CN106778519A - A kind of face verification method by matching user identity certificate and take pictures certainly - Google Patents
A kind of face verification method by matching user identity certificate and take pictures certainly Download PDFInfo
- Publication number
- CN106778519A CN106778519A CN201611054308.XA CN201611054308A CN106778519A CN 106778519 A CN106778519 A CN 106778519A CN 201611054308 A CN201611054308 A CN 201611054308A CN 106778519 A CN106778519 A CN 106778519A
- Authority
- CN
- China
- Prior art keywords
- feature
- user identity
- pictures certainly
- identity certificate
- classification
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/513—Sparse representations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Collating Specific Patterns (AREA)
- Image Analysis (AREA)
Abstract
A kind of face verification method by matching user identity certificate and take pictures certainly proposed in the present invention, its main contents include：Import user identity certificate and take pictures certainly, face detection and cutting, image enhaucament, feature extraction, characteristic normal, feature merge, classification, the result, its process is, guide's access customer identity document and take pictures certainly, carry out face detection and be cropped to appropriate pixels region, then image enhaucament is carried out to mitigate the serious illumination effects produced with noise, feature extraction is carried out again, characteristic normal and feature merge to calculate two similitudes of vector, classified using SVMs (SVM) linear discriminant function, finally obtained the result.The present invention reduces the influence of illumination variation and domain transfer by strengthening image with different technologies；Easy to operate, recall rate is high；The time is saved, man power and material facilitates user.
Description
Technical field
The present invention relates to field of face identification, more particularly, to a kind of by matching user identity certificate and taking pictures certainly
Face verification method.
Background technology
With the development of modern science and technology, application of the face recognition in fields such as business, security systems is more and more extensive.
For example, when company needs to verify the information of client, can be by contrasting the identity document information and use of client reservation
The auto heterodyne that family sends over shines into capable contrast, so as to verify the identity of client.Meanwhile, it can be also used in security system, and
Can be combined with other biological recognition systems, such as fingerprint, iris and voice etc..Traditional method generally needs artificial contrast
The certificate of user and user, required time are long, also consume manpower and materials；User is also required to scene and is just able to verify that, non-
It is often inconvenient.
The present invention proposes a kind of face verification method by matching user identity certificate and take pictures certainly, guide's access customer
Identity document and take pictures certainly, carry out face detection and be cropped to appropriate pixels region, then carry out image enhaucament with mitigate with
Noise and the serious illumination effects that produce, then feature extraction is carried out, characteristic normal and feature merge to calculate two vectors
Similitude, classified using SVMs (SVM) linear discriminant function, finally obtain the result.The present invention passes through
Strengthen image with different technologies, reduce the influence of illumination variation and domain transfer；Easy to operate, recall rate is high；Save
Time, man power and material facilitates user.
The content of the invention
For the problem that traditional verification method is inconvenient, it is an object of the invention to provide one kind by matching user identity
Certificate and the face verification method taken pictures certainly, guide's access customer identity document and take pictures certainly, carry out face detection and are cropped to conjunction
Suitable pixel region, then carries out image enhaucament to mitigate the serious illumination effects produced with noise, then carries out feature to carry
Take, characteristic normal and feature merge to calculate two similitudes of vector, using SVMs (SVM) linear discriminant function
Classified, finally obtained the result.
To solve the above problems, the present invention provides a kind of face verification side by matching user identity certificate and take pictures certainly
Method, its main contents include：
(1) import user identity certificate and take pictures certainly；
(2) face detection and cutting；
(3) image enhaucament；
(4) feature extraction；
(5) characteristic normal；
(6) feature merges；
(7) classify；
(8) the result.
Wherein, described importing user identity certificate and take pictures certainly, during identifying user identity, extractable user reserves in advance
Identity document information and taking pictures certainly of sending over of user, by face verification, confirm the identity of user.
Wherein, described face detection and cutting, a conventional detection include the outside of ear, chin or hair, detection
Region expands；Geometrical normalization is carried out to eyes coordinates, eyes are positioned in zero angle by Plane Rotation；Finally, using two-wire
Property interpolation, facial image is simplified to 224 × 224 pixel regions.
Wherein, described image enhaucament, in order to mitigate the serious illumination effects produced with noise, we have evaluated
Three kinds of algorithms below：
The first is the color constancy Retinex theories of view-based access control model, that is, think white with human visual system most
Auger signal is related, and it can reduce the color intensity change in different regions；
Second is Automatic Color Equalization Algorithm (ACE), and it is the human visual system based on unified global and local influence
Computation model, achieves good contrast enhancing, approximate two different input picture sources；
The third is Contrast-limited adaptive histogram equalization (CLAHE), the image block that it will be input into, and each piece is applicable
In traditional histogram equalization, then check whether histogram exceedes contrast and limit；
Present invention employs Automatic Color Equalization Algorithm (ACE).
