CN107577987A - Identity authentication method, system and device - Google Patents
Identity authentication method, system and device Download PDFInfo
- Publication number
- CN107577987A CN107577987A CN201710647019.9A CN201710647019A CN107577987A CN 107577987 A CN107577987 A CN 107577987A CN 201710647019 A CN201710647019 A CN 201710647019A CN 107577987 A CN107577987 A CN 107577987A
- Authority
- CN
- China
- Prior art keywords
- image
- face
- infrared light
- visible images
- near infrared
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
Abstract
The present invention relates to a kind of identity authentication method, its method comprises the following steps：Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；ID Card Image, face visible images and face near infrared light image are inputted to being previously-completed in the Triplets CNN models of training, the convolution feature of ID Card Image, face visible images and face near infrared light image is extracted, obtains corresponding characteristic vector；Calculate the similarity of any two image in ID Card Image, face visible images and face near infrared light image respectively according to characteristic vector；The uniformity of ID Card Image, face visible images and face near infrared light image is judged according to similarity, and exports identity authentication result.Triplets and depth convolutional neural networks CNN models are effectively combined in the present invention, the robustness of authentication, and then the accuracy rate of structural reform authentication can be effectively improved.
Description
Technical field
The present invention relates to pattern-recognition and field of artificial intelligence, more particularly to a kind of identity authentication method, is
System and device.
Background technology
Now, information security is increasingly valued by people, in order to ensure information is not repaiied by the people of no access rights
Change or random spurious information, carrying out authentication to people is just particularly important.At present, many occasions in daily life
(such as in life, tourism, work etc.) is required for showing and examining identity document, and carries out personnel identity according to identity document
Certification, such as banking are handled, take the vehicles such as high ferro or aircraft, move in hotel etc..In verification process, mainly
Relevant information is examined, to ensure being consistent property between holder and certificate, i.e., " testimony of a witness unification ".
At present, conventional identity identifying method mainly has two kinds, and a kind of is first by data of the identity card in public security internet
The verification of essential information is carried out in storehouse, by manually by the information contrast in holder and identity card, the verification method efficiency
It is lower than relatively low and recognition accuracy.Another method is to utilize identification authentication system, while captured identity card image, and utilization can
See that light catches the image of holder, the ID Card Image of collection and holder image are then subjected to image comparison, tied comparing
When fruit is consistent, it is verified.
However, in authentication procedures, during visible light collection image, illumination, posture and expression etc. pair
IMAQ influence is bigger, and causing image, rate is low respectively, so as to cause the accuracy rate of authentication poor.
The content of the invention
Based on this, it is necessary to the problem of for existing identity identifying method accuracy rate difference, there is provided a kind of side of authentication
Method and system.
A kind of identity authentication method system, including：
Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；
The ID Card Image, the face visible images and the face near infrared light image are inputted to complete in advance
Into in the Triplets CNN models of training, it is near to extract the ID Card Image, the face visible images and the face
The convolution feature of infrared light image, obtains corresponding characteristic vector；
It is near that the ID Card Image, the face visible images and the face are calculated according to the characteristic vector respectively
The similarity of any two image in infrared light image；
According to the similarity to the ID Card Image, the face visible images and the face near infrared light figure
The uniformity of picture is judged, and exports identity authentication result.
A kind of identity authentication method system, including：
Image collection module, the ID Card Image, face visible images and face for obtaining personnel to be certified are closely red
Outer light image；
Characteristic vector obtains module, for the ID Card Image, the face visible images and the face is near
Infrared light image is inputted to being previously-completed in the Triplets CNN models of training, extracts the ID Card Image, the face
The convolution feature of visible images and the face near infrared light image, obtains corresponding characteristic vector；
Similarity calculation module, for calculating the ID Card Image respectively according to the characteristic vector, the face can
See the similarity of any two image in light image and the face near infrared light image；
Authentication judge module, for according to the similarity to the ID Card Image, the face visible ray figure
The uniformity of picture and the face near infrared light image is judged, and exports identity authentication result.
A kind of computer-readable recording medium, is stored thereon with computer program, it is characterised in that the program is by processor
Following step is realized during execution：
Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；
The ID Card Image, the face visible images and the face near infrared light image are inputted to complete in advance
Into in the Triplets CNN models of training, it is near to extract the ID Card Image, the face visible images and the face
The convolution feature of infrared light image, obtains corresponding characteristic vector；
It is near that the ID Card Image, the face visible images and the face are calculated according to the characteristic vector respectively
The similarity of any two image in infrared light image；
According to the similarity to the ID Card Image, the face visible images and the face near infrared light figure
The uniformity of picture is judged, and exports identity authentication result.
A kind of device of authentication, it is characterised in that including shell body, be installed in the display of the shell body face side
Shield, be installed in the binocular camera of the shell body top surface side, be installed in the ID Card Image harvester of shell body bottom surface side
And it is installed in the processor inside the shell body；
The binocular camera is used for the face visible images and face near infrared light image for gathering personnel to be certified；
The ID Card Image harvester is used for the ID Card Image for gathering personnel to be certified；
The processor is used to perform following steps：
Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；
The ID Card Image, the face visible images and the face near infrared light image are inputted to complete in advance
Into in the Triplets CNN models of training, it is near to extract the ID Card Image, the face visible images and the face
The convolution feature of infrared light image, obtains corresponding characteristic vector；
It is near that the ID Card Image, the face visible images and the face are calculated according to the characteristic vector respectively
The similarity of any two image in infrared light image；
According to the similarity to the ID Card Image, the face visible images and the face near infrared light figure
The uniformity of picture is judged, and exports identity authentication result；
The display screen is used to show the authentication result.
