden
Private leakage, and often computation complexity is very high for these methods;In addition a kind of method passes through functional encryption lifting system
Computational efficiency, but the feature extracted is required to stablize to guarantee precision.
Summary of the invention
In order to overcome the deficiencies of the prior art, the present invention provides a kind of face retrieval method and system of double blind secret protection
(Double Blind Finder) characterize to face and be matched using blind person's face, realize efficiently peace by the more attributes of low-dimensional
Full face retrieval.Herein, the operation in client and cloud that double blind refers to is carried out under security context, suspicious missing
The privacy of personnel's (such as suspicious missing child) and real missing crew (such as real missing child) picture all obtains at this both ends
Protection.
The method of the present invention and system are realized by client and cloud, and are related to interaction between the parties: 1) social networks is helped
The person of helping (Social Network Helpers, SNH), they clap after finding suspicious missing crew such as missing child according to scene
Photo is taken the photograph, Cloud Server is uploaded to by mark, extraction and encryption that system carries out children's face information, these information constitute
Children's face characteristic library to be retrieved.2) missing crew relatives, such as missing child parent (Parents of Missing
Children, PMC), the photo of real missing child can be carried out information labeling, extraction and encryption by system and uploaded by them
To Cloud Server;On the other hand, PMC can also be browsed the photo to match with the real missing child of oneself by system and be carried out
Confirmation.3) missing crew helps center, such as missing to help center (CME) with battered child, generates key for encrypting children people
Face information feeds back matching and confirmation result, if it is confirmed that being real missing child, then executes assistance, give missing youngster for change
It is virgin.
In order to solve the above technical problems, the present invention is achieved through the following technical solutions:
A kind of face retrieval method of double blind secret protection, comprising the following steps:
1) image of the client detection comprising suspicious missing crew or the image comprising real missing crew, obtain face
Image;
2) client carries out feature extraction to the facial image detected, obtains face characterization vector;
3) client obtains face characterization vector to facial image and extraction and is encrypted and upload to Cloud Server;So as to
Cloud Server is using face characterization vector to the human face image information of suspicious missing crew and the facial image of real missing crew
Information is matched in encrypted domain, obtains matching result, and matching result is decrypted into facial image;
4) client receives the facial image decrypted according to matching result that Cloud Server is sent, and is confirmed whether to be true
Positive missing crew.
A kind of face retrieval method of double blind secret protection, comprising the following steps:
1) Cloud Server receives the encrypted facial image and face characterization vector that client uploads;Wherein facial image
Be image of the client detection comprising suspicious missing crew or image comprising real missing crew and obtain, face characterize to
Amount is that client carries out feature extraction to the facial image detected and obtains;
2) human face image information and real missing crew of the Cloud Server using face characterization vector to suspicious missing crew
Human face image information matched in encrypted domain, obtain matching result;
3) matching result is decrypted into facial image and issues client by Cloud Server, is confirmed whether it is true so as to client
Positive missing crew.
Further, the missing crew be missing child, client detection from SNH shoot include suspicious missing youngster
Virgin image or the image comprising real missing child provided from PMC, obtain children's facial image;Cloud Server is to next
It is matched in encrypted domain from children's human face image information of SNH with children's face information from PMC, obtains matching knot
Fruit;The client of PMC receives the facial image decrypted according to matching result that Cloud Server is sent, and is confirmed whether it is oneself
Real missing child.
Further, the client use multiattribute recognition of face device, extract obtain face identity vector and
The attribute of two auxiliary, i.e. gender and age.
Further, the client encrypts the image comprising suspicious missing crew using following sub-step:
A) help center using the more attribute feature vectors and missing crew that arrive according to the image zooming-out of suspicious missing crew
The common parameter of offer generates a pair of of public key PKFAnd PKI;
B) PK is usedFMatched indicia F is encrypted, ciphertext tokens C is formedF;
C) PK is used to suspicious missing crew's facial imageIIt is encrypted, forms ciphertext image CI;
D) by ciphertext tokens CFWith ciphertext image CIUpload to Cloud Server.
Further, the client uses the more attribute feature vectors extracted according to real missing crew's photo and mistake
The master key that track personnel help center to provide generates a pair of of private key SKFAnd SKI, the two private keys return to missing crew relatives.
Further, the Cloud Server using blind matching algorithm in encrypted domain by SKFIt is close with each of Cloud Server
Text label CFCarry out retrieval matching;If it does, then by corresponding ciphertext image CIMissing crew relatives are pushed to, and use SKI
It is decrypted, obtains decrypted image I;So that missing crew relatives check to decrypted image I be confirmed whether it is really missing
Personnel.