Wherein, described feature extraction, using the works of the shift learning technology based on CNN, extracts third layer to finally
One layer of feature, referred to as fc6, there is 4096 dimensions；
Activation functions after fc6 are to correct linear unit (ReLU), and (0, x), these layers are often to be defined as f (x)=max
Sparse output can be produced, important information may finally be lost；The characteristics of independently fc6 being analyzed before activation primitive；However, by
When in training, network never occurs negative value, it is impossible to assuming that whether non-sparse features may be useful for or simply as random value；
It should be noted that feature is extracted from identity and auto heterodyne image respectively and the Output Size that activation does not change this layer is removed.
Wherein, described characteristic normal, because cross-domain (isomery source) is set, the function of being extracted in each domain may have
Significant different amplitude range；Consider a characteristic vector p-norm,ByBe given；L1 returns
One change characteristic vector byBe given due to these standardized techniques according to scalar simply by original feature vector, it
Maintain it is original openness.
Wherein, described feature merges, and a and b is respectively identity document and the d dimensional feature vectors taken pictures certainly, and we assess
Four kinds of methods calculate two similitudes of vector, and keep original dimension d in the same time；These four analytical technologies,
Last feature is generated to f：
Subtraction absolute value：F=| a-b |；
Element-Level multiplication：
Correlation：0 is set in the index or not scope；
Phase is related：F=IDFT (G/ ‖ G ‖2), wherein G=DFT (a) ° DFT (b), IDFT is inverse discrete Fourier transform,
DFT is discrete Fourier transform；
When vectorial a and b are similar, the absolute value of subtraction should produce less function, and the present invention is absolute using subtraction
The method of value.
Wherein, described classification, in sorting phase, using SVMs (SVM) linear discriminant function.
Further, described SVMs (SVM), principle is：
If linear separability sample set and be (xi,yi), i=1 ..., n, x ∈ Rd, y ∈ {+1, -1 } are category labels, then
Wx+b=0
It is the classifying face equation of SVM classifier；
Classification when, in order that classifying face to all samples correctly classification and class interval reach maximum, it is necessary to meet under
Two, face condition：
Φ (x)=min (wTw)
yi(w·xi+b)-1≥0
Optimal classification surface is can be obtained by by solving this constrained optimization problem, and crosses nearest from classifying face in two class samples
Point and those special samples that equal sign is set up in formula are just so that parallel to the training sample on the hyperplane of optimal classification surface,
Because they support optimal classification surface, therefore are referred to as support vector；Fusion output is input to SVM points as characteristic vector
Among class device, final classification result is obtained.
Wherein, described the result, classification results are obtained from grader, and whether display user identity certificate takes pictures with certainly
Matching.
Brief description of the drawings
Fig. 1 is a kind of system flow of face verification method by matching user identity certificate and taking pictures certainly of the invention
Figure.
Fig. 2 is that a kind of flow by the face verification method for matching user identity certificate and taking pictures certainly of the present invention is illustrated
Figure.
Fig. 3 is that a kind of three kinds of images by the face verification method for matching user identity certificate and taking pictures certainly of the present invention increase
The Contrast on effect of strong algorithms.
Fig. 4 is a kind of personnel identity core of face verification method by matching user identity certificate and taking pictures certainly of the invention
Check system interface.
Specific embodiment
It should be noted that in the case where not conflicting, the feature in embodiment and embodiment in the application can phase
Mutually combine, the present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings.
Fig. 1 is a kind of system flow of face verification method by matching user identity certificate and taking pictures certainly of the invention
Figure.It is main to include importing user identity certificate and take pictures certainly, face detection and cutting, image enhaucament, feature extraction, characteristic normal
Change, feature merges, classify and the result.
Wherein, import user identity certificate and take pictures certainly, during identifying user identity, can extract user's identity reserved in advance
What certificate information and user sended over takes pictures certainly, by face verification, confirms the identity of user.
Wherein, face detection and cutting, a conventional detection include the outside of ear, chin or hair, and detection zone expands
Greatly；Geometrical normalization is carried out to eyes coordinates, eyes are positioned in zero angle by Plane Rotation；Finally, using bilinear interpolation,
Facial image is simplified to 224 × 224 pixel regions.
Wherein, feature extraction, using the works of the shift learning technology based on CNN, extracts third layer to last layer
Feature, referred to as fc6, there is 4096 dimensions；
Activation functions after fc6 are to correct linear unit (ReLU), and (0, x), these layers are often to be defined as f (x)=max
Sparse output can be produced, important information may finally be lost；The characteristics of independently fc6 being analyzed before activation primitive；However, by
When in training, network never occurs negative value, it is impossible to assuming that whether non-sparse features may be useful for or simply as random value；
It should be noted that feature is extracted from identity and auto heterodyne image respectively and the Output Size that activation does not change this layer is removed.