The ID Card Image, face visible images and face near infrared light image of personnel to be certified is obtained in the present invention,
Then the ID Card Image of collection, face visible images and face near infrared light image are inputted to being previously-completed training
In Triplets CNN models, the convolution feature of extraction ID Card Image, face visible images and face near infrared light image,
Obtain corresponding characteristic vector；It is near to calculate the ID Card Image, face visible images and face respectively according to characteristic vector
The similarity of any two image in infrared light image；According to similarity to ID Card Image, face visible images and face
The uniformity of near infrared light image is judged, and exports identity authentication result.By Triplets and depth convolution in the present invention
Neutral net CNN models effectively combine, and can improve the robustness of authentication, and then improve the accuracy rate of authentication.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the identity authentication method of the present invention in one of the embodiments；
Fig. 2 is the schematic flow sheet of the identity authentication method of the present invention in one of the embodiments；
Fig. 3 is the schematic flow sheet of the identity authentication method of the present invention in one of the embodiments；
Fig. 4 is the schematic diagram of Triplets CNN models in identity authentication method of the invention；
Fig. 5 is the schematic flow sheet of the identity authentication method of the present invention in one of the embodiments；
Fig. 6 a, Fig. 6 b and Fig. 6 c in the identity authentication method of the present invention according to default triple alternative condition from institute
State the schematic diagram for the triple image that the condition that meets is chosen in triple image；
Fig. 7 is the schematic flow sheet of the identity authentication method of the present invention in one of the embodiments；
Fig. 8 is the schematic flow sheet of the identity authentication method of the present invention in one of the embodiments；
Fig. 9 is the schematic flow sheet of the device of the authentication of the present invention in one of the embodiments；
Figure 10 is the structural representation of the device of the authentication of the present invention in one embodiment.
Embodiment
Present disclosure is described in further detail below in conjunction with preferred embodiment and accompanying drawing.Obviously, hereafter institute
The embodiment of description is only used for explaining the present invention, rather than limitation of the invention.It is general based on the embodiment in the present invention, this area
The every other embodiment that logical technical staff is obtained under the premise of creative work is not made, belong to what the present invention protected
Scope.It should be noted that for the ease of describing, part related to the present invention rather than full content are illustrate only in accompanying drawing.
Fig. 1 is the schematic flow sheet of the identity authentication method of the present invention in one embodiment, as shown in figure 1, this hair
Identity authentication method in bright embodiment, comprises the following steps：
Step S110, obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified.
Personnel to be certified described in the present embodiment are the people for arbitrarily needing to receive authentication.Carrying out image authentication
When, gather the face visible images and face near infrared light image of personnel to be certified simultaneously using binocular camera, then will
The face visible images and face near infrared light image collected detect human face region by human-face detector.In addition, treat
The ID Card Image of certification personnel can be directly obtained using reader device or scan personnel's to be certified by scanning means
Identity card obtains.
It should be appreciated that the ID Card Image of personnel to be certified, face visible images and face near infrared light image will can
Obtained with other conventional image-pickup methods.
Step S120, ID Card Image, face visible images and face near infrared light image are inputted to being previously-completed
In the Triplets CNN models of training, the volume of extraction ID Card Image, face visible images and face near infrared light image
Product feature, obtains corresponding characteristic vector.
Specifically, when carrying out authentication, first choice first to gather obtain personnel to be measured ID Card Image, face it is visible
Light image and face near infrared light image, then these images are inputted to the Triplets CNN models completed to training in advance,
Using Triplets and CNN related algorithms, ID Card Image, face visible images and face near infrared light image are extracted respectively
Convolution feature, obtain the characteristic vector of each image in these images, i.e., by Triplets CNN models, calculate class origin respectively
The characteristic vector of part card image, face visible images and face near infrared light image.
Step S130, calculate ID Card Image, face visible images and face near infrared light respectively according to characteristic vector
The similarity of any two image in image.
Step S140, according to similarity to ID Card Image, face visible images and face near infrared light image one
Cause property is judged, and exports identity authentication result.
In the present embodiment, using calculating ID Card Image, face visible images and face near infrared light image
Characteristic vector calculates ID Card Image, face visible images and face near infrared light image between any two similar respectively
Degree, then according to the result of similarity, to the uniformity of ID Card Image, face visible images and face near infrared light image
Judged, and export identity authentication result.Wherein identity authentication result can be the ID Card Image and face figure of people to be measured
Seem it is consistent, i.e., " certification unification "；Can also be the ID Card Image of people to be measured and facial image is inconsistent.
Above-mentioned identity authentication method, gather first the ID Card Image of personnel to be certified, face visible images and
Face near infrared light image, then the ID Card Image of collection, face visible images and face near infrared light image are inputted
To being previously-completed in the Triplets CNN models of training, ID Card Image, face visible images and face near-infrared are extracted
The convolution feature of light image, obtains corresponding characteristic vector；The ID Card Image is calculated according to characteristic vector respectively, face can
See the similarity of any two image in light image and face near infrared light image；According to similarity to ID Card Image, face
The uniformity of visible images and face near infrared light image is judged, and exports identity authentication result.Will in the present invention
Triplets and depth convolutional neural networks CNN models effectively combine, and can effectively improve the robustness of authentication, and then
Improve the accuracy rate of authentication.
In wherein a kind of embodiment, as shown in Fig. 2 ID Card Image, face visible images and face is closely red
Outer light image is inputted to being previously-completed in the Triplets CNN models of training, extracts ID Card Image, face visible images
With the convolution feature of face near infrared light image, obtain corresponding to characteristic vector the step of before, in addition to：
Step S150, using quality evaluation algorithm to ID Card Image, face visible images and face near infrared light figure
As carrying out quality evaluation.
When specifically, due to gathering facial image (face visible images and face near infrared light image) at the scene, figure
As quality easily by when ambient lighting (intensity or dim light), camera collection image focal length alignment etc. it is multifactor influenceed, once
The poor image quality (such as resolution ratio is low) collected, or the image that once gathers of identity card harvester are smudgy etc.,
These images can all influence the accuracy of the authentication in later stage.Therefore, after facial image and ID Card Image is collected
Quality evaluation is carried out to image, sees whether it is applied to the authentication in later stage.If poor image quality cannot be used for later stage body
Part certification, just resurveys facial image and ID Card Image.In the present embodiment, scene is adopted using quality evaluation algorithm
The facial image and identity card of collection carry out quality evaluation.In addition, after facial image is collected, live body inspection can also be carried out to it
Survey, to ensure that the facial image collected is the live body image from personnel on site to be measured collection, rather than utilize personnel to be measured
The image that collects such as photo.
Step S160, when ID Card Image, face visible images and face near infrared light image quality meet to require,
Image preprocessing is carried out to ID Card Image, face visible images and face near infrared light image.