Further, the Cloud Server is matched using safe inner product algorithm, comprising the following steps:
A) public key PK is inputtedF, private key SKF, ciphertext tokens collection { CFAnd matching score threshold range;
B) to each ciphertext tokens CFCompared retrieval one by one, specifically to score threshold t in threshold range by small
To being recycled greatly, to each threshold value t, the ciphertext tokens C under the threshold value is utilizedF,tDecryption obtains ciphertext m;
C) judge whether ciphertext m belongs to suspicious missing crew, if it is deconditioning, by the corresponding letter of ciphertext m
List of matches is added in breath.
A kind of client of the face retrieval for double blind secret protection comprising;
Face detection module is responsible for image of the detection comprising suspicious missing crew or the figure comprising real missing crew
Picture obtains facial image;
Face characteristic extraction module is responsible for carrying out feature extraction to the facial image detected, obtains face characterization vector;
Face information encrypting module is responsible for obtaining facial image and extraction face characterization vector to be encrypted and uploaded to
Cloud Server.
A kind of Cloud Server of the face retrieval for double blind secret protection comprising;
Blind matching module is responsible for receiving the encrypted facial image and face characterization vector that the client uploads, and
The human face image information of suspicious missing crew is matched in encrypted domain with the human face image information of real missing crew, is obtained
To matching result;
Matching confirmation and processing module are responsible for that matching result is decrypted into facial image and issues the client, so as to
The client is confirmed whether it is real missing crew.
A kind of face retrieval system of double blind secret protection comprising client recited above and Cloud Server.
The beneficial effects of the present invention are:
Children's face inspection for missing crew's face retrieval problem under secret protection, especially under double blind secret protection
Suo Wenti, search method of the invention and system have good Privacy Safeguarding, higher precision and processing speed, can
There is provided effective missing crew such as missing child gives scheme for change.
Detailed description of the invention
Fig. 1 is the flow chart of children's face retrieval method of double blind secret protection in the embodiment of the present invention.
Fig. 2 is children's face processing flow chart in the embodiment of the present invention.
Fig. 3 is the deep learning schematic network structure of more attribute recognitions of face in the embodiment of the present invention.
Fig. 4 is blind matching flow chart in the embodiment of the present invention.
Specific embodiment
To be clearer and more comprehensible above scheme and beneficial effect of the invention, hereafter by embodiment, and attached drawing is cooperated to make
It is described in detail.
The present embodiment by taking children's face retrieval as an example, provide a kind of double blind secret protection children's face retrieval method and
The system for realizing this method.Method includes the following steps:
Face datection is carried out using trained human-face detector in advance to the suspicious missing child photo of SNH shooting, and
Using preparatory trained recognition of face device to the more attribute feature vectors of children's face extraction detected;
The common parameter provided using the more attribute feature vectors and CME extracted generates a pair of of public key PKFAnd PKI.Its
Middle common parameter can be one be all 1 vector, be also possible to other suitable forms.
Using PKFMatched indicia F is encrypted, ciphertext tokens C is formedF;F is a two-value scalar, is marked whether
Match;
PK is used to the suspicious missing child facial image of SNH shootingIIt is encrypted, forms ciphertext image CI;
By ciphertext tokens CFWith ciphertext image CIUpload to shared cloud;
Face datection is carried out using trained human-face detector in advance to the real missing child photo that PMC is provided, and
Using preparatory trained recognition of face device to the more attribute feature vectors of children's face extraction detected;
The master key provided using the more attribute feature vectors and CME extracted generates a pair of of private key SKFAnd SKI, this two
A private key returns to PMC;
Using blind matching algorithm in encrypted domain by SKFWith each ciphertext tokens C in shared cloudFCarry out retrieval matching;
If it does, then by corresponding ciphertext image CIIt is pushed to PMC, and uses SKIIt is decrypted, obtains decrypted image
I;
PMC check being confirmed whether it is oneself real missing child to decrypted image I.
Children's face retrieval system of the double blind secret protection of the present embodiment, including client and Cloud Server, wherein visitor
Family end includes children's face detection module, children's face characteristic extraction module, children's face information encrypting module, Cloud Server packet
Containing blind matching module and matching confirmation and processing module, as shown in Figure 1.The specific mistake of children's face retrieval is carried out using the system
Journey includes:
1) image is received.