Wherein, characteristic normal, due to cross-domain (isomery source) set, in each domain extract function may have it is significant not
Same amplitude range；Consider a characteristic vector p-norm,ByBe given；L1 normalization is special
Levy vector byBe given due to these standardized techniques according to scalar simply by original feature vector, they keep
Original is openness.
Wherein, feature merges, and a and b is respectively identity document and the d dimensional feature vectors taken pictures certainly, and we have evaluated four kinds
Method calculates two similitudes of vector, and in the same time keeps original dimension d；These four analytical technologies, generate
Last feature is to f：
Subtraction absolute value：F=| a-b |；
Element-Level multiplication：
Correlation：0 is set in the index or not scope；
Phase is related：F=IDFT (G/ ‖ G ‖2), wherein G=DFT (a) ° DFT (b), IDFT is inverse discrete Fourier transform,
DFT is discrete Fourier transform；
When vectorial a and b are similar, the absolute value of subtraction should produce less function, and the present invention is absolute using subtraction
The method of value.
Wherein, classify, in sorting phase, using SVMs (SVM) linear discriminant function.
The principle of SVMs (SVM) is：
If linear separability sample set and be (xi,yi), i=1 ..., n, x ∈ Rd, y ∈ {+1, -1 } are category labels, then
Wx+b=0
It is the classifying face equation of SVM classifier；
Classification when, in order that classifying face to all samples correctly classification and class interval reach maximum, it is necessary to meet under
Two, face condition：
Φ (x)=min (wTw)
yi(w·xi+b)-1≥0
Optimal classification surface is can be obtained by by solving this constrained optimization problem, and crosses nearest from classifying face in two class samples
Point and those special samples that equal sign is set up in formula are just so that parallel to the training sample on the hyperplane of optimal classification surface,
Because they support optimal classification surface, therefore are referred to as support vector；Fusion output is input to SVM points as characteristic vector
Among class device, final classification result is obtained.
Wherein, the result, classification results are obtained from grader, and whether display user identity certificate matches with from taking pictures.
Fig. 2 is that a kind of flow by the face verification method for matching user identity certificate and taking pictures certainly of the present invention is illustrated
Figure.Guide's access customer identity document and take pictures certainly, carry out face detection and be cropped to appropriate pixels region, then carry out image increasing
It is strong then to carry out feature extraction mitigating the serious illumination effects produced with noise, characteristic normal and feature merge with
Two similitudes of vector are calculated, is classified using SVMs (SVM) linear discriminant function, finally obtain checking knot
Really.
Fig. 3 is that a kind of three kinds of images by the face verification method for matching user identity certificate and taking pictures certainly of the present invention increase
The Contrast on effect of strong algorithms.In order to mitigate the serious illumination effects produced with noise, we have evaluated following three kinds of calculations
Method：
The first is the color constancy Retinex theories of view-based access control model, that is, think white with human visual system most
Auger signal is related, and it can reduce the color intensity change in different regions；
Second is Automatic Color Equalization Algorithm (ACE), and it is the human visual system based on unified global and local influence
Computation model, achieves good contrast enhancing, approximate two different input picture sources；
The third is Contrast-limited adaptive histogram equalization (CLAHE), the image block that it will be input into, and each piece is applicable
In traditional histogram equalization, then check whether histogram exceedes contrast and limit；
Present invention employs Automatic Color Equalization Algorithm (ACE).
Fig. 4 is a kind of personnel identity core of face verification method by matching user identity certificate and taking pictures certainly of the invention
Check system interface.Its process is：User's brush identity card, imports the information and photograph of identity document in systems, and user shoots certainly
After taking pictures, system carries out portrait contrast, if user is consistent with identity document photo from taking pictures, completes checking.
For those skilled in the art, the present invention is not restricted to the details of above-described embodiment, without departing substantially from essence of the invention
In the case of god and scope, the present invention can be realized with other concrete forms.Additionally, those skilled in the art can be to this hair
Bright to carry out various changes and modification without departing from the spirit and scope of the present invention, these improvement also should be regarded as of the invention with modification
Protection domain.Therefore, appended claims are intended to be construed to include preferred embodiment and fall into all changes of the scope of the invention
More and modification.
Claims (10)
1. a kind of face verification method by matching user identity certificate and take pictures certainly, it is characterised in that main to include importing
User identity certificate and (one) is taken pictures certainly；Face detection and cutting (two)；Image enhaucament (three)；Feature extraction (four)；Feature is just
Ruleization (five)；Feature merges (six)；Classification (seven)；The result (eight).
