CN105844206A - Identity authentication method and identity authentication device - Google Patents
Identity authentication method and identity authentication device Download PDFInfo
- Publication number
- CN105844206A CN105844206A CN201510020554.2A CN201510020554A CN105844206A CN 105844206 A CN105844206 A CN 105844206A CN 201510020554 A CN201510020554 A CN 201510020554A CN 105844206 A CN105844206 A CN 105844206A
- Authority
- CN
- China
- Prior art keywords
- facial image
- key point
- image
- identity document
- holder
- 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
Landscapes
- Collating Specific Patterns (AREA)
Abstract
The invention discloses an identity authentication method and an identity authentication device. The method comprises the following steps: collecting an image of an identity document; automatically collecting a face image of a holder of the identity document in response to the detection of the identity document; getting the information recorded in the identity document from the collected identity document image; getting the holder information related to the holder from the collected face image; and comparing the information recorded in the identity document with the holder information to authenticate the identity of the holder. A convenient and accurate way of identity authentication without the need for users to go in person and based on user information and face comparison is provided for users and user identity authentication demanders.
Description
Technical field
The present invention relates to the technical field of pattern recognition, be specifically related to identification based on face and ratio
To identity identifying method and equipment.
Background technology
The appearance of the Internet and development are very easy to the life of people, under many traditional lines
Business can be carried out, such as transaction payment, payment etc. online.But additionally there are some business to consider
The problems such as bigger safety, still need to the accredited part of user and physically go to handle, as opened a bank account,
Modify password etc..Limit these business handlings is not the problem of operating process, and is because
Financial transaction is more sensitive, if bank cannot verify the true identity of user, then cannot be carried out
Corresponding operation, to avoid some undesirable unexpected generations.
In these need the business of true identity information of checking user, most common method is
The figure information on identity document that on-site verification transactor is held with him is the most consistent, Yi Jilu
The personal information entered and verify on the identity document of transactor.
But, if user can automatically obtain in self-service mode and verify above two information, this
Sample both can facilitate user, has saved the time of user, can be again that service side reduces business
Processing cost, obtains the subscriber data of reliable specification.
Open a bank account scene corresponding to online application, if bank can receive online, permissible
Confirm the information of user identity, and these Information Authentications are by meeting the requirements, then be the most permissible
Realize online opening an account.
The overall comparative maturity of face recognition technology, more in safety-security area application, most common of which
Mode be that the image prestored in the facial image collected and data base is compared, to test
Demonstrate,prove the identity of this people.But this authentication mode existing defects, most common of which deception side
Formula is photo deception.
The purpose of the present invention at least that by provide one need not user come to personally, facilitate, accurate
True identity identifying method and equipment, thus solve lacking of prior art the most to a certain extent
Fall into.
Summary of the invention
One side according to the disclosure, it is provided that a kind of identity identifying method, including: gather body
The image of part certificate;In response to described identity document being detected, automatically gather and hold described identity
The facial image of the holder of certificate;Described identity card is obtained from the identity document image gathered
Information described in part;Obtain relevant to described holder from the described facial image gathered
Holder information;And the information described in described identity document is entered with described holder information
Row compares, so that described holder is carried out authentication.
According to embodiment of the present disclosure, the information described in described identity document includes described body
Face key point information on part certificate, and obtain information described in described identity document
Step includes: detect human face region in the described identity document image gathered;And to institute
State the human face region detected in identity document image and carry out face key point identification, to determine
The face key point of the human face region of described identity document.
According to embodiment of the present disclosure, the holder of described identity document is held in collection automatically
The step of facial image includes: i) in response to described identity document being detected, adopt the most continuously
Collect multiple facial images of described holder;Ii) in each facial image of being gathered of detection whether
There is human face region, and give up the facial image being defined as that not there is human face region;Iii) when not
The time interval of the described facial image with human face region is described many more than giving up when setting threshold value
Individual facial image;And iv) repeat step i) to step iii), until obtained face figure
The number of picture reaches described predetermined number.
According to another embodiment of the disclosure, automatically gather and hold the accredited of described identity document
The step of the facial image of people includes: i) in response to described identity document being detected, automatically adopt
Collect the facial image of described holder;Ii) whether the described facial image that detection is gathered has
Human face region, and when described facial image is confirmed as the facial image without human face region,
Give up described facial image;And iii) repeat step i) to step ii), until obtained
The number of facial image reaches described predetermined number.
According to embodiment of the present disclosure, described holder information includes the face of described holder
Key point information, and the step obtaining the holder information relevant to described holder includes:
Described facial image is carried out face key point identification, to determine the face of described facial image
Key point.
According to embodiment of the present disclosure, described holder information also includes the people of described holder
The live body degree information of face image, and obtain the holder information relevant to described holder
Step also includes: based on a determination that the face key point of the described facial image gone out calculates described face
The live body degree value of image, so that it is determined that whether described facial image is live body image.
According to embodiment of the present disclosure, the information described in described identity document also includes described
The character information of identity document, and obtain the step of information described in described identity document also
Including: in the described identity document image gathered, identify the character information of described identity document;
And determine whether described character information meets the character information standard of described identity document.
According to embodiment of the present disclosure, the step carrying out authentication includes: based on being determined
The face key point of the human face region of the described identity document gone out and the described face figure determined
The face key point of picture, determines human face similarity degree.
According to embodiment of the present disclosure, detect institute according to cascade homing method based on SURF
State human face region and the human face region of described facial image of identity document image.
According to embodiment of the present disclosure, to the face district detected in described identity document image
Territory carries out the step of face key point identification and enters the facial image of the described predetermined number obtained
The step of pedestrian's face key point identification all includes: use the degree of depth convolutional neural networks of three-stage cascade
The first order estimate described face key point initial position;And based on estimated described people
The described initial position of face key point, uses supervision gradient descent method to calculate described face crucial
The exact position of point.
According to embodiment of the present disclosure, based on a determination that the face of the described facial image gone out is crucial
The step of the live body degree value that point calculates described facial image includes:
The region of the face key point according to the described facial image determined, extracts each described
Multiple non-rigid characteristic areas of facial image, described non-rigid characteristic area closes with described face
The region of key point is corresponding;Determine that each described facial image is in each described non-rigid characteristic area
Feature in territory;And go out each described non-rigid district according to the described feature calculation determined
The non-rigid degree in territory is as the live body degree value of described facial image.
According to another aspect of the present disclosure, it is provided that a kind of ID authentication device, it includes first
Harvester, is configured to the image of captured identity certificate;Second harvester, is configured to response
In described identity document being detected by described first harvester, automatically gather and hold described identity
The facial image of the holder of certificate;Information obtaining device, is configured to from the identity card gathered
Part image obtains the information described in described identity document and from by from the described face gathered
Image obtains the holder information relevant to described holder;And certification device, be configured to by
Information described in described identity document compares with described holder information, to hold described
Witness carries out authentication.
According to embodiment of the present disclosure, described information obtaining device includes: detection device, joins
It is set in the described identity document image gathered by described first harvester detect face district
Territory, and in the described facial image gathered by described second harvester, detect face district
Territory;And identification device, be configured to by described detection device in described identity document image
The human face region detected carries out face key point identification, to determine the people of described identity document
The face key point in face region, and determine that device detects in described facial image to by described
To human face region carry out face key point identification, with determine described facial image face close
Key point, and the information described in described identity document includes being determined by described identification device
The face key point of human face region of described identity document, described holder information includes by institute
State the face key point of the described facial image that identification device is determined, and described certification dress
Put the human face region of described identity document that is configured to be determined by described identification device
Face key point is crucial with the face of the described facial image determined by described identification device
Point, determines human face similarity degree.
According to embodiment of the present disclosure, described holder information also includes the people of described holder
The live body degree information of face image, described information obtaining device also includes that live body degree value calculates dress
Putting, described live body degree value calculation apparatus is configured to the institute determined by described identification device
The face key point stating facial image calculates the live body degree value of described facial image, so that it is determined that
Whether described facial image is live body image.
According to embodiment of the present disclosure, when described detection device is not examined in described facial image
When measuring human face region, described detection device gives up described facial image.
According to embodiment of the present disclosure, the information described in described identity document also includes described
The character information of identity document, described information obtaining device also includes character information device, described
Character information device identifies in the described identity document image gathered by described first harvester
The character information of described identity document and determine whether described character information meets described identity card
The character information standard of part.
According to embodiment of the present disclosure, described detection device returns according to cascade based on SURF
Method is returned to detect human face region and the face district of described facial image of described identity document image
Territory.