These images are from the SNH image comprising suspicious missing child shot or from PMC offer comprising real
The image of missing child.
2) it detects to obtain children's facial image by children's face detection module of client.
Children's face detection module is first using preparatory trained cascade multitask convolutional neural networks human-face detector
Automatic Face datection is carried out, children's face candidate is obtained and is shown on a user interface to SNH or PMC, these face candidates
In may include children's face of adult face, quality lower (such as attitudes vibration is big, excessively fuzzy), these faces are unfavorable
In children's face retrieval, therefore SNH or PMC selects quality higher (such as towards more front, than more visible) by interactive mode
Children's face candidate, obtain children's facial image.Human-face detector is made up of 3 multitask convolutional neural networks of cascade,
Each multitask convolutional neural networks include 3 convolutional layers and 1 full articulamentum, and predict 3 tasks: classification task judges whether
It is face, 2 recurrence tasks predict that face frame position and 5 face key points are set respectively.
3) feature extraction is carried out to the children's facial image detected by children's face characteristic extraction module of client,
Obtain children's face characterization vector.
Children's face characteristic extraction module realizes that extraction obtains the identity of children's face using multiattribute recognition of face device
The attribute of vector and two auxiliary, i.e. gender and age.
4) by children's face information encrypting module to children's facial image and relevant information carry out encryption upload to it is shared
Cloud.
Children's face information encrypting module handles children's facial image from SNH and PMC in different ways.It is right
Children's facial image from SNH, using public key PKIIt is encrypted, PKIIt is provided by the more attribute feature vectors and CME that extract
Common parameter generates;To children's facial image from PMC, using public key SKIIt is encrypted, SKIBy the more attributive character extracted
The master key that vector sum CME is provided generates.
5) children's human face image information from SNH is believed with children's facial image from PMC by blind matching module
Breath is matched in encrypted domain, obtains matching result.
Blind matching module is compared matching to face characteristic by safe inner product algorithm, obtains matching result.
6) result that matching retrieves is decrypted into children's facial image and shown to PMC by matching confirmation and processing module,
PMC is confirmed whether to be oneself real missing child by browsing matched children's facial image.If it is, notice CME into
Row processing.
It is the process of children's face processing as shown in Figure 2.The present invention is in one interactive interface of Client Design for mentioning
Rise the detection, identification, confirmation performance of children's face.Firstly, in client, (such as mobile phone etc. is with camera and mobile network
Portable device), children's face is detected by preparatory trained human-face detector and interactive selection obtains;Then children people
Face is described as one comprising multiattribute vector by children's face characteristic extraction module, the vector from the identity of 128 dimensions to
The age scalar composition of amount, the gender probability vector of 2 dimensions and 1 dimension, identity vector, gender probability vector, age scalar all pass through
Quantization and fixed point processing, age scalar are measured as unit of the moon;Then, the characteristic information of children's face, facial image
Information is encrypted handle and uploads to cloud;Then, beyond the clouds, children's face from PMC is by safety ratio to from SNH
Children's face characteristic library in retrieve, obtain matching result;Followed by matching result is returned to PMC, after being decrypted
Matched children's facial image is obtained, it is that oneself is true that PMC, which browses these matched children's facial images and is confirmed whether,
Just missing children;Finally, notifying the center CME to carry out giving processing for change if retrieving oneself real missing child.
As shown in figure 3, in order to extract children's face characteristic with strong discriminating power and efficient stable, youngster of the invention
Virgin face characteristic extraction module uses a multitask depth convolutional neural networks, and network includes 3 tasks: face identities
Task, Gender Classification task and age return task.Further, the identity characteristic that the present invention extracts is low-dimensional fixed point
, it ensure that the recall precision under security context.The network is mainly made of 3 × 3 convolutional layer, and 1 × 1 convolutional layer is put
It is placed between 3 × 3 convolutional layer or between maximum pond layer and 3 × 3 convolutional layer, slows down characteristic pattern number of active lanes gradually
Increase, enhances the non-thread sexuality of network, and reduce reasoning memory;In addition, global average pond layer be used to carry out at prediction
Reason, maximum pond layer is for reducing characteristic dimension.Finally, which includes 10 convolutional layers, 3 maximum pond layers, 1 overall situation
Average pond layer and 2 full articulamentums.To which obtained network parameter amount is 0.79M, therefore the treatment effeciency of network is very high.