In the present embodiment, meeting in ID Card Image, face visible images and face near infrared light image quality will
When asking, ID Card Image, face visible images and face near infrared light image are pre-processed, such as to identity card figure
Picture, face visible images and face near infrared light image carry out face and it handled, to ID Card Image, face visible ray figure
Picture and face near infrared light image carry out resolution processes, are consistent its resolution ratio.These images are pre-processed,
The accuracy rate of the characteristics of image of extraction can effectively be strengthened, and then increase the accuracy of authentication.
In wherein a kind of embodiment, as shown in figure 3, the Triplets CNN models for being previously-completed training pass through following step
It is rapid to obtain：
Step S170, several ID Card Images, face visible images corresponding with ID Card Image and people are obtained in advance
Face near infrared light image, and with image, face visible images and face near infrared light image structure training set in several identity
Triple image, triple image includes reference sample image, similar sample image and foreign peoples's sample image.
Specifically, triple is the concept in the Triplet loss functions based on metric learning, a triple
Triplets is made up of (i.e. anchor samples (i.e. reference sample), positive samples (i.e. similar sample), negative samples
Foreign peoples's sample), so-called triple is exactly three samples, such as (anchor, positive, negative), wherein, a and p are same
One kind, a and n are inhomogeneities.The process so learnt is exactly to acquire a kind of expression, for triple as much as possible so that
Anchor and positive distance is less than anchor and negative distance.I.e.：
WhereinRepresent reference sample,Similar sample is represented,Foreign peoples's sample is represented, α represents specific threshold, 0.0
Between~1.0, it is proposed that be worth for 0.2.Inequality substantially defines the distance between similar sample and foreign peoples's sample relation, i.e.,：
The distance between all similar samples+threshold alpha is less than the distance between foreign peoples's sample.Above-mentioned formula is changed, obtained
Object function based on Triplets：
The implication of object function is exactly the triple for being unsatisfactory for condition, is optimized；Ternary for meeting condition
Group, just first no matter.
In the present embodiment, several ID Card Images, face visible images corresponding with ID Card Image are gathered first
With face near infrared light image, trained with image, face visible images and face near infrared light image structure in several identity
The triple image of collection.
Step S180, the triple figure for the condition that meets is chosen from triple image according to default triple alternative condition
Picture.
In the present embodiment, default triple alternative condition can be with ID Card Image, face visible images and
An image in face near infrared light image in any one image is as reference sample, using other kind of image as similar sample
Or foreign peoples's sample, select in similar sample with reference sample in farthest image and foreign peoples's sample with reference sample distance most
Near image, generation meet the triple image of condition.In addition, triple alternative condition can be according to real in authentication procedures
The demand on border is designed, and selection mode is not unique.
Step S190, the triple image for the condition that meets is inputted into CNN models and is trained, obtain what training was completed
Triplets CNN models.
In the present embodiment, then these are met the three of condition by the triple image for meeting condition that will be screened
Tuple image inputs into CNN models (such as Fig. 4), and the Triplets CNN models that training is completed can just be obtained by being trained.Its
Middle Triplets CNN models (i.e. Triplets depth convolutional neural networks model) are mainly by convolutional layer, pond layer, Quan Lian
Connect layer (average pond layer), Triplets loss layers composition；Generally the number of convolutional layer is different, can basis
Actual demand is adjusted, and can be used for local average and sub-sample immediately following a pond layer behind each convolutional layer, in convolution and
Continuous alternating between sampling, is finally exported by full articulamentum.Authentication procedures by triple (personnel's i.e. to be measured
ID Card Image, face visible images and face near infrared light image) it is put into Triplets CNN network models, through excessive
Secondary computing, make Triplets loss as small as possible and restrain, finally when extracting the feature of facial image, be extracted
The output of the full articulamentum of last in Triplets CNN (average pond layer), according to the Triplets CNN models of training,
Full articulamentum (average pond layer) feature of identity card picture, near-infrared facial image, visible ray facial image is extracted respectively.
In wherein a kind of embodiment, as shown in figure 5, the first reference sample triple image, the second reference sample ternary
Group image and the triple image that the 3rd reference sample triple image is the condition that meets, according to default triple alternative condition
Chosen from triple image in the step of meeting the triple image of condition, including：
Step S181, any one ID Card Image is chosen from reference sample image as the first reference sample, from same
One is chosen in class sample image with the first reference sample apart from farthest face near infrared light image, and from foreign peoples's sample image
A face near infrared light image closest with the first reference sample is chosen, generates the first reference sample triple image；
Step S182, any one face near infrared light image is chosen from reference sample image and refers to sample as second
This, chooses one with the second reference sample apart from farthest face visible images from similar sample image, and from foreign peoples's sample
This image chooses a face visible images closest with the first reference sample, generates the second reference sample triple figure
Picture；
Step S183, any one people's visible images are chosen from reference sample image as the 3rd reference sample, from
One is chosen in similar sample image to choose apart from farthest ID Card Image, and from foreign peoples's sample image with the 3rd reference sample
One ID Card Image closest with the 3rd reference sample, generate the 3rd reference sample triple image.
In the present embodiment, ID Card Image is designated as IDPIC, face near-infrared image is designated as NIR, visible images letter
Title is designated as VIS.The step of the triple image for the condition that meets is chosen from triple image according to default triple alternative condition
It is rapid as follows：
(1)IDPIC-NIR：An identity card picture is chosen from anchor sample images as the first reference sample,
In theorem in Euclid space, one is chosen from nositive sample images with identity card picture apart from farthest face near-infrared face figure
Picture, hard anchor-nositive are formed, selection one is closest with identity card picture from negative sample images
Face near-infrared facial image, form hard anchor-negative, as shown in Figure 6 a；Wherein anchor, hard
Anchor-nositive and hard anchor-negative constitute the first reference sample triple image.Similarly, NIR-
IDPIC is also as implied above.
(2)NIR-VIS：A near-infrared image is chosen from anchor sample images as the second reference sample, in Europe
In formula space, one is chosen from positive sample images with near-infrared image apart from farthest visible ray facial image, structure
Into hard anchor-Positive, the closest visible ray of a near-infrared image is chosen from negative sample images
Facial image, hard anchor-negative are formed, as shown in Figure 6 b；Wherein anchor, hard anchor-nositive
The second reference sample triple image is constituted with hard anchor-negative.Similarly, VIS-NIR is also as implied above.