According to embodiment of the present disclosure, described identification device includes: initial position estimation module,
Based on the human face region detected in described facial image by described detection device, use three grades
The first order of the degree of depth convolutional neural networks of cascade estimates the initial position of described face key point;
And exact position computing module, based on by described in estimated by described initial position estimation module
The described initial position of face key point, uses supervision gradient descent method accurately to calculate described people
The exact position of face key point, to determine described face key point.
According to embodiment of the present disclosure, described live body degree value calculation apparatus includes: non-rigid
Characteristic area extraction module, people based on the described facial image determined by described identification device
Face key point extracts multiple non-rigid characteristic areas of each described facial image, described non-rigid
Characteristic area is corresponding with the region of described face key point;Characteristic determination module, determines each
Described facial image feature in each described non-rigid characteristic area;And live body degree value
Computing module, the described feature calculation according to being determined by described characteristic determination module goes out each
The non-rigid degree in described non-rigid region is as the live body degree value of described facial image.
According to embodiment of the present disclosure, described detection device includes: the first detector unit, joins
It is set in the described identity document image gathered by described first harvester detect face district
Territory;And second detector unit, it is configured to described in gathered by described second harvester
Facial image detects human face region.
According to embodiment of the present disclosure, described identification device includes: the first recognition unit, joins
It is set to the human face region detected in described identity document image by described first detector unit
Carry out face key point identification, crucial to determine the face of the human face region of described identity document
Point;And second recognition unit, be configured to by described second detector unit at described face figure
The human face region detected in Xiang carries out face key point identification, to determine described facial image
Face key point.
According to embodiment of the present disclosure, described ID authentication device is included in and can communicate
Terminal in.
Accompanying drawing explanation
Fig. 1 shows the schematic diagram of the ID authentication device according to illustrative embodiments;
Fig. 2 shows the schematic diagram of the information obtaining device according to illustrative embodiments;
Fig. 3 shows the flow chart of the identity identifying method according to illustrative embodiments;
Fig. 4 shows the captured identity certificate of the identity identifying method according to illustrative embodiments
Image and therefrom obtain the flow chart of the step of information being documented on identity document;
Fig. 5 shows the collection holder of the identity identifying method according to illustrative embodiments
The flow chart of facial image the therefrom step of the holder information that acquisition is relevant to holder;
Fig. 6 is that the disclosure for carrying out 21 key points that key point identification is used to face
Example;And
Fig. 7 is that the identity identifying method according to the disclosure is applied to opening a bank account and conclude the business scene
System block diagram.
Detailed description of the invention
Provide referring to the drawings following description with help be apparent from as claims and etc.
Each embodiment with the disclosure that scheme is limited.Hereinafter describe and include that each detail is with side
Assistant solves, but these details are considered as being only exemplary.Therefore, this area is common
It is to be understood by the skilled artisans that in the case of without departing substantially from the spirit and scope of the disclosure, can be right
Each embodiment described herein makes various changes and modifications.It addition, in order to clear and simple
See from tomorrow, the description of known function and structure may be eliminated.
Fig. 1 shows the schematic diagram of the ID authentication device 100 according to illustrative embodiments.
In FIG, ID authentication device 100 comprises the steps that housing the 110, first harvester
120, the second harvester 130, information obtaining device 140 and certification device 150.
First harvester 120 and the second harvester 130 may be provided at ID authentication device
Opening in the housing 110 of 100 and from housing 110 exposes to gather image.Specifically,
First harvester 120 may be provided at the back side of ID authentication device 100 and from housing 110
Opening on the back side exposes to gather image, and the second harvester 130 may be provided at identity
The front of authenticating device 100 opening from the front of housing 110 expose to gather image.
In the present embodiment, the first harvester 120 can be used for the image of captured identity certificate, with
And second harvester 130 can be used for gathering the image of the holder holding this identity document.So
And, it will be understood by those skilled in the art that the first harvester 120 can also be used for collection and holds body
The image of the holder of part certificate, the second harvester 130 can also be used for captured identity certificate
Image, the disclosure does not the most limit.
It addition, in the present embodiment, it is preferable that examine in response to by the first harvester 120
Measuring identity document, the second harvester 130 automatically (" silently ") gathers holds this body
The facial image of the holder of part certificate.Specifically, the holder of identity document is held towards body
The front of part authenticating device 100, the screen being arranged on ID authentication device 100 front (does not shows
Go out) show the view-finder of the first harvester 120.First harvester 120 is directed at identity
Certificate, with the image of captured identity certificate, is simultaneously in response to the first harvester 120 and detects
Identity document, the second harvester 130 automatically picks up in the case of holder is unwitting to be held
The facial image of witness.Preferably, the first harvester 120 and the second harvester 130 are right
The collection of the image of identity document and the facial image of holder is real-time, automatic, continual
Gather, gatherer process need not holder ID authentication device 100 is carried out any operation.
First harvester 120 and the second harvester 130 should within a certain period of time, such as in the several seconds,
The collection of the facial image of the image completing identity document and the holder of holding this identity document,
The image of identity document and the collection of facial image should not have bigger time interval, to ensure to adopt
Collect to image be the image of identity document and the image of holder at same scene.
Information obtaining device 140 may be provided in the housing 110 of ID authentication device 100, and
For obtaining described in identity document from the identity document image gathered by the first harvester 120
Information and from the holder holding this identity document gathered by the second harvester 130
Facial image obtains the holder information relevant to this holder.In the present embodiment, information
Acquisition device 140 obtains the information described in identity document from the image of identity document and such as can wrap
Include the face key point letter of human face region on the character information on identity document and identity document
Breath, and information obtaining device 140 obtain from the facial image of holder with this holder phase
Close holder information such as can include this holder facial image face key point information with
And the live body degree value of the facial image of holder, this is described hereinafter with reference to Fig. 2.
Certification device 150 may also be arranged in the housing 110 of ID authentication device 100.Certification
Device 150 can be based on the human face region of the identity document determined by information obtaining device 140
The facial image of face key point and the holder determined by information obtaining device 140
Face key point determines human face similarity degree.
Fig. 2 is the schematic diagram of the information obtaining device 140 according to illustrative embodiments.Such as figure
Shown in 2, information obtaining device 140 can include detecting device 141, identifying device 142, character
Massaging device 143 and live body degree value calculation apparatus 144.
Detection device 141 is at the identity document image gathered by the first harvester 120
Middle detection human face region and at the face figure of the holder gathered by the second harvester 130
Human face region is detected in Xiang.Detection device 141 can use such as based on SURF (Speeded Up
Robust Features) cascade homing method carry out the detection of human face region, this recurrence accords with substantially
Close the algorithm frame of Viola-Jone Face datection.
Information obtaining device 140 is described first below obtain identity document from identity document image
The exemplary operation of the information recorded.
Detection device 141 can detect that the square of face on gathered identity document image
Frame.When there being multiple frame on identity document, the most preferentially choose bigger face frame, have bigger
It is probably holder.Identify that detection device 141 can be detected in ID Card Image by device 142
To human face region carry out face key point identification.In the present embodiment, device 142 is identified
In detected human face region 21 can be carried out for the face key point of face alignment
Identify, the wherein position of 21 face key points (face key point 1 to face key point 21)
Put as shown in Figure 6.
Describe in detail below with reference to this example and identify device 142.As in figure 2 it is shown, identify dress
Put 142 and include initial position estimation module 1421 and exact position computing module 1422.Initially
Position estimation module 1421 is based on the people detected in identity document image by detection device 141
Face region uses first in the degree of depth convolutional neural networks (CNN) of a kind of three-stage cascade
Level carries out the general estimation of 21 face key points, to determine the initial position of face key point.
Exact position computing module 1422 is based on by face determined by initial position estimation module 1421
The initial position of key point uses and gradient such as carries out the supervision gradient descent method that cascade returns
(Supervised Descent Method (SDM)) is precisely located the accurate of face key point
Position, so that it is determined that go out face key point.
Specifically, the thinking of SDM is, by iteration repeatedly gradually accurately to face key point
Estimation.In the disclosure, from LFPW (Labeled Face Parts in the Wild) data
35 key point marks of collection choose sampling and obtains 21 required key points, and thus train
Obtaining iteration desired parameters in SDM, wherein the initial key point of SDM is by CNN first
Level processes and obtains.
In CNN method, the network in ground floor has obtained general to key point position
Estimating, rear two-layer makes this estimation the most accurate successively.But this method disadvantageously,
Amount of calculation is huge, and speed is slower, it is difficult to accomplish to calculate in real time under the computing capability of mobile terminal.