P indicates the side length of input facial image in Fig. 3, and D indicates the feature vector dimension to be distilled, and identity layer is for compressing face characteristic
To more low dimensional, softmax layers belong to all kinds of probability for calculating.
As shown in figure 4, executing the retrieval of children's face beyond the clouds.In order to guarantee personal secrets, the present invention uses blind matching
It realizes.Specifically, the present invention uses safe inner product algorithm, to really missing children's face characteristic from PMC and comes from SNH
Suspicious missing children's face characteristic match, in encrypted domain complete face characteristic matching.Firstly, the input of algorithm
For public key PKF, private key SKF, ciphertext tokens collection { CFAnd matching score threshold range [ts,tS];Normally, score threshold is matched
Range is rule of thumb set in systems in advance;Then, to each ciphertext tokens CFCompared retrieval one by one, specifically to point
Threshold value t is ascending in threshold range is recycled for number, to each threshold value t, utilizes the ciphertext tokens C under the threshold valueF,tDecryption
Ciphertext m is obtained, then judges whether the ciphertext belongs to suspicious missing child, if it is deconditioning, the ciphertext is corresponding
List of matches is added in information.Smc indicates suspicious missing child in Fig. 4.
Using face retrieval method and system of the invention, tested on children's human face data collection, test method
The precision of k search result before return, the results showed that, when k=20, precision reaches 98.5%.This demonstrate the party
The validity of method.
In addition, it is also tested for computational efficiency of the present invention program in terms of blind matching, under the conditions of monokaryon single thread, from
The time that PMC obtains private key is 0.14 second, and the time for carrying out photo upload from SNH is 0.18 second, and the face blind matched time is
39 seconds.In view of being run using multithreading, layering and matching strategy and distributed deployment, PMC is after issuing a request, and general 15
Search result can be received in minute, this demonstrate the practicabilities of this method.
The implementation process of the method for the present invention is illustrated by taking children's face retrieval as an example above, it should be noted that, side of the present invention
Method is a kind of general face retrieval method, is not limited only to the face retrieval of children, is also applied for adult face retrieval, that is, uses
In giving missing adult for change.In addition, the neural network structure that face detection module of the invention, face characteristic extraction module use
The structure being also not necessarily limited in above-described embodiment can also use other applicable structure types.
It is above to implement to be merely illustrative of the technical solution of the present invention rather than be limited, the ordinary skill people of this field
Member can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the spirit and scope of the present invention, this hair
Bright protection scope should be subject to described in claims.
Claims (11)
1. a kind of face retrieval method of double blind secret protection, which comprises the following steps:
1) image of the client detection comprising suspicious missing crew or the image comprising real missing crew, obtain face figure
Picture;
2) client carries out feature extraction to the facial image detected, obtains face characterization vector;
3) client obtains face characterization vector to facial image and extraction and is encrypted and upload to Cloud Server;So as to cloud clothes
Business device is using face characterization vector to the human face image information of suspicious missing crew and the human face image information of real missing crew
It is matched in encrypted domain, obtains matching result, and matching result is decrypted into facial image;
4) client receives the facial image decrypted according to matching result that Cloud Server is sent, and is confirmed whether to be real
Missing crew.
2. a kind of face retrieval method of double blind secret protection, which comprises the following steps:
1) Cloud Server receives the encrypted facial image and face characterization vector that client uploads;Wherein facial image is visitor
The detection of family end includes the image of suspicious missing crew or the image comprising real missing crew and obtains, and face characterization vector is
Client carries out feature extraction to the facial image detected and obtains;
2) Cloud Server is using face characterization vector to the human face image information of suspicious missing crew and the people of real missing crew
Face image information is matched in encrypted domain, obtains matching result;
3) matching result is decrypted into facial image and issues client by Cloud Server, is confirmed whether it is real so as to client
Missing crew.
3. method according to claim 1 or 2, which is characterized in that the missing crew is missing child, client detection
The image comprising suspicious missing child from SNH shooting or the image comprising real missing child from PMC offer, obtain
To children's facial image;Cloud Server is adding children's human face image information from SNH with children's face information from PMC
It is matched on close domain, obtains matching result;What the client reception Cloud Server of PMC was sent decrypts to obtain according to matching result
Facial image, be confirmed whether to be oneself real missing child.
4. method according to claim 1 or 2, which is characterized in that the client uses multiattribute recognition of face device,
Extract the attribute of the identity vector for obtaining face and two auxiliary, i.e. gender and age.