(3)VIS-IDPIC：A visible images are chosen from anchor sample images and refer to sample as the 3rd
This, in theorem in Euclid space, chooses one with visible images apart from farthest identity card picture from positive sample images,
Hard anchor-positive are formed, a body closest with visible images is chosen from negative sample images
Part license piece, forms hard anchor-negative, as fig. 6 c；Wherein anchor, hard anchor-nositive
The 3rd reference sample triple image is constituted with hard anchor-negative.Similarly, IDPIC-VIS is also as implied above.
In wherein a kind of embodiment, it is trained, obtains the triple image for the condition that meets is inputted into CNN models
To in the step of training the Triplets CNN models completed, including：
Step S191, the triple image for the condition that meets is inputted multiple convolutional layers and pond layer are carried out into CNN models
Learning training, and calculate Triplets target loss functional value, in Triplets target loss functional value convergence, obtain
The Triplets CNN models completed to training.
In the present embodiment, the triple image for meeting condition in figure is inputted and multiple convolution is carried out into CNN models
The learning training of layer and pond layer, and Triplets target loss functional value is calculated, in Triplets target loss function
During value convergence, the Triplets CNN models that training is completed are obtained.Wherein the number of convolutional layer and pond layer can be according to reality
Demand during image procossing is adjusted.A pond layer can be closelyed follow behind wherein each convolutional layer to take out for local average and son
Sample, the continuous alternating between convolution and sampling, is finally exported by full articulamentum.
In wherein a kind of embodiment, as shown in fig. 7, ID Card Image is calculated according to characteristic vector respectively, face can
In the step of seeing the similarity of any two image in light image and face near infrared light image, including：
Step S131, ID Card Image, face visible images and face near infrared light image are calculated according to below equation
The similarity of middle any two image：
sim(I1,I2) expression characteristic vector is I1Image and characteristic vector be I2Image similarity, n be feature to
The dimension of amount, f1kIt is characteristic vector I1K-th of element, f2kIt is characteristic vector I2K-th of element.
Specifically, in wherein a kind of embodiment, as shown in fig. 7, visible to ID Card Image, face according to similarity
The uniformity of light image and face near infrared light image judged, and the step of export identity authentication result in, including：
Step S141, in similarity or ID Card Image and the face near-infrared of ID Card Image and face visible images
When the similarity of light image is more than default threshold value, judge that the authentication of personnel to be certified passes through.
In the present embodiment, calculating using calculating ID Card Image, face visible images and face near-infrared
The characteristic vector of light image calculates ID Card Image, face visible images and face near infrared light image between any two respectively
Similarity after, using the decision-making way of output：
If result is 1, representative capacity license piece, face visible images, the identity one of face near infrared light image
Cause, authentication is by the way that on the contrary inconsistent for 0 identity for representing three images, authentication does not pass through.Wherein T is default
Threshold value, IDPIC represents that identity card picture, NIR represent that face near infrared light image, VIS represent face visible images.
According to the identity authentication method of the invention described above, the present invention also provides a kind of identity authentication method system, under
The identity authentication method system of the present invention is described in detail with reference to accompanying drawing and preferred embodiment for face.
Figure is the structural representation of the identity authentication method system of the present invention in one embodiment.As shown in figure 8, should
Identity authentication method system in embodiment, including：
Image collection module 10, the ID Card Image, face visible images and face for obtaining personnel to be certified are near
Infrared light image.
Characteristic vector obtains module 20, for by ID Card Image, face visible images and face near infrared light image
To being previously-completed in the Triplets CNN models of training, extraction ID Card Image, face visible images and face are near for input
The convolution feature of infrared light image, obtains corresponding characteristic vector.
Similarity calculation module 30, for calculated respectively according to characteristic vector ID Card Image, face visible images and
The similarity of any two image in face near infrared light image.
Authentication judge module 40, near to ID Card Image, face visible images and face according to similarity
The uniformity of infrared light image is judged, and exports identity authentication result.
In one of the embodiments, the system of authentication, in addition to：
Quality assessment modules 50, for utilizing quality evaluation algorithm to ID Card Image, face visible images and face
Near infrared light image carries out quality evaluation.
Image pre-processing module 60, in ID Card Image, face visible images and face near infrared light image matter
When amount satisfaction requires, image preprocessing is carried out to ID Card Image, face visible images and face near infrared light image.
In one of the embodiments, the system of authentication, in addition to：
Training set triple builds module 70, for obtaining several ID Card Images, corresponding with ID Card Image in advance
Face visible images and face near infrared light image, with image, face visible images and face near-infrared in several identity
Light image builds the triple image of training set, and triple image includes reference sample image, similar sample image and foreign peoples's sample
This image.
Triple Image selection module 80 is full for being chosen according to default triple alternative condition from triple image
The triple image of sufficient condition；
Triplets CNN model training modules 90, for the triple image for the condition that meets to be inputted into CNN models
It is trained, obtains the Triplets CNN models that training is completed.
In one of the embodiments, the system of authentication, in addition to：
Similarity calculation module 30, it is additionally operable to calculate ID Card Image, face visible images and people according to below equation
The similarity of any two image in face near infrared light image：
sim(I1,I2) expression characteristic vector is I1Image and characteristic vector be I2Image similarity, n be feature to
The dimension of amount, f1kIt is characteristic vector I1K-th of element, f2kIt is characteristic vector I2K-th of element.
In one of the embodiments, the first reference sample triple image, the second reference sample triple image and
Three reference sample triple images are the triple image for the condition that meets, triple Image selection module 80, in addition to：
First reference sample triple image generation module 81, for choosing any one identity from reference sample image
Image is demonstrate,proved as the first reference sample, selection one is near apart from farthest face with the first reference sample from similar sample image
Infrared light image, and a face near infrared light image closest with the first reference sample is chosen from foreign peoples's sample image,
Generate the first reference sample triple image；
Second reference sample triple image generation module 82, for choosing any one face from reference sample image
Near infrared light image chooses one with the second reference sample distance farthest as the second reference sample from similar sample image
Face visible images, and choose a face visible ray figure closest with the first reference sample from foreign peoples's sample image
Picture, generate the second reference sample triple image；
3rd reference sample triple image generation module 83, for choosing any one Zhang Renke from reference sample image
See that light image as the 3rd reference sample, chooses one with the 3rd reference sample apart from farthest identity from similar sample image
Image is demonstrate,proved, and an ID Card Image closest with the 3rd reference sample, the ginseng of generation the 3rd are chosen from foreign peoples's sample image
Examine sample triple image.