The advantage of SDM method is that speed is fast, but the method relies on initial key point position
Selecting, what bad initial position was possible causes reverting to locally optimal solution, to people
There is deviation in the estimation of face key point position.
Above two method, initial position is combined in the identification device 142 according to the disclosure
Estimation module 1421 uses the ground floor of CNN network to obtain relatively accurate face key point
Initial position, exact position computing module 1422 utilizes SDM method little by little linear regression to go out
Accurate face key point position.The advantage thus combining two kinds of methods so that the disclosure
Identify face key point the existing preferable robustness of process, have again speed faster.
As in figure 2 it is shown, information obtaining device also includes character information device 143, character information
Device 143 can be used for determining record on the identity document gathered by the first harvester 120
Character information.People's Republic of China's second generation identity card pair will be combined below in an illustrative manner
The exemplary operation of character information device 143 is described in detail.But those skilled in the art should
Understand, identity document may also include any kind of certificate being able to demonstrate that identity, as driver's license,
Passports etc., the disclosure does not the most limit.For identity card as identity document, character
Massaging device 143 can use " realization of Chinese name card for business identification system " of such as Zhang Chun et al.
In method carry out character recognition, and be appropriately modified the character arrangements adapting on identity card
Feature.This identification operation includes Image semantic classification, printed page analysis, character recognition, comprehension of information
Deng child-operation, finally it is output as the content of each character information item of identity card.
It is printed on holder personal information and facial image at People's Republic of China's second generation identity card
One side in, the arrangement of each character information item and human face region has its distinctive feature, such as,
Comprise name, sex, nationality, date of birth, address, citizenship number etc. in appearance
Character information, and the head portrait of holder.These information are all fixed on identity card and really position
Put, and the rule that every kind of information has it to determine, such as nationality only 56 kinds is possible, Gong Minshen
Part card number all has 18 etc..
Character information device 143 can pass through the character information on character recognition identification identity card, and
And judge whether these character informations meet the requirement of identity card character information.If identity card is few
The certificate of number ethnic mimority area, wherein comprises minority language, need to be according to minority language
The difference of content and typesetting exclusively carries out process.
Information obtaining device 140 is described below and obtains from the facial image of the holder gathered
The exemplary operation of the holder information relevant to this holder.
Detection device 141 can be additionally used in and detects everyone that gathered by the second harvester 130
Whether face image has human face region.In embodiment of the present disclosure, be to holder
Facial image carries out In vivo detection, it is thus possible to need gather continuous multiple frames facial image for
Live body degree value calculation apparatus 144, this will be described in detail hereinafter.Dress is gathered second
During putting 130 collection facial images, it is likely to be due to the acutely shake of holder and makes inspection
Surveying device 141 has a frame human face region cannot be detected in the facial image gathered, then give up
Abandon this frame;If human face region can not be detected in continuous multiple frames or is not detected by human face region
Time interval more than certain set threshold value, then give up the face images collected before,
Again facial image is gathered by the second harvester 130, until acquiring enough continuous print bags
Facial image containing human face region.
Additionally, detection device 141 based on the mistake determining human face region in identity document image
The process of Cheng Xiangtong detects face district in the facial image gathered by the second harvester 130
Territory.To this, its repetitive description will be omitted herein.
Identify that device 142 carries out face pass to by the human face region detected by detection device 141
Key point identification.Identify that device 142 is based on crucial with identification face in the human face region of identity card
The method that the method for point is identical carries out face pass in the human face region of the facial image of holder
Key point identification.Similarly, initial position estimation module 1421 uses the ground floor of CNN network
The initial bit of relatively accurate face key point in the human face region of the facial image obtaining holder
Putting, exact position computing module 1422 utilizes SDM method little by little linear regression to go out holder
Facial image human face region in accurate face key point position.To this, will save herein
Slightly its repetitive description.
As in figure 2 it is shown, information obtaining device also includes live body degree value calculation apparatus 144, live
Body degree value calculation apparatus 144 face figure based on the holder determined by identification device 142
The face key point of picture calculates the live body degree value of this facial image, so that it is determined that this face figure
Seem no for live body image.Specifically, live body degree value calculation apparatus 144 based on the most non-just
Body is analyzed (Non-rigid Motion Analysis) method and is detected the live body degree of face, this
It is described in detail below by reference to Fig. 2.
As in figure 2 it is shown, live body degree value calculation apparatus 144 includes that non-rigid characteristic area extracts
Module 1441, characteristic determination module 1442 and live body degree value computing module 1443.
In the present embodiment, non-rigid characteristic area extraction module 1441 is used for extracting face district
4 non-rigid characteristic area Ω: left ocular Ω of territory1, right ocular Ω2, nose region
Territory Ω3, mouth region Ω4, and have detected the non-rigid degree in these four regions respectively, i.e. live body
Degree.These 4 regions with by identifying that device 142 is at the human face region of the facial image of holder
The region of middle 21 determined face key points is generally corresponding to, can be according to identifying device 142
Key point recognition result determine 4 non-rigid characteristic areas.Specifically, left ocular Ω1
Include face key point 1,2,3,7,8 and 17, right ocular Ω2Include face key point
4,5,6,9,10 and 18, nasal area Ω3Include face key point 11,12,13 and 19, and
Mouth region Ω4Include face key point 14,15,16,20 and 21.In the disclosure, with one
The individual rectangle comprising key point corresponding to each characteristic area to represent each characteristic area,
The width of this rectangle and height need to individually need to be configured according to the difference in this feature region.
Characteristic determination module 1442 is for calculating in each frame facial image j according to formula (1)
Human face region FacejIn feature T of each non-rigid characteristic area Ωi,j, i=1...4,1: left
Ocular, 2: right ocular, 3: nasal area, 4: mouth region,
Wherein (x, y) is the coordinate of a pixel in image, and E is the non-rigid fortune in former method
Dynamic matrix,Represent the area in region ().
Live body degree value computing module 1443 is for calculating each non-rigid feature according to formula (2)
Feature T in regioniVariance as non-rigid degree S of each non-rigid characteristic areai(left eye portion
Region S1, right ocular S2, nasal area S3, mouth region S4), SiValue the least, table
Levying this facial image, to be not from the probability of live body the biggest:
Wherein, n is the totalframes of the image for In vivo detection,Average for ith feature.
Live body degree value calculation apparatus 144 is by the non-rigid degree of each non-rigid characteristic area
(S1,S2,S3,S4) as the overall live body degree of facial image of holder.
Certification device 150 is described more fully below based on by the body identifying that device 142 is determined
The face key point of the human face region of part certificate and the holder determined by identification device 142
The face key point of facial image determine human face similarity degree.
In the present embodiment, certification device 150 is respectively the people of the identity document determined
21 face key points in face region and 21 people of the facial image of holder determined
LBP (Local Binary Patterns) is extracted special on the different scale of face key point surrounding neighbors
Levy.LBP intrinsic dimensionality owing to extracting on different graphical rules is the highest, the most permissible
Reaching dimension up to ten thousand, be unfavorable for follow-up process, certification device 150 is according to such as PLDA
(Probabilistic Linear DiscriminantAnalysis) method is by Feature Dimension Reduction, by feature
The dimension of vector is down to hundreds of magnitude.
After extracting feature, certification device 150 such as Joint Bayesian model conduct
The algorithm of face alignment.As the one in Bayesian Estimation algorithm, this algorithm is by existing
Data be trained setting up two assume under Gauss model, assume that two faces respectively
The comparison sample model from same people and the model from different people.New for two afterwards
Face alignment sample, i.e. the face key point of the ID Card Image in the disclosure and facial image
Face key point, by compare they two assume under posterior probability whether calculate it
Belong to the probability of same person.Joint Bayesian algorithm is relative to traditional bayes method
Being made that improvement, the Joint Distribution of two sampling feature vectors is modeled rather than two sample spies by it
Levy the difference modeling of vector, so can capture differentiation information more rich between face, improve
The accuracy rate of comparison.Above-mentioned training sample set is all from including but not limited to LFPW
The data set of (Labeled Face Parts in the Wild) data set.
So, certification device 150 obtains the face figure of the face in ID Card Image and holder
The similarity degree of the face in Xiang, the facial image in this representative capacity card and holder face figure
Seem no be the degree of consistency of same person, its value is the biggest, it is believed that two images are same people
Probability the biggest.
The face live body exported by live body degree value calculation apparatus 144 and certification device 150 respectively
Degree value and human face similarity degree are as the two class figures gathered according to the ID authentication device of the disclosure
The reference of the human face similarity degree in Xiang, can be different according to the real needs of authentication application, and
Different modes is taked to integrate and use.