5. method according to claim 1 or 2, which is characterized in that the client is using following steps to comprising suspicious
The image of missing crew encrypts:
A) center is helped to provide using the more attribute feature vectors and missing crew that are arrived according to the image zooming-out of suspicious missing crew
Common parameter, generate a pair of of public key PKFAnd PKI;
B) PK is usedFMatched indicia F is encrypted, ciphertext tokens C is formedF;
C) PK is used to suspicious missing crew's facial imageIIt is encrypted, forms ciphertext image CI;
D) by ciphertext tokens CFWith ciphertext image CIUpload to Cloud Server.
6. according to the method described in claim 5, it is characterized in that, the client according to real missing crew's photo using mentioning
The master key that the more attribute feature vectors got and missing crew help center to provide generates a pair of of private key SKFAnd SKI, the two
Private key returns to missing crew relatives.
7. according to the method described in claim 6, it is characterized in that, the Cloud Server uses blind matching algorithm will in encrypted domain
SKFWith each ciphertext tokens C in Cloud ServerFCarry out retrieval matching;If it does, then by corresponding ciphertext image CIPush
Missing crew relatives are given, and use SKIIt is decrypted, obtains decrypted image I;So as to missing crew relatives to decrypted image I into
Row checks be confirmed whether it is real missing crew.
8. the method according to the description of claim 7 is characterized in that the Cloud Server is using the progress of safe inner product algorithm
Match, comprising the following steps:
A) public key PK is inputtedF, private key SKF, ciphertext tokens collection { CFAnd matching score threshold range;
B) to each ciphertext tokens CFCompared retrieval one by one, specifically to score threshold t in threshold range it is ascending into
Row circulation, to each threshold value t, utilizes the ciphertext tokens C under the threshold valueF,tDecryption obtains ciphertext m;
C) judge whether ciphertext m belongs to suspicious missing crew, if it is deconditioning, the corresponding information of ciphertext m is added
Enter list of matches.
9. a kind of client of the face retrieval for double blind secret protection, which is characterized in that including;
Face detection module is responsible for image of the detection comprising suspicious missing crew or the image comprising real missing crew, is obtained
To facial image;
Face characteristic extraction module is responsible for carrying out feature extraction to the facial image detected, obtains face characterization vector;
Face information encrypting module, be responsible for facial image and extraction obtain face characterization vector encrypted and upload to cloud clothes
Business device.
10. a kind of Cloud Server of the face retrieval for double blind secret protection, which is characterized in that including;
Blind matching module, be responsible for receive claim 9 described in client upload encrypted facial image and face characterize to
Amount, and the human face image information of the human face image information of suspicious missing crew and real missing crew are carried out in encrypted domain
Match, obtains matching result;
Matching confirmation and processing module are responsible for that matching result is decrypted into facial image and issues the client, so as to described
Client is confirmed whether it is real missing crew.
11. a kind of face retrieval system of double blind secret protection, which is characterized in that including client as claimed in claim 9 and
Cloud Server described in any one of claim 10.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810903319.3A CN109359210A (en) | 2018-08-09 | 2018-08-09 | The face retrieval method and system of double blind secret protection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810903319.3A CN109359210A (en) | 2018-08-09 | 2018-08-09 | The face retrieval method and system of double blind secret protection |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109359210A true CN109359210A (en) | 2019-02-19 |
Family
ID=65349849
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810903319.3A Pending CN109359210A (en) | 2018-08-09 | 2018-08-09 | The face retrieval method and system of double blind secret protection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109359210A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110598464A (en) * | 2019-10-10 | 2019-12-20 | 山东浪潮人工智能研究院有限公司 | Data and model safety protection method of face recognition system |
CN110909189A (en) * | 2019-12-03 | 2020-03-24 | 支付宝(杭州)信息技术有限公司 | Method and device for processing face picture |
CN110932946A (en) * | 2019-11-25 | 2020-03-27 | 广州富港万嘉智能科技有限公司 | User meaning expression real-time judgment system with privacy protection and intelligent living room system |
CN111553320A (en) * | 2020-05-14 | 2020-08-18 | 支付宝(杭州)信息技术有限公司 | Feature extraction method for protecting personal data privacy, model training method and hardware |