In one of the embodiments, the system of authentication, in addition to：
Triplets CNN model training modules 90, it is additionally operable to input the triple image for the condition that meets to CNN models
It is middle to carry out the learning training of multiple convolutional layers and pond layer, and Triplets target loss functional value is calculated, in Triplets
Target loss functional value convergence when, obtain training complete Triplets CNN models.
Above-mentioned identity authentication method system can perform the identity authentication method that the embodiment of the present invention is provided, and possesses and holds
The corresponding functional module of row method and beneficial effect.As for the processing method performed by wherein each functional module, such as image
Acquisition module 10, characteristic vector obtain module 20, similarity calculation module 30, authentication judge module 40, Triplets
CNN model training modules 90 etc., the description in above method embodiment is can refer to, is no longer repeated herein.
Standby according to the identity authentication method of the invention described above and system, the present invention also provides a kind of computer-readable storage
Medium, below in conjunction with the accompanying drawings and the computer-readable recording medium of the present invention is described in detail for preferred embodiment.
Computer-readable recording medium in the embodiment of the present invention, is stored thereon with computer program, and the program is processed
Device can realize all method and steps in the inventive method embodiment when performing.
One of ordinary skill in the art will appreciate that realize all or part of flow in above-described embodiment method, being can be with
The hardware of correlation is instructed to be stored in a computer read/write memory medium come the program completed by computer program, should
Program is upon execution, it may include such as the flow of the embodiment of above-mentioned each method.Wherein, storage medium can be magnetic disc, CD, only
Read storage memory (Read-Only Memory, ROM) or random access memory (Random Access Memory, RAM)
Deng ".
Above computer readable storage medium storing program for executing is used for the journey for storing the identity authentication method that the embodiment of the present invention is provided
Sequence (instruction), wherein the identity authentication method that the embodiment of the present invention is provided can be performed by performing the program, possesses the side of execution
The corresponding beneficial effect of method.The description in above method embodiment is can refer to, is no longer repeated herein.
According to the identity authentication method and system of the invention described above, the present invention also provides a kind of device of authentication,
Below in conjunction with the accompanying drawings and the computer equipment of the present invention is described in detail for preferred embodiment.
Fig. 9 is the structural representation of the device of the authentication of the present invention in one embodiment.As shown in figure 9, the reality
Apply the device of the authentication in example, including binocular camera 901, ID Card Image harvester 902, processor 903 and
Display screen 904,；
Binocular camera 901 is used for the face visible images and face near infrared light image for gathering personnel to be certified；
ID Card Image harvester 902 is used for the ID Card Image for gathering personnel to be certified；
Processor 903 is used to perform following steps：
Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；
ID Card Image, face visible images and face near infrared light image are inputted to being previously-completed training
In Triplets CNN models, the convolution feature of extraction ID Card Image, face visible images and face near infrared light image,
Obtain corresponding characteristic vector；
Calculated respectively according to characteristic vector any in ID Card Image, face visible images and face near infrared light image
The similarity of two images；
The uniformity of ID Card Image, face visible images and face near infrared light image is sentenced according to similarity
It is disconnected, and export identity authentication result；
Display screen 904 is used to show identity authentication result.
Above-mentioned identification authentication system can utilize binocular camera shooting figure while gather the face visible ray figure of personnel to be certified
Picture and face near infrared light image, the ID Card Image of personnel to be certified is gathered using ID Card Image harvester, then will
The image processor collected, it is near to calculate ID Card Image, face visible images and face using Triplets CNN models
The similarity of any two image in infrared light image, according to similarity to ID Card Image, face visible images and face
The uniformity of near infrared light image is judged, and exports identity authentication result.The identification authentication system is easy to use, can be quick
Carry out authentication.
In a kind of specific embodiment, the device of authentication, including housing 100, it is installed in the aobvious of the side of housing 100
Display screen 200, the binocular camera 300 for being installed in the side upper end of housing 100, it is installed in the side lower end ID Card Image of housing 100
Harvester 400 and the processor (not shown) being installed in inside housing 100.Binocular is taken the photograph in the identification authentication system
As first 300 are arranged at the convenient collection to personnel's facial image to be certified in the side upper end of housing 100, ID Card Image is gathered and filled
400 settings are put with facilitating personnel to be certified to place identity card, and captured identity card image, the identity in the side lower end of housing 100
Authentication device is simple in construction, convenient use.
Further, display screen 200 is touch display screen.It is conveniently used for operating identification authentication system, such as selects people
Face image resurveys.
Each technical characteristic of above example can be combined arbitrarily, to make description succinct, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, lance is not present in the combination of these technical characteristics
Shield, all it is considered to be the scope of this specification record.
Above example only expresses the several embodiments of the present invention, and its description is more specific and detailed, but can not
Therefore it is construed as limiting the scope of the patent.It should be pointed out that for the person of ordinary skill of the art,
On the premise of not departing from present inventive concept, various modifications and improvements can be made, these belong to protection scope of the present invention.
Therefore, the protection domain of patent of the present invention should be determined by the appended claims.
Claims (10)
1. a kind of identity authentication method, it is characterised in that comprise the following steps：
Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；
By the ID Card Image, the face visible images and the face near infrared light image input to be previously-completed instruction
In experienced Triplets CNN models, the ID Card Image, the face visible images and the face near-infrared are extracted
The convolution feature of light image, obtains corresponding characteristic vector；
The ID Card Image, the face visible images and the face near-infrared are calculated according to the characteristic vector respectively
The similarity of any two image in light image；
According to the similarity to the ID Card Image, the face visible images and the face near infrared light image
Uniformity is judged, and exports identity authentication result.