Alternately, in another embodiment, detection device 141 also include can be used for by
The described identity document image that first harvester 120 is gathered detects the first of human face region
Detector unit and can be used at the described facial image gathered by described second harvester 130
Second detector unit of middle detection human face region.It addition, identify that device 142 can include the first knowledge
Other unit and the second recognition unit, wherein the first recognition unit can be used for by the first detector unit
The human face region detected in identity document image carries out face key point identification, to determine
The face key point of the human face region of identity document, and the second recognition unit can be used for by
The human face region that two detector units detect in the facial image of holder carries out face key point
Identify, to determine the face key point of facial image.
Alternately, in another embodiment, it is possible in ID authentication device 100
Detection device 141 includes can be used for the described identity card being gathered by the first harvester 120
Part image detects the first detector unit of human face region and can be used for being gathered dress by described second
Put the second detector unit detecting human face region in the 130 described facial images gathered, identify
The device 142 human face region to being detected in identity document image by the first detector unit is carried out
Face key point identification, to determine the face key point of the human face region of identity document, and
The human face region detected in the facial image of holder by the second detector unit is carried out face
Key point identification, to determine the face key point of facial image.
Alternately, in yet, it is possible in ID authentication device 100
Identify that device 142 includes the first recognition unit and the second recognition unit, wherein the first recognition unit
Can be used for the human face region to being detected in identity document image by detection device 141 and carry out people
Face key point identification, to determine the face key point of the human face region of identity document, Yi Ji
Two recognition units can be used for by detecting what device 141 detected in the facial image of holder
Human face region carries out face key point identification, to determine the face key point of facial image.
Below with reference to Fig. 3, the identity identifying method according to illustrative embodiments is described.
In the exemplary embodiment, will be applied to comprise according to the identity identifying method of the disclosure
One in front and one in back on the mobile phone of two photographic head, the identity document for authentication is the China people
Republic's second generation identity card.It is understood by one skilled in the art that without departing substantially from the disclosure
In the case of scope, according to the identity identifying method of the disclosure can apply to various equipment such as mobile phone,
Computer etc., the identity document for authentication include but not limited to identity card, driver's license,
Passport etc..
In step 1000, the image in first captured identity card personal information face, then to word therein
Symbol and human face region extract respectively, finally know the face key point of human face region
, the face key point not obtained is for the face alignment of step 3000.
In step 2000, first gather the facial image of the holder of this identity card, then to wherein
Human face region extract, finally the face key point of the holder in human face region is carried out
Identifying, the face key point obtained is for the face alignment of step 3000.This step is also counted
The live body degree value of the facial image of gathered holder of being gone, to judge that this facial image is
Live body image.
In step 3000, according to the people of the identity card human face region identified in step 1000
Two faces are entered by face key point and the face key point of holder identified in step 2000
The comparison of row similarity degree, to determine human face similarity degree.
Preferably, in the specific implementation, the image acquisition in step 1000 and step 2000 by
Same equipment is implemented, as applied to having when the identity identifying method according to disclosure embodiment
One in front and one in back two photographic head mobile phone and be middle Chinese for carrying out the identity document of authentication
During people republic second generation identity card, one in front and one in back two photographic head of available mobile phone enter simultaneously
OK.In preferred image acquisition process example, holder should be towards mobile phone screen, and one is hand-held
Mobile phone, a hand-held identity card, cell phone front camera gather facial image, rear camera collection
Identity card personal information face image." simultaneously " refer to that holder need to be at synchronization respectively by face
Being placed in one in front and one in back before two photographic head with identity card, two photographic head should within a certain period of time, example
Within the several seconds, completing to gather, the collection of two images should not have bigger time interval, to ensure
The image collected is all ID Card Image and the facial image of holder at same scene.
It addition, it will be understood by those skilled in the art that according to the identity document used different, from
The information that this identity document is obtained may be different, and the disclosure does not the most limit.
Additionally, the information obtained from the facial image of holder is also not limited to above-mentioned face key point
With live body degree value, may also include any information that can be used for carrying out authentication.
Below in conjunction with Fig. 4 and Fig. 5 step 1000 to the identity identifying method according to the disclosure
It is described in detail with step 2000.
As shown in Figure 4, step 1000 includes the step 1001 of captured identity card image, identifies
Face district in the character recognition step 1002 of character, detection ID Card Image in ID Card Image
The step 1003 in territory and identify the step of face key point in the human face region detected
1004。
In step 1001, captured identity card image.Identity card holder towards mobile phone screen,
Mobile phone screen display post-positioned pick-up head view-finder, identity card is placed under post-positioned pick-up head by holder,
Post-positioned pick-up head is used to gather the identity card personal information face image of holder.This gatherer process is excellent
Select and enter with the process of collection holder facial image in the step 2001 that will be described below simultaneously
OK, to ensure that the image collected is certificate photograph and the holder photo at same scene.This is adopted
Collection is collection real-time, automatic, continual, need not holder to mobile phone in gatherer process
Carry out any operation.
In step 1002, the character in ID Card Image is identified.Make for identity card
For identity document, " realization of Chinese name card for business identification system " of such as Zhang Chun et al. can be used
In method carry out character recognition, and be appropriately modified the character arrangements adapting on identity card
Feature.This identification process includes Image semantic classification, printed page analysis, character recognition, comprehension of information
Etc. sub-process, thus identify the content of each character information item of identity card.
It is printed on holder personal information and facial image at People's Republic of China's second generation identity card
One side in, the arrangement of each character information item and human face region has its distinctive feature, such as,
Comprise name, sex, nationality, date of birth, address, citizenship number etc. in appearance
Character information, and the head portrait of holder.These information are all fixed on identity card and really position
Put, and the rule that every kind of information has it to determine, such as nationality only 56 kinds is possible, Gong Minshen
Part card number all has 18 etc..
If the character information detected by character recognition on identity card, and these characters letter
Breath meets the requirement of identity card character information, the then character information that output detections arrives, and otherwise returns
1001, the image of identity card is re-started collection.If identity card is the card of minority area
Part, wherein comprises minority language, need to according to the content of minority language and typesetting not
With exclusively carrying out process.
In step 1003, from the identity card personal information face image gathered step 1001
Middle detection human face region.This step have employed such as based on SURF (Speeded Up Robust
Features) cascade homing method carries out the detection of human face region, and this recurrence substantially conforms to
The algorithm frame of Viola-Jone Face datection.This step finally determines that on image, face is just
Square box.When there being multiple frame on image, the most preferentially choose bigger face frame, have bigger
It is probably holder.
If be detected that human face region, then enter step 1004, human face region is carried out key point
Identifying, if being not detected by human face region in the picture, otherwise returning step 1001, to identity
The image of card re-starts collection.
In step 1004, perform face key point identification.Bag is detected in step 1003
After square-shaped frame containing human face region, in this step 1004,21 in this frame are used for
The face key point of face alignment is identified, wherein 21 face key point (face key points
1 to face key point 21) position as shown in Figure 6.This step of the disclosure first uses such as
The first order in the degree of depth convolutional neural networks (CNN) of three-stage cascade carries out 21 key points
General estimation, uses afterwards and gradient such as carries out the supervision gradient descent method that cascade returns
(Supervised Descent Method (SDM)) carries out follow-up key point and is accurately positioned.
Specifically, the thinking of SDM is, by iteration repeatedly gradually accurately to face key point
Estimation.In the disclosure, from LFPW (Labeled Face Parts in the Wild) data
35 key point marks of collection choose sampling and obtains 21 required key points, and thus train
Obtaining iteration desired parameters in SDM, wherein the initial key point of SDM is by CNN first
Level processes and obtains.
In CNN method, the network in ground floor has obtained general to key point position
Estimating, rear two-layer makes this estimation the most accurate successively.But this method disadvantageously,
Amount of calculation is huge, and speed is slower, it is difficult to accomplish to calculate in real time under the computing capability of mobile terminal.
The advantage of SDM method is that speed is fast, but the method relies on initial key point position
Selecting, what bad initial position was possible causes reverting to locally optimal solution, to people
There is deviation in the estimation of face key point position.
The disclosure combines above two method, first obtains a phase with the ground floor of CNN network
Initial position to accurate face key point, then recycling SDM method, gradually line
Property returns out accurate face key point position.The advantage thus combining two kinds of methods, makes
Obtain the existing preferable robustness of process of the disclosure, have again speed faster.