WO2020252911A1 (en) * | 2019-06-19 | 2020-12-24 | 平安科技(深圳)有限公司 | Facial recognition method for missing individual, apparatus, computer device and storage medium |
CN112163238A (en) * | 2020-09-09 | 2021-01-01 | 中国科学院信息工程研究所 | Network model training method for multi-party participation data unshared |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101989300A (en) * | 2010-10-15 | 2011-03-23 | 江苏省莱科信息技术有限公司 | Missing person seeking system and implementation method for seeking missing person |
CN105046234A (en) * | 2015-08-04 | 2015-11-11 | 北京电子科技学院 | Invisible recognition method used for human face image in cloud environment and based on sparse representation |
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 |
-
2018
- 2018-08-09 CN CN201810903319.3A patent/CN109359210A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101989300A (en) * | 2010-10-15 | 2011-03-23 | 江苏省莱科信息技术有限公司 | Missing person seeking system and implementation method for seeking missing person |
CN105046234A (en) * | 2015-08-04 | 2015-11-11 | 北京电子科技学院 | Invisible recognition method used for human face image in cloud environment and based on sparse representation |
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 |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020252911A1 (en) * | 2019-06-19 | 2020-12-24 | 平安科技(深圳)有限公司 | Facial recognition method for missing individual, apparatus, computer device and storage medium |
CN110598464A (en) * | 2019-10-10 | 2019-12-20 | 山东浪潮人工智能研究院有限公司 | Data and model safety protection method of face recognition system |
CN110598464B (en) * | 2019-10-10 | 2021-04-27 | 浪潮集团有限公司 | Data and model safety protection method of face recognition system |
CN110932946A (en) * | 2019-11-25 | 2020-03-27 | 广州富港万嘉智能科技有限公司 | User meaning expression real-time judgment system with privacy protection and intelligent living room system |
CN110909189A (en) * | 2019-12-03 | 2020-03-24 | 支付宝(杭州)信息技术有限公司 | Method and device for processing face picture |
CN111553320A (en) * | 2020-05-14 | 2020-08-18 | 支付宝(杭州)信息技术有限公司 | Feature extraction method for protecting personal data privacy, model training method and hardware |
CN114419712A (en) * | 2020-05-14 | 2022-04-29 | 支付宝(杭州)信息技术有限公司 | Feature extraction method for protecting personal data privacy, model training method and hardware |
CN112163238A (en) * | 2020-09-09 | 2021-01-01 | 中国科学院信息工程研究所 | Network model training method for multi-party participation data unshared |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109359210A (en) | The face retrieval method and system of double blind secret protection | |
JP5383705B2 (en) | Determining social relationships from personal photo collections | |
US8542886B2 (en) | System for secure face identification (SCIFI) and methods useful in conjunction therewith | |
CN105260676B (en) | Image privacy decision-making commending system and its method | |
CN107480163A (en) | The efficient ciphertext image search method of secret protection is supported under a kind of cloud environment | |
CN110472519A (en) | A kind of human face in-vivo detection method based on multi-model | |
Shang et al. | A duo-generative approach to explainable multimodal covid-19 misinformation detection | |
Aly et al. | Intelligent recognition system for identifying items and pilgrims | |
Patel et al. | Trans-DF: a transfer learning-based end-to-end deepfake detector | |
Fitwi et al. | Minor privacy protection by real-time children identification and face scrambling at the edge | |
Naveen et al. | Face recognition and authentication using LBP and BSIF mask detection and elimination | |
Sharma et al. | A fully integrated violence detection system using CNN and LSTM. | |
Palanivinayagam et al. | An optimized machine learning and big data approach to crime detection | |
CN106296949A (en) | A kind of gate control system based on H264 video decoding techniques and method | |
Dandage et al. | Review of violence detection system using deep learning | |
Ajesh et al. | A hybrid method for fake profile detection in social networkusing artificial intelligence | |
CN114373200A (en) | Face acquisition system based on edge intelligent perceptron and privacy protection method | |
Li et al. | SRIM scheme: an impression-management scheme for privacy-aware photo-sharing users | |
CN108319659B (en) | Social contact discovery method based on encrypted image quick search | |
Kumar et al. | Image steganography analysis based on deep learning | |
Honda et al. | A study on fuzzy clustering-based k-anonymization for privacy preserving crowd movement analysis with face recognition | |
Yang et al. | Federation Learning of Optimized Convolutional Neural Network Structure for Intrusion Detection | |
Mukhiddin et al. | Privacy-preserving of human identification in CCTV data using a novel deep learning-based method | |
Bekeneva et al. | Local image processing in distributed monitoring system | |
Ataliya et al. | Web based access card generation using face recognition: a technical review |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190219 |