2. identity authentication method according to claim 1, it is characterised in that by the ID Card Image, the people
Face visible images and the face near infrared light image are inputted to being previously-completed in the Triplets CNN models of training, are extracted
The convolution feature of the ID Card Image, the face visible images and the face near infrared light image, obtain corresponding
Before the step of characteristic vector, in addition to：
Using quality evaluation algorithm to the ID Card Image, the face visible images and the face near infrared light image
Carry out quality evaluation；
When the ID Card Image, the face visible images and the face near infrared light image quality meet to require,
Image preprocessing is carried out to the ID Card Image, the face visible images and the face near infrared light image.
3. identity authentication method according to claim 1, it is characterised in that the Triplets for being previously-completed training
CNN models are obtained by following steps：
Several ID Card Images, face visible images corresponding with the ID Card Image and face near infrared light are obtained in advance
Image, and with image, the face visible images and face near infrared light image structure training in identity several described
The triple image of collection, the triple image include reference sample image, similar sample image and foreign peoples's sample image；
The triple image for the condition that meets is chosen from the triple image according to default triple alternative condition；
The triple image for meeting condition is inputted into CNN models and is trained, obtains the Triplets that training is completed
CNN models.
4. identity authentication method according to claim 3, it is characterised in that the first reference sample triple figure
As, the second reference sample triple image and triple figure that the 3rd reference sample triple image is the condition that meets
Picture, the step of meeting the triple image of condition, is chosen from the triple image according to default triple alternative condition
In, including：
Any one ID Card Image is chosen from the reference sample image as the first reference sample, from the similar sample
One is chosen in image with first reference sample apart from farthest face near infrared light image, and from foreign peoples's sample graph
As choosing a face near infrared light image closest with first reference sample, the first reference sample triple is generated
Image；
Any one face near infrared light image is chosen from the reference sample image as the second reference sample, from described same
One is chosen in class sample image with second reference sample apart from farthest face visible images, and from foreign peoples's sample
This image chooses a face visible images closest with first reference sample, generates the second reference sample ternary
Group image；
Any one people's visible images are chosen from the reference sample image as the 3rd reference sample, from the similar sample
One is chosen in this image to select apart from farthest ID Card Image, and from foreign peoples's sample image with the 3rd reference sample
An ID Card Image closest with the 3rd reference sample is taken, generates the 3rd reference sample triple image.
5. identity authentication method according to claim 3, it is characterised in that by the triple figure for meeting condition
As input be trained into CNN models, obtain training complete Triplets CNN models the step of in, including：
The triple image for meeting condition is inputted to the study instruction that multiple convolutional layers and pond layer are carried out into CNN models
Practice, and calculate Triplets target loss functional value, in the target loss functional value convergence of the Triplets, instructed
Practice the Triplets CNN models completed.
6. identity authentication method according to claim 1, it is characterised in that calculated respectively according to the characteristic vector
The similarity of any two image in the ID Card Image, the face visible images and the face near infrared light image
The step of in, including：
Calculated according to below equation in the ID Card Image, the face visible images and the face near infrared light image
The similarity of any two image：
<mrow>
<mi>s</mi>
<mi>i</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>I</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<msub>
<mi>I</mi>
<mn>2</mn>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</msubsup>
<msub>
<mi>f</mi>
<mi>k</mi>
</msub>
<msub>
<mi>f</mi>
<mrow>
<mn>2</mn>
<mi>k</mi>
</mrow>
</msub>
</mrow>
<mrow>
<msqrt>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</msubsup>
<msup>
<msub>
<mi>f</mi>
<mrow>
<mn>1</mn>
<mi>k</mi>
</mrow>
</msub>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
<msqrt>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</msubsup>
<msup>
<msub>
<mi>f</mi>
<mrow>
<mn>2</mn>
<mi>k</mi>
</mrow>
</msub>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
</mfrac>
</mrow>
sim(I1,I2) expression characteristic vector is I1Image and characteristic vector be I2Image similarity, n is characteristic vector
Dimension, f1kIt is characteristic vector I1K-th of element, f2kIt is characteristic vector I2K-th of element.
7. the identity authentication method according to claim 1 or 6, it is characterised in that according to the similarity to described
The uniformity of ID Card Image, the face visible images and the face near infrared light image is judged, and exports body
In the step of part authentication result, including：
It is near in the ID Card Image and the similarity of the face visible images or the ID Card Image and the face
When the similarity of infrared light image is more than default threshold value, judge that the authentication of the personnel to be certified passes through.
A kind of 8. identity authentication method system, it is characterised in that including：
Image collection module, for obtaining the ID Card Image, face visible images and face near infrared light of personnel to be certified
Image；
Characteristic vector obtains module, for by the ID Card Image, the face visible images and the face near-infrared
Light image is inputted to being previously-completed in the Triplets CNN models of training, and it is visible to extract the ID Card Image, the face
Light image and the convolution feature of the face near infrared light image, obtain corresponding characteristic vector；
Similarity calculation module, for calculating the ID Card Image, the face visible ray respectively according to the characteristic vector
The similarity of any two image in image and the face near infrared light image；
Authentication judge module, for according to the similarity to the ID Card Image, the face visible images and
The uniformity of the face near infrared light image is judged, and exports identity authentication result.
9. a kind of computer-readable recording medium, is stored thereon with computer program, it is characterised in that the program is held by processor
The step of claim 1-7 methods describeds are realized during row.