It will be understood by those skilled in the art that step 1002 and step 1003-1004 can parallel, together
Time or be sequentially performed, the disclosure does not the most limit.
As it is shown in figure 5, step 2000 includes: gather holder facial image step 2001,
Detect the step 2002 of the human face region in the facial image of the holder gathered, detecting
Human face region in identify face key point step 2003 and calculate holder face figure
The step 2004 of the live body degree value of picture.
In step 2001, gather the facial image of holder.Holder towards mobile phone screen,
" mourn in silence " with mobile phone front-facing camera and gather the facial image of holder, the most not at screen
The view-finder of upper display front-facing camera, in the case of user's unaware, collects truer
Naturally facial image.The figure of the identity card in this gatherer process and step 1001 mentioned above
Carry out in gathering at preset time intervals, carry out the most simultaneously, the image collected with guarantee
It is certificate photograph and the holder photo at same scene.This collection is in real time, automatically, uninterruptedly
Collection, gatherer process need not holder mobile phone carried out any operation.
It addition, the disclosure to carry out In vivo detection to the facial image of holder, therefore in step
The facial image gathering continuous multiple frames may be needed in 2001 for follow-up calculating live body degree
The step 2004 of value.
In step 2002, the face of location from the facial image gathered step 2001
Region.Being embodied as of this step is essentially identical with step 1003.If acquiring enough companies
The continuous holder facial image comprising human face region, then carry out the face key point of step 2004
Identify.
" continuously " image in step 2001 and step 2002 does not represent strict sequential even
Continuous image, is likely to be due to what the acutely shake of holder face caused being gathered in step 2001
Some two field pictures that sequential is connected there is a frame human face region cannot be detected, then in step
Just skip detection human face region in this frame, then frame later in 2002, repeat step 2001
With step 2002 until acquiring the facial image that enough continuous print comprise human face region.Separately
Outward, if continuous multiple frames does not comprise face or do not comprises the time interval of human face region more than certain
Set threshold value, then should give up all images collected before, again be carried out by step 2001
Gather, until the image gathered has met requirement.
Alternately, as required and various condition limit (such as hardware condition etc.), also can be in step
Only gathering the facial image of a frame holder in rapid 2001, then in step 2002, detection should
The human face region of frame facial image, and if be not detected by face district in this frame facial image
Territory, then give up this frame.Repeat step 2001 and step 2002, until acquiring enough companies
The continuous facial image comprising human face region.If it addition, continuous multiple frames does not comprise face or not
Comprise the time interval of human face region and set threshold value more than certain, then should give up and collect before
All images, the collection again carried out by step 2001, until the image gathered has met requirement.
In step 2004, the human face region detected identifies face key point.This step
Specific implementation method essentially identical with step 1004, therefore, by omission, it is retouched in detail herein
State.
In step 2005, the face live body degree of the facial image of detection holder.The disclosure
Based on such as non-rigid analysis (the Non-rigid Motion Analysis) method live body journey to face
Degree detects.
In this step, when carrying out In vivo detection, the human face region of needs is obtained by step 2002.
As example, this step is extracted 4 non-rigid characteristic area Ω of human face region: left
Ocular Ω1, right ocular Ω2, nasal area Ω3, mouth region Ω4, and examine respectively
The non-rigid degree in these four regions, i.e. live body degree are surveyed.These 4 regions are crucial with face
The region of 21 the face key points obtained in some identification step 2003 is generally corresponding to, can basis
The key point recognition result of step 2003 determines 4 non-rigid characteristic areas.Specifically, left
Ocular Ω1Include face key point 1,2,3,7,8 and 17, right ocular Ω2Include
Face key point 4,5,6,9,10 and 18, nasal area Ω3Include face key point 11,12,13
With 19, and mouth region Ω4Include face key point 14,15,16,20 and 21.In these public affairs
In opening, represent each with a rectangle comprising key point corresponding to each characteristic area
Characteristic area, the width of this rectangle and height need to individually need to be configured according to the difference in this feature region.
The human face region Face in each two field picture j is calculated based on formula (1) abovejIn
Feature T of each non-rigid characteristic area Ωi,j, i=1...4,1: left ocular, 2: right eye
Region, portion, 3: nasal area, 4: mouth region.
Non-rigid degree S of each non-rigid characteristic area is calculated based on formula (2) abovei
(left ocular S1, right ocular S2, nasal area S3, mouth region S4), Mei Gefei
Non-rigid degree S of rigid body characteristic areaiFeature T for this regioniVariance, SiValue the least,
Characterizing this facial image, to be not from the probability of live body the biggest.
Finally by the non-rigid degree (S of each non-rigid characteristic area1,S2,S3,S4) as holder
The overall live body degree of facial image as output.
Step 3000 is described more fully below.
In step 3000, according to the identity card figure recognized in step 1004 and step 2003
Two faces are compared by the face key point of picture and the face key point of facial image.?
Behind the position of 21 face key points, at the different scale of these 21 key point surrounding neighbors
Upper extraction LBP (Local Binary Patterns) feature.Different graphical rules extracts
The LBP intrinsic dimensionality arrived is the highest, often can reach dimension up to ten thousand, be unfavorable for follow-up process.
Use PLDA (Probabilistic Linear Discriminant Analysis) method that feature is dropped
Dimension, is down to hundreds of magnitude by the dimension of characteristic vector.
After extracting feature, the disclosure such as can be with Joint Bayesian model as face
The algorithm of comparison.As the one in Bayesian Estimation algorithm, this algorithm is by existing number
According to be trained setting up two assume under Gauss model, assume that two face alignments respectively
The sample model from same people and the model from different people.Afterwards for two new people
Face comparison sample, i.e. according to face key point and the facial image of the ID Card Image in the disclosure
The face characteristic that obtains of face key point, general by the posteriority under comparing them and assuming at two
Rate calculates its probability whether belonging to same person.Joint Bayesian algorithm is relative to biography
The bayes method of system is made that improvement, and the Joint Distribution of two sampling feature vectors is modeled by it
Rather than the difference modeling of two sampling feature vectors, so can capture between face more rich
Differentiation information, improves the accuracy rate of comparison.Above-mentioned training sample set is all from including
But it is not limited to the data set of LFPW (Labeled Face Parts in the Wild) data set.
In step 3000, obtain in the facial image of the face in ID Card Image and holder
The similarity degree of face, facial image and holder facial image in this representative capacity card are
No is the degree of consistency of same person, and its value is the biggest, it is believed that two images be same people can
Energy property is the biggest.
In step 2003 and step 3000, face live body degree and the human face similarity degree of output is made
For the reference of the human face similarity degree in the two class images that the disclosure gathers, can answer according to authentication
Real needs different, and take different modes to integrate and use.
Fig. 7 shows that the identity identifying method application of the disclosure is opened an account and trading floor online to bank
The schematic diagram of scape.
To be applied to comprise one in front and one in back below in conjunction with by the identity identifying method according to the disclosure
It is the People's Republic of China (PRC) on the mobile phone of two photographic head and for the identity document of authentication
The example of second generation identity card describes the identity identifying method application of the disclosure and opens online to bank
Family and the process of transaction scene.
When opening an account, interior by the two of mobile phone photographic head, preferably same
Time, captured identity card and the facial image of holder, the word content of identity card is realized in this locality
Identification, the detection of human face region in ID Card Image, holder facial image in face district
The detection in territory, and complete face and the comparison of face in facial image in identity card, it is calculated
The similarity degree of two faces and the live body degree value of holder facial image, finally by network (example
As, wireless network, GPRS, WIFI etc.) immediately by the ID Card Image gathered above with hold
Character information, the similarity degree of two faces and holder on the facial image of witness, identity card
The live body degree value of facial image be uploaded to bank server, by server end according to the Ministry of Public Security
Identity information database judges the credibility of information above and retains.
When carrying out bank's online transaction, during face alignment can being joined authentication,
The modes such as comprehensive word password, gesture password, dynamic password carry out safety certification, also can be independent
It is authenticated.Local device is the most still used to complete the identification of character information of identity card, identity
Demonstrate,prove the detection of human face region in the facial image of the detection of human face region in image, holder, and
Complete face and the comparison of face in the facial image of holder in ID Card Image, be calculated
The live body degree value of the similarity degree of two faces and the facial image of holder, but last need to lead to
Cross network (e.g., wireless network, GPRS, WIFI etc.) immediately to be believed by the word on identity card
The live body degree of breath, the similarity of two faces and facial image is uploaded to bank server, by taking
Business device end judges the credibility of information above and the most permissible according to the User Information Database of bank
Carry out subsequent operation.The process carrying out image in this locality as above, and do not upload image
Information, to server end, is effectively reduced and is wirelessly transferred expense, and the computing of server end and depositing
Storage expense.