2. it is based on the importing user identity certificate described in claims 1 and takes pictures certainly (one), it is characterised in that identifying user body
During part, what the identity document information and user that extractable user reserves in advance were sended over takes pictures certainly, by face verification, confirms
The identity of user.
3. based on the face detection described in claims 1 and cutting (two), it is characterised in that a conventional detection includes ear
Piece, the outside of chin or hair, detection zone expands；Geometrical normalization is carried out to eyes coordinates, by Plane Rotation in zero degree
Angle positions eyes；Finally, using bilinear interpolation, facial image is simplified to 224 × 224 pixel regions.
4. based on the image enhaucament (three) described in claims 1, it is characterised in that in order to mitigate produced with noise tight
The illumination effects of weight, we have evaluated following three kinds of algorithms：
The first is the color constancy Retinex theories of view-based access control model, that is, think the most auger of white and human visual system
Signal is related, and it can reduce the color intensity change in different regions；
Second is Automatic Color Equalization Algorithm (ACE), and it is the calculating of the human visual system based on unified global and local influence
Model, achieves good contrast enhancing, approximate two different input picture sources；
The third is Contrast-limited adaptive histogram equalization (CLAHE), the image block that it will be input into, and each piece is applied to biography
The histogram equalization of system, then checks whether histogram exceedes contrast and limit；
Present invention employs Automatic Color Equalization Algorithm (ACE).
5. based on the feature extraction (four) described in claims 1, it is characterised in that use the shift learning technology based on CNN
Works, extract third layer to the feature of last layer, referred to as fc6 has 4096 dimensions；
Activation functions after fc6 are to correct linear unit (ReLU), and (0, x), these layers are often produced to be defined as f (x)=max
The sparse output of life, may finally lose important information；The characteristics of independently fc6 being analyzed before activation primitive；However, due to
During training, network never occurs negative value, it is impossible to assuming that whether non-sparse features may be useful for or simply as random value；Need
It is noted that feature is extracted from identity and auto heterodyne image respectively and the Output Size that activation does not change this layer is removed.
6. based on the characteristic normal (five) described in claims 1, it is characterised in that because cross-domain (isomery source) is set,
The function of being extracted in each domain may have significant different amplitude range；Consider a characteristic vector p-norm,ByBe given；L1 normalization characteristics vector byBe given due to these standardized technique roots
According to scalar simply by original feature vector, they maintain original openness.
7. (six) are merged based on the feature described in claims 1, it is characterised in that a and b are respectively identity documents and take pictures certainly
D dimensional feature vectors, we have evaluated four kinds of methods to calculate two similitudes of vector, and keep original in the same time
Dimension d；These four analytical technologies, generate last feature to f：
Subtraction absolute value：F=| a-b |；
Element-Level multiplication：
Correlation：0 is set in the index or not scope；
Phase is related：F=IDFT (G/ ‖ G ‖2), whereinIDFT is inverse discrete Fourier transform, DFT
It is discrete Fourier transform；
When vectorial a and b are similar, the absolute value of subtraction should produce less function, and the present invention is using subtraction absolute value
Method.
8. based on the classification (seven) described in claims 1, it is characterised in that in sorting phase, using SVMs (SVM)
Linear discriminant function.
9. based on the SVMs (SVM) described in claims 8, it is characterised in that principle is：
If linear separability sample set and be (xi,yi), i=1 ..., n, x ∈ Rd, y ∈ {+1, -1 } are category labels, then
Wx+b=0
It is the classifying face equation of SVM classifier；
In classification, in order that classifying face is correctly classified by all samples and class interval reaches maximum, it is necessary to meet following two
Individual condition：
Φ (x)=min (wTw)
yi(w·xi+b)-1≥0
Can be obtained by optimal classification surface by solving this constrained optimization problem, and cross point nearest from classifying face in two class samples and
Those special samples of equal sign establishment in formula are just so that parallel to the training sample on the hyperplane of optimal classification surface, because
They support optimal classification surface, therefore are referred to as support vector；Fusion output is input to SVM classifier as characteristic vector
Among, obtain final classification result.
10., based on the result (eight) described in claims 1, it is characterised in that obtain classification results from grader, show
Whether user identity certificate matches with from taking pictures.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611054308.XA CN106778519A (en) | 2016-11-25 | 2016-11-25 | A kind of face verification method by matching user identity certificate and take pictures certainly |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611054308.XA CN106778519A (en) | 2016-11-25 | 2016-11-25 | A kind of face verification method by matching user identity certificate and take pictures certainly |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106778519A true CN106778