10. a kind of device of authentication, it is characterised in that including display screen, binocular camera, ID Card Image harvester
And processor；
The binocular camera is used for the face visible images and face near infrared light image for gathering personnel to be certified；
The ID Card Image harvester is used for the ID Card Image for gathering personnel to be certified；
The processor is used to perform following steps：
Obtain the ID Card Image, face visible images and face near infrared light image of personnel to be certified；
By the ID Card Image, the face visible images and the face near infrared light image input to be previously-completed instruction
In experienced Triplets CNN models, the ID Card Image, the face visible images and the face near-infrared are extracted
The convolution feature of light image, obtains corresponding characteristic vector；
The ID Card Image, the face visible images and the face near-infrared are calculated according to the characteristic vector respectively
The similarity of any two image in light image；
According to the similarity to the ID Card Image, the face visible images and the face near infrared light image
Uniformity is judged, and exports identity authentication result；
The display screen is used to show the identity authentication result.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710647019.9A CN107577987A (en) | 2017-08-01 | 2017-08-01 | Identity authentication method, system and device |
| PCT/CN2018/093787 WO2019024636A1 (en) | 2017-08-01 | 2018-06-29 | Identity authentication method, system and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710647019.9A CN107577987A (en) | 2017-08-01 | 2017-08-01 | Identity authentication method, system and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107577987A true CN107577987A (en) | 2018-01-12 |
Family
ID=61035777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710647019.9A Pending CN107577987A (en) | 2017-08-01 | 2017-08-01 | Identity authentication method, system and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107577987A (en) |
| WO (1) | WO2019024636A1 (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108197563A (en) * | 2017-12-29 | 2018-06-22 | 百度在线网络技术（北京）有限公司 | For obtaining the method and device of information |
| CN108416326A (en) * | 2018-03-27 | 2018-08-17 | 百度在线网络技术（北京）有限公司 | Face identification method and device |
| CN108460366A (en) * | 2018-03-27 | 2018-08-28 | 百度在线网络技术（北京）有限公司 | Identity identifying method and device |
| CN108922542A (en) * | 2018-06-01 | 2018-11-30 | 平安科技（深圳）有限公司 | Acquisition methods, device, computer equipment and the storage medium of sample triple |
| CN108961485A (en) * | 2018-05-07 | 2018-12-07 | 金联汇通信息技术有限公司 | Intelligent door lock, auth method and device |
| CN109089052A (en) * | 2018-10-18 | 2018-12-25 | 浙江宇视科技有限公司 | A kind of verification method and device of target object |
| CN109145991A (en) * | 2018-08-24 | 2019-01-04 | 北京地平线机器人技术研发有限公司 | Image group generation method, image group generating means and electronic equipment |
| WO2019024636A1 (en) * | 2017-08-01 | 2019-02-07 | 广州广电运通金融电子股份有限公司 | Identity authentication method, system and apparatus |
| CN109325448A (en) * | 2018-09-21 | 2019-02-12 | 广州广电卓识智能科技有限公司 | Face identification method, device and computer equipment |
| CN109753934A (en) * | 2019-01-09 | 2019-05-14 | 中控智慧科技股份有限公司 | A kind of method and identification device identifying image true-false |
| WO2019192121A1 (en) * | 2018-04-04 | 2019-10-10 | 平安科技（深圳）有限公司 | Dual-channel neural network model training and human face comparison method, and terminal and medium |
| CN110765933A (en) * | 2019-10-22 | 2020-02-07 | 山西省信息产业技术研究院有限公司 | Dynamic portrait sensing comparison method applied to driver identity authentication system |
| CN112036277A (en) * | 2020-08-20 | 2020-12-04 | 浙江大华技术股份有限公司 | Face recognition method, electronic equipment and computer readable storage medium |
| CN112199975A (en) * | 2019-07-08 | 2021-01-08 | 中国移动通信集团浙江有限公司 | Identity verification method and device based on human face features |
| CN117172783A (en) * | 2023-07-17 | 2023-12-05 | 湖北盈嘉集团有限公司 | Credit and debt cross checking system for confirming account receivables |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN111008550A (en) * | 2019-09-06 | 2020-04-14 | 上海芯灵科技有限公司 | Identification method for finger vein authentication identity based on Multiple loss function |
| CN110956080B (en) * | 2019-10-14 | 2023-11-03 | 京东科技信息技术有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN113158712A (en) * | 2020-08-07 | 2021-07-23 | 西安天和防务技术股份有限公司 | Personnel management system |
| CN112699803B (en) * | 2020-12-31 | 2024-01-16 | 竹间智能科技（上海）有限公司 | Face recognition method, system, equipment and readable storage medium |
| CN114049289B (en) * | 2021-11-10 | 2024-03-05 | 合肥工业大学 | Near infrared-visible light face image synthesis method based on contrast learning and StyleGAN2 |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976356A (en) * | 2010-09-30 | 2011-02-16 | 惠州市华阳多媒体电子有限公司 | Face identification method for surfing identification policy, and identification system thereof |
| CN105023005A (en) * | 2015-08-05 | 2015-11-04 | 王丽婷 | Face recognition apparatus and recognition method thereof |
| CN105608450A (en) * | 2016-03-01 | 2016-05-25 | 天津中科智能识别产业技术研究院有限公司 | Heterogeneous face identification method based on deep convolutional neural network |
| CN106203533A (en) * | 2016-07-26 | 2016-12-07 | 厦门大学 | The degree of depth based on combined training study face verification method |
| CN106780906A (en) * | 2016-12-28 | 2017-05-31 | 北京品恩科技股份有限公司 | A kind of testimony of a witness unification recognition methods and system based on depth convolutional neural networks |
| CN106778607A (en) * | 2016-12-15 | 2017-05-31 | 国政通科技股份有限公司 | A kind of people based on recognition of face and identity card homogeneity authentication device and method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9836641B2 (en) * | 2014-12-17 | 2017-12-05 | Google Inc. | Generating numeric embeddings of images |
| CN106339695B (en) * | 2016-09-20 | 2019-11-15 | 北京小米移动软件有限公司 | Face similarity detection method, device and terminal |
| CN107577987A (en) * | 2017-08-01 | 2018-01-12 | 广州广电卓识智能科技有限公司 | Identity authentication method, system and device |
-
2017
- 2017-08-01 CN CN201710647019.9A patent/CN107577987A/en active Pending
-
2018
- 2018-06-29 WO PCT/CN2018/093787 patent/WO2019024636A1/en active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976356A (en) * | 2010-09-30 | 2011-02-16 | 惠州市华阳多媒体电子有限公司 | Face identification method for surfing identification policy, and identification system thereof |