The disclosure utilizes wireless intelligent communications terminal, simultaneously, automatically gathers user true man live body people
Face photo and identity card picture thereof, the comparison of portrait in this locality carries out living body faces and identity card,
Obtaining the similarity of two faces, this similarity result can be as subsequent user true identity certification
A reference index.Convenient, quick, safe identity document is provided for consumer, businessman
Data Enter and authenticity verification technology.
One of ordinary skill in the art will appreciate that the whole or portion described in the above-described embodiment
Step by step or unit can in a software form and/or example, in hardware realizes, the disclosure is not restricted to appoint
The combination of the hardware and software of what particular form.
Claims (21)
1. an identity identifying method, including:
The image of captured identity certificate;
In response to described identity document being detected, automatically gather and hold the accredited of described identity document
The facial image of people;
The information described in described identity document is obtained from the identity document image gathered;
The holder information relevant to described holder is obtained from the described facial image gathered;
And
Information described in described identity document is compared with described holder information, with right
Described holder carries out authentication.
Method the most according to claim 1, wherein, the letter described in described identity document
Breath includes the face key point information on described identity document, and obtains in described identity document
The step of the information recorded includes:
Human face region is detected in the described identity document image gathered;And
The human face region detected in described identity document image is carried out face key point identification,
To determine the face key point of the human face region of described identity document.
Method the most according to claim 2, wherein, automatically gathers and holds described identity card
The step of the facial image of the holder of part includes:
I) in response to described identity document being detected, automatically holder described in continuous acquisition is many
Individual facial image;
Ii) whether each facial image of being gathered of detection has human face region, and give up and determine
For not having the facial image of human face region;
Iii) threshold value is set when the time interval of the described facial image without human face region is more than
Time give up the plurality of facial image;And
Iv) step i) is repeated to step iii), until the number of obtained facial image reaches
Described predetermined number.
Method the most according to claim 2, wherein, automatically gathers and holds described identity card
The step of the facial image of the holder of part includes:
I) in response to described identity document being detected, the face figure of described holder is automatically picked up
Picture;
Ii) whether the described facial image that detection is gathered has human face region, and as described people
When face image is confirmed as the facial image without human face region, give up described facial image;
And
Iii) step i) is repeated to step ii), until the number of obtained facial image reaches
Described predetermined number.
5., according to the method described in claim 3 or 4, wherein said holder information includes institute
State the face key point information of holder, it is thus achieved that the holder information relevant to described holder
Step includes:
Described facial image is carried out face key point identification, to determine described facial image
Face key point.
Method the most according to claim 5, wherein, described holder information also includes institute
State the live body degree information of the facial image of holder, and obtain relevant to described holder
The step of holder information also includes:
Based on a determination that the face key point of the described facial image gone out calculates the work of described facial image
Body degree value, so that it is determined that whether described facial image is live body image.
Method the most according to claim 2, wherein, the letter described in described identity document
Breath also includes the character information of described identity document, and obtains described in described identity document
The step of information also includes:
The character information of described identity document is identified in the described identity document image gathered;
And
Determine whether described character information meets the character information standard of described identity document.
Method the most according to claim 5, wherein,
The step carrying out authentication includes: face based on the described identity document determined
The face key point in region and the face key point of the described facial image determined, determine
Human face similarity degree.
9. according to the method described in claim 3 or 4, wherein, according to level based on SURF
Connection homing method detects human face region and the people of described facial image of described identity document image
Face region.
Method the most according to claim 5, wherein, in described identity document image
The human face region detected carries out the step of face key point identification and to the described predetermined number obtained
Purpose facial image carries out the step of face key point identification and all includes:
The first order using the degree of depth convolutional neural networks of three-stage cascade estimates described face key point
Initial position;And
Described initial position based on estimated described face key point, uses under supervision gradient
Fall method calculates the exact position of described face key point.
11. methods according to claim 6, wherein, based on a determination that the described face gone out
The step of the live body degree value that the face key point of image calculates described facial image includes:
The region of the face key point according to the described facial image determined, extracts each described
Multiple non-rigid characteristic areas of facial image, described non-rigid characteristic area closes with described face
The region of key point is corresponding;
Determine each described facial image feature in each described non-rigid characteristic area;With
And
The non-rigid journey in each described non-rigid region is gone out according to the described feature calculation determined
Spend the live body degree value as described facial image.
12. 1 kinds of ID authentication devices, including:
First harvester, is configured to the image of captured identity certificate;
Second harvester, is configured to described body be detected in response to by described first harvester
Part certificate, gathers the facial image of the holder holding described identity document automatically;
Information obtaining device, is configured to the identity document image from being gathered and obtains described identity card
Information described in part and from by obtaining from the described facial image gathered and described holder
Relevant holder information;And
Certification device, is configured to believe the information described in described identity document with described holder
Breath compares, so that described holder is carried out authentication.
13. ID authentication devices according to claim 12, wherein,
Described information obtaining device includes:
Detection device, is configured in the described identity gathered by described first harvester
Certificate image detects human face region, and is being gathered by described second harvester
Described facial image detects human face region;And
Identify device, be configured to by described detection device in described identity document image
The human face region detected carries out face key point identification, to determine described identity document
The face key point of human face region, and determine that device is at described face figure to by described
The human face region detected in Xiang carries out face key point identification, to determine described face
The face key point of image, and
Information described in described identity document includes described in described identification device is determined
The face key point of the human face region of identity document,
The described facial image that described holder information includes being determined by described identification device
Face key point, and
Described certification device is configured to the described identity card determined by described identification device
The face key point of the human face region of part and the described face figure determined by described identification device
The face key point of picture, determines human face similarity degree.
14. ID authentication devices according to claim 13, wherein, described holder is believed
Breath also includes the live body degree information of the facial image of described holder, described information obtaining device
Also include live body degree value calculation apparatus, described live body degree value calculation apparatus be configured to by
The face key point of the described facial image that described identification device is determined calculates described facial image
Live body degree value, so that it is determined that whether described facial image is live body image.
15. ID authentication devices according to claim 13, wherein, described identity document
Described in information also include the character information of described identity document, described information obtaining device is also
Including character information device, described character information device is by described first harvester collection
Described identity document image identifies the character information of described identity document and determines described character
Whether information meets the character information standard of described identity document.
16. ID authentication devices according to claim 13, wherein, described detection device
According to based on SURF cascade homing method detect described identity document image human face region and
The human face region of described facial image.
17. ID authentication devices according to claim 13, wherein, described identification device
Including:
Initial position estimation module, detects in described facial image based on by described detection device
The human face region arrived, uses the first order of the degree of depth convolutional neural networks of three-stage cascade to estimate described
The initial position of face key point;And
Exact position computing module, based on by described in estimated by described initial position estimation module
The described initial position of face key point, uses supervision gradient descent method accurately to calculate described people
The exact position of face key point, to determine described face key point.
18. ID authentication devices according to claim 14, wherein, described live body degree
Value calculation apparatus includes:
Non-rigid characteristic area extraction module, based on the described people determined by described identification device
The face key point of face image extracts multiple non-rigid characteristic areas of each described facial image,
Described non-rigid characteristic area is corresponding with the region of described face key point;
Characteristic determination module, determines that each described facial image is in each described non-rigid characteristic area
Feature in territory;And
Live body degree value computing module, according to being determined by described characteristic determination module
Feature calculation goes out the non-rigid degree work as described facial image in each described non-rigid region
Body degree value.
19. ID authentication devices according to claim 13, wherein, described detection device
Including:
First detector unit, is configured in the described identity gathered by described first harvester
Certificate image detects human face region;And
Second detector unit, is configured at the described face gathered by described second harvester
Image detects human face region.
20. ID authentication devices according to claim 19, wherein, described identification device
Including:
First recognition unit, be configured to by described first detector unit at described identity document figure
The human face region detected in Xiang carries out face key point identification, to determine described identity document
The face key point of human face region;And
Second recognition unit, be configured to by described second detector unit in described facial image
The human face region detected carries out face key point identification, to determine the people of described facial image
Face key point.
21. according to the ID authentication device according to any one of claim 14-20, wherein,
Described ID authentication device is included in the terminal that can communicate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510020554.2A CN105844206A (en) | 2015-01-15 | 2015-01-15 | Identity authentication method and identity authentication device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510020554.2A CN105844206A (en) | 2015-01-15 | 2015-01-15 | Identity authentication method and identity authentication device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105844206A true CN105844206A (en) | 2016-08-10 |
Family
ID=56580110
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510020554.2A Pending CN105844206A (en) | 2015-01-15 | 2015-01-15 | Identity authentication method and identity authentication device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105844206A (en) |
Cited By (34)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106330970A (en) * | 2016-11-01 | 2017-01-11 | 广东技术师范学院 | Mobile learning monitoring method and system |
CN106407912A (en) * | 2016-08-31 | 2017-02-15 | 腾讯科技(深圳)有限公司 | Face verification method and apparatus |
CN106446855A (en) * | 2016-09-30 | 2017-02-22 | 深圳市商汤科技有限公司 | Real name authentication device |
CN106529414A (en) * | 2016-10-14 | 2017-03-22 | 国政通科技股份有限公司 | Method for realizing result authentication through image comparison |
CN106570489A (en) * | 2016-11-10 | 2017-04-19 | 腾讯科技(深圳)有限公司 | Living body determination method and apparatus, and identity authentication method and device |
CN106682645A (en) * | 2017-01-16 | 2017-05-17 | 中兴智能视觉大数据技术(湖北)有限公司 | Intelligent visual big data airport security inspecting and security-protection method |
CN106874968A (en) * | 2016-12-21 | 2017-06-20 | 江苏国光信息产业股份有限公司 | A kind of China second-generation identity card distinguishing method between true and false of efficient robust |
CN107220633A (en) * | 2017-06-15 | 2017-09-29 | 苏州科达科技股份有限公司 | A kind of intelligent mobile enforcement system and method |
CN108494778A (en) * | 2018-03-27 | 2018-09-04 | 百度在线网络技术(北京)有限公司 | Identity identifying method and device |
CN108596011A (en) * | 2017-12-29 | 2018-09-28 | 中国电子科技集团公司信息科学研究院 | A kind of face character recognition methods and device based on combined depth network |
CN108875480A (en) * | 2017-08-15 | 2018-11-23 | 北京旷视科技有限公司 | A kind of method for tracing of face characteristic information, apparatus and system |
CN109145768A (en) * | 2018-07-31 | 2019-01-04 | 北京旷视科技有限公司 | Obtain the method and device of the human face data with face character |
CN109255299A (en) * | 2018-08-13 | 2019-01-22 | 北京市商汤科技开发有限公司 | Identity identifying method and device, electronic equipment and storage medium |
CN109344747A (en) * | 2018-09-17 | 2019-02-15 | 平安科技(深圳)有限公司 | A kind of recognition methods that distorting figure, storage medium and server |
CN109359502A (en) * | 2018-08-13 | 2019-02-19 | 北京市商汤科技开发有限公司 | False-proof detection method and device, electronic equipment, storage medium |
CN109410393A (en) * | 2018-10-11 | 2019-03-01 | 郴州金通信息科技有限公司 | A kind of ordering dishes and remotely enter device for hotel's catering business |
CN109508705A (en) * | 2018-12-29 | 2019-03-22 | 河北三川科技有限公司 | The dynamic threshold face alignment setting method of identity-based card validity period |
CN109977867A (en) * | 2019-03-26 | 2019-07-05 | 厦门瑞为信息技术有限公司 | A kind of infrared biopsy method based on machine learning multiple features fusion |
CN110019909A (en) * | 2017-12-13 | 2019-07-16 | 航天信息股份有限公司 | A kind of method and device thereof for realizing Identity Management using portrait alignment algorithm |
CN110059579A (en) * | 2019-03-27 | 2019-07-26 | 北京三快在线科技有限公司 | For the method and apparatus of test alive, electronic equipment and storage medium |
CN110214320A (en) * | 2017-08-09 | 2019-09-06 | 居米奥公司 | The certification compared using face-image |
CN110414454A (en) * | 2019-07-31 | 2019-11-05 | 南充折衍智能光电科技有限公司 | A kind of testimony of a witness unification identifying system based on machine vision |
WO2020034733A1 (en) * | 2018-08-13 | 2020-02-20 | 北京市商汤科技开发有限公司 | Identity authentication method and apparatus, electronic device, and storage medium |
CN111626274A (en) * | 2020-07-30 | 2020-09-04 | 四川骏逸富顿科技有限公司 | Social security card identification method and identification system thereof |
TWI706264B (en) * | 2017-12-14 | 2020-10-01 | 大陸商深圳市商湯科技有限公司 | Method and apparatus for verifying identity document, electronic device, and storage medium |
CN111860078A (en) * | 2019-04-30 | 2020-10-30 | 北京眼神智能科技有限公司 | Face silence living body detection method and device, readable storage medium and equipment |
WO2020220453A1 (en) * | 2019-04-29 | 2020-11-05 | 众安信息技术服务有限公司 | Method and device for verifying certificate and certificate holder |
CN111968264A (en) * | 2020-10-21 | 2020-11-20 | 东华理工大学南昌校区 | Sports event time registration device |
CN112767586A (en) * | 2020-12-31 | 2021-05-07 | 深圳市商汤科技有限公司 | Passage detection method and device, electronic equipment and computer readable storage medium |
CN112912876A (en) * | 2018-06-28 | 2021-06-04 | 因文提亚有限责任公司 | System and method for online verification of an object's identity |
CN113283359A (en) * | 2021-06-02 | 2021-08-20 | 万达信息股份有限公司 | Authentication method and system for handheld certificate photo and electronic equipment |
CN113486727A (en) * | 2021-06-11 | 2021-10-08 | 福建数博讯信息科技有限公司 | Face recognition optimization method based on certificate photo as data source |
CN113781209A (en) * | 2021-09-27 | 2021-12-10 | 度小满科技(北京)有限公司 | Credit service data processing method and related device |
WO2022057682A1 (en) * | 2020-09-15 | 2022-03-24 | 中国银联股份有限公司 | Data processing method and device, and computer-readable storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103425972A (en) * | 2013-09-02 | 2013-12-04 | 北方工业大学 | Method and system for automatically identifying true and false of identity document possessor |
CN103440482A (en) * | 2013-09-02 | 2013-12-11 | 北方工业大学 | Method, system and device for identifying identity document holder based on hidden video |
CN103559485A (en) * | 2013-11-08 | 2014-02-05 | 广州广电运通金融电子股份有限公司 | Identity authentication method and device based on remote bank VTM |
CN103678984A (en) * | 2013-12-20 | 2014-03-26 | 湖北微模式科技发展有限公司 | Method for achieving user authentication by utilizing camera |
CN103824054A (en) * | 2014-02-17 | 2014-05-28 | 北京旷视科技有限公司 | Cascaded depth neural network-based face attribute recognition method |
CN103824050A (en) * | 2014-02-17 | 2014-05-28 | 北京旷视科技有限公司 | Cascade regression-based face key point positioning method |
CN103886301A (en) * | 2014-03-28 | 2014-06-25 | 中国科学院自动化研究所 | Human face living detection method |
CN103914686A (en) * | 2014-03-11 | 2014-07-09 | 辰通智能设备(深圳)有限公司 | Face comparison authentication method and system based on identification photo and collected photo |
-
2015
- 2015-01-15 CN CN201510020554.2A patent/CN105844206A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103425972A (en) * | 2013-09-02 | 2013-12-04 | 北方工业大学 | Method and system for automatically identifying true and false of identity document possessor |
CN103440482A (en) * | 2013-09-02 | 2013-12-11 | 北方工业大学 | Method, system and device for identifying identity document holder based on hidden video |
CN103559485A (en) * | 2013-11-08 | 2014-02-05 | 广州广电运通金融电子股份有限公司 | Identity authentication method and device based on remote bank VTM |
CN103678984A (en) * | 2013-12-20 | 2014-03-26 | 湖北微模式科技发展有限公司 | Method for achieving user authentication by utilizing camera |
CN103824054A (en) * | 2014-02-17 | 2014-05-28 | 北京旷视科技有限公司 | Cascaded depth neural network-based face attribute recognition method |
CN103824050A (en) * | 2014-02-17 | 2014-05-28 | 北京旷视科技有限公司 | Cascade regression-based face key point positioning method |
CN103914686A (en) * | 2014-03-11 | 2014-07-09 | 辰通智能设备(深圳)有限公司 | Face comparison authentication method and system based on identification photo and collected photo |
CN103886301A (en) * | 2014-03-28 | 2014-06-25 | 中国科学院自动化研究所 | Human face living detection method |
Cited By (40)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106407912A (en) * | 2016-08-31 | 2017-02-15 | 腾讯科技(深圳)有限公司 | Face verification method and apparatus |