| CN105023005A (en) * | 2015-08-05 | 2015-11-04 | 王丽婷 | Face recognition apparatus and recognition method thereof |
| CN105608450A (en) * | 2016-03-01 | 2016-05-25 | 天津中科智能识别产业技术研究院有限公司 | Heterogeneous face identification method based on deep convolutional neural network |
| CN106203533A (en) * | 2016-07-26 | 2016-12-07 | 厦门大学 | The degree of depth based on combined training study face verification method |
| CN106778607A (en) * | 2016-12-15 | 2017-05-31 | 国政通科技股份有限公司 | A kind of people based on recognition of face and identity card homogeneity authentication device and method |
| CN106780906A (en) * | 2016-12-28 | 2017-05-31 | 北京品恩科技股份有限公司 | A kind of testimony of a witness unification recognition methods and system based on depth convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| XIAOXIANG LIU等: "Transferring Deep Representation for NIR-VIS Heterogeneous Face Recognition", 《2016 INTERNATIONAL CONFERENCE ON BIOMETRICS(ICB)》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019024636A1 (en) * | 2017-08-01 | 2019-02-07 | 广州广电运通金融电子股份有限公司 | Identity authentication method, system and apparatus |
| CN108197563B (en) * | 2017-12-29 | 2022-03-11 | 百度在线网络技术（北京）有限公司 | Method and device for acquiring information |
| CN108197563A (en) * | 2017-12-29 | 2018-06-22 | 百度在线网络技术（北京）有限公司 | For obtaining the method and device of information |
| CN108416326A (en) * | 2018-03-27 | 2018-08-17 | 百度在线网络技术（北京）有限公司 | Face identification method and device |
| CN108460366A (en) * | 2018-03-27 | 2018-08-28 | 百度在线网络技术（北京）有限公司 | Identity identifying method and device |
| CN108416326B (en) * | 2018-03-27 | 2021-07-16 | 百度在线网络技术（北京）有限公司 | Face recognition method and device |
| WO2019192121A1 (en) * | 2018-04-04 | 2019-10-10 | 平安科技（深圳）有限公司 | Dual-channel neural network model training and human face comparison method, and terminal and medium |
| CN108961485A (en) * | 2018-05-07 | 2018-12-07 | 金联汇通信息技术有限公司 | Intelligent door lock, auth method and device |
| CN108922542A (en) * | 2018-06-01 | 2018-11-30 | 平安科技（深圳）有限公司 | Acquisition methods, device, computer equipment and the storage medium of sample triple |
| WO2019227613A1 (en) * | 2018-06-01 | 2019-12-05 | 平安科技（深圳）有限公司 | Method and device for obtaining triple of samples, computer device and storage medium |
| CN109145991A (en) * | 2018-08-24 | 2019-01-04 | 北京地平线机器人技术研发有限公司 | Image group generation method, image group generating means and electronic equipment |
| CN109145991B (en) * | 2018-08-24 | 2020-07-31 | 北京地平线机器人技术研发有限公司 | Image group generation method, image group generation device and electronic equipment |
| CN109325448A (en) * | 2018-09-21 | 2019-02-12 | 广州广电卓识智能科技有限公司 | Face identification method, device and computer equipment |
| CN109089052A (en) * | 2018-10-18 | 2018-12-25 | 浙江宇视科技有限公司 | A kind of verification method and device of target object |
| CN109089052B (en) * | 2018-10-18 | 2020-09-01 | 浙江宇视科技有限公司 | Target object verification method and device |
| CN109753934A (en) * | 2019-01-09 | 2019-05-14 | 中控智慧科技股份有限公司 | A kind of method and identification device identifying image true-false |
| CN112199975A (en) * | 2019-07-08 | 2021-01-08 | 中国移动通信集团浙江有限公司 | Identity verification method and device based on human face features |
| CN110765933A (en) * | 2019-10-22 | 2020-02-07 | 山西省信息产业技术研究院有限公司 | Dynamic portrait sensing comparison method applied to driver identity authentication system |
| CN112036277A (en) * | 2020-08-20 | 2020-12-04 | 浙江大华技术股份有限公司 | Face recognition method, electronic equipment and computer readable storage medium |
| CN112036277B (en) * | 2020-08-20 | 2023-09-29 | 浙江大华技术股份有限公司 | Face recognition method, electronic equipment and computer readable storage medium |
| CN117172783A (en) * | 2023-07-17 | 2023-12-05 | 湖北盈嘉集团有限公司 | Credit and debt cross checking system for confirming account receivables |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019024636A1 (en) | 2019-02-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107577987A (en) | Identity authentication method, system and device | |
| CN106503687B (en) | Merge the monitor video system for identifying figures and its method of face multi-angle feature | |
| CN109522819B (en) | Fire image identification method based on deep learning | |
| CN107066942A (en) | A kind of living body faces recognition methods and system | |
| Xie et al. | Background learning based on target suppression constraint for hyperspectral target detection | |
| WO2019134536A1 (en) | Neural network model-based human face living body detection | |
| CN106446872A (en) | Detection and recognition method of human face in video under low-light conditions | |
| Fang et al. | A visual attention model combining top-down and bottom-up mechanisms for salient object detection | |
| CN103793690A (en) | Human body biotic living body detection method based on subcutaneous bloodstream detection and application | |
| CN110032915A (en) | A kind of human face in-vivo detection method, device and electronic equipment | |
| CN106295501A (en) | The degree of depth based on lip movement study personal identification method | |
| CN111222564B (en) | Image identification system, method and device based on image channel correlation | |
| CN107798279A (en) | Face living body detection method and device | |
| CN109255319A (en) | For the recognition of face payment information method for anti-counterfeit of still photo | |
| CN108664839A (en) | A kind of image processing method and equipment | |
| CN109190456A (en) | Pedestrian detection method is overlooked based on the multiple features fusion of converging channels feature and gray level co-occurrence matrixes | |
| CN103593648A (en) | Face recognition method for open environment | |
| CN112668557A (en) | Method for defending image noise attack in pedestrian re-identification system | |
| CN115131880A (en) | Multi-scale attention fusion double-supervision human face in-vivo detection method | |
| Ru et al. | Bita-net: Bi-temporal attention network for facial video forgery detection | |
| Pal et al. | Implementation of hand vein structure authentication based system | |
| Cai et al. | Image fusion of palmprint and palm vein: Multispectral palm image fusion | |
| CN112052730B (en) | 3D dynamic portrait identification monitoring equipment and method | |
| Manssor et al. | TIRFaceNet: thermal IR facial recognition | |
| CN112861588A (en) | Living body detection method and device |
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 |
Application publication date: 20180112 |
|
| RJ01 | Rejection of invention patent application after publication |