CN106446855A (en) * | 2016-09-30 | 2017-02-22 | 深圳市商汤科技有限公司 | Real name authentication device |
CN106529414A (en) * | 2016-10-14 | 2017-03-22 | 国政通科技股份有限公司 | Method for realizing result authentication through image comparison |
CN106330970A (en) * | 2016-11-01 | 2017-01-11 | 广东技术师范学院 | Mobile learning monitoring method and system |
CN106570489A (en) * | 2016-11-10 | 2017-04-19 | 腾讯科技(深圳)有限公司 | Living body determination method and apparatus, and identity authentication method and device |
CN106874968A (en) * | 2016-12-21 | 2017-06-20 | 江苏国光信息产业股份有限公司 | A kind of China second-generation identity card distinguishing method between true and false of efficient robust |
CN106874968B (en) * | 2016-12-21 | 2020-03-24 | 江苏国光信息产业股份有限公司 | Second-generation identity card authenticity identification method |
CN106682645A (en) * | 2017-01-16 | 2017-05-17 | 中兴智能视觉大数据技术(湖北)有限公司 | Intelligent visual big data airport security inspecting and security-protection method |
CN107220633A (en) * | 2017-06-15 | 2017-09-29 | 苏州科达科技股份有限公司 | A kind of intelligent mobile enforcement system and method |
CN110214320A (en) * | 2017-08-09 | 2019-09-06 | 居米奥公司 | The certification compared using face-image |
CN108875480A (en) * | 2017-08-15 | 2018-11-23 | 北京旷视科技有限公司 | A kind of method for tracing of face characteristic information, apparatus and system |
CN110019909A (en) * | 2017-12-13 | 2019-07-16 | 航天信息股份有限公司 | A kind of method and device thereof for realizing Identity Management using portrait alignment algorithm |
TWI706264B (en) * | 2017-12-14 | 2020-10-01 | 大陸商深圳市商湯科技有限公司 | Method and apparatus for verifying identity document, electronic device, and storage medium |
US11256943B2 (en) | 2017-12-14 | 2022-02-22 | Shenzhen Sensetime Technology Co., Ltd. | Method and apparatus for verifying identity document, electronic device, and storage medium |
CN108596011A (en) * | 2017-12-29 | 2018-09-28 | 中国电子科技集团公司信息科学研究院 | A kind of face character recognition methods and device based on combined depth network |
CN108494778A (en) * | 2018-03-27 | 2018-09-04 | 百度在线网络技术(北京)有限公司 | Identity identifying method and device |
CN112912876A (en) * | 2018-06-28 | 2021-06-04 | 因文提亚有限责任公司 | System and method for online verification of an object's identity |
CN109145768A (en) * | 2018-07-31 | 2019-01-04 | 北京旷视科技有限公司 | Obtain the method and device of the human face data with face character |
CN109255299A (en) * | 2018-08-13 | 2019-01-22 | 北京市商汤科技开发有限公司 | Identity identifying method and device, electronic equipment and storage medium |
WO2020034733A1 (en) * | 2018-08-13 | 2020-02-20 | 北京市商汤科技开发有限公司 | Identity authentication method and apparatus, electronic device, and storage medium |
CN109359502A (en) * | 2018-08-13 | 2019-02-19 | 北京市商汤科技开发有限公司 | False-proof detection method and device, electronic equipment, storage medium |
CN109344747A (en) * | 2018-09-17 | 2019-02-15 | 平安科技(深圳)有限公司 | A kind of recognition methods that distorting figure, storage medium and server |
CN109344747B (en) * | 2018-09-17 | 2024-01-05 | 平安科技(深圳)有限公司 | Tamper graph identification method, storage medium and server |
CN109410393A (en) * | 2018-10-11 | 2019-03-01 | 郴州金通信息科技有限公司 | A kind of ordering dishes and remotely enter device for hotel's catering business |
CN109508705A (en) * | 2018-12-29 | 2019-03-22 | 河北三川科技有限公司 | The dynamic threshold face alignment setting method of identity-based card validity period |
CN109977867A (en) * | 2019-03-26 | 2019-07-05 | 厦门瑞为信息技术有限公司 | A kind of infrared biopsy method based on machine learning multiple features fusion |
CN110059579A (en) * | 2019-03-27 | 2019-07-26 | 北京三快在线科技有限公司 | For the method and apparatus of test alive, electronic equipment and storage medium |
CN112507889A (en) * | 2019-04-29 | 2021-03-16 | 众安信息技术服务有限公司 | Method and system for verifying certificate and certificate holder |
WO2020220453A1 (en) * | 2019-04-29 | 2020-11-05 | 众安信息技术服务有限公司 | Method and device for verifying certificate and certificate holder |
CN111860078B (en) * | 2019-04-30 | 2024-05-14 | 北京眼神智能科技有限公司 | Face silence living body detection method, device, readable storage medium and equipment |
CN111860078A (en) * | 2019-04-30 | 2020-10-30 | 北京眼神智能科技有限公司 | Face silence living body detection method and device, readable storage medium and equipment |
CN110414454A (en) * | 2019-07-31 | 2019-11-05 | 南充折衍智能光电科技有限公司 | A kind of testimony of a witness unification identifying system based on machine vision |
CN111626274A (en) * | 2020-07-30 | 2020-09-04 | 四川骏逸富顿科技有限公司 | Social security card identification method and identification system thereof |
CN111626274B (en) * | 2020-07-30 | 2020-10-27 | 四川骏逸富顿科技有限公司 | Social security card identification method |
WO2022057682A1 (en) * | 2020-09-15 | 2022-03-24 | 中国银联股份有限公司 | Data processing method and device, and computer-readable storage medium |
CN111968264A (en) * | 2020-10-21 | 2020-11-20 | 东华理工大学南昌校区 | Sports event time registration device |
CN112767586A (en) * | 2020-12-31 | 2021-05-07 | 深圳市商汤科技有限公司 | Passage detection method and device, electronic equipment and computer readable storage medium |
CN113283359A (en) * | 2021-06-02 | 2021-08-20 | 万达信息股份有限公司 | Authentication method and system for handheld certificate photo and electronic equipment |
CN113486727A (en) * | 2021-06-11 | 2021-10-08 | 福建数博讯信息科技有限公司 | Face recognition optimization method based on certificate photo as data source |
CN113781209A (en) * | 2021-09-27 | 2021-12-10 | 度小满科技(北京)有限公司 | Credit service data processing method and related device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105844206A (en) | Identity authentication method and identity authentication device | |
WO2020155939A1 (en) | Image recognition method and device, storage medium and processor | |
CN108009528B (en) | Triple Loss-based face authentication method and device, computer equipment and storage medium | |
CN109858371A (en) | The method and device of recognition of face | |
US9514356B2 (en) | Method and apparatus for generating facial feature verification model | |
US20180034852A1 (en) | Anti-spoofing system and methods useful in conjunction therewith | |
CN107844748A (en) | Auth method, device, storage medium and computer equipment | |
CN104700094B (en) | A kind of face identification method for intelligent robot and system | |
CN106557726A (en) | A kind of band is mourned in silence the system for face identity authentication and its method of formula In vivo detection | |
WO2022206319A1 (en) | Image processing method and apparatus, and device, storage medium and computer program product | |
TWI712980B (en) | Claim information extraction method and device, and electronic equipment | |
CN110321792A (en) | Without card enchashment method, apparatus, equipment and computer readable storage medium | |
CN107622246A (en) | Face identification method and Related product | |
CN108875509A (en) | Biopsy method, device and system and storage medium | |
CN109657627A (en) | Auth method, device and electronic equipment | |
KR20210155303A (en) | Online Test System using face contour recognition AI to prevent the cheating behaviour and method thereof | |
CN111738199A (en) | Image information verification method, image information verification device, image information verification computing device and medium | |
CN109448193A (en) | Identity information recognition methods and device | |
CN106022272A (en) | Front face determining method and device | |
CN113591603A (en) | Certificate verification method and device, electronic equipment and storage medium | |
Surantha et al. | Lightweight face recognition-based portable attendance system with liveness detection | |
CN108734099A (en) | Auth method and device, electronic equipment, computer program and storage medium | |
Castillo et al. | Class attendance generation through multiple facial detection and recognition using artificial neural network | |
CN107025435A (en) | A kind of face recognition processing method and system | |
Yuan et al. | SALM: Smartphone-based identity authentication using lip motion characteristics |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160810 |
|
RJ01 | Rejection of invention patent application after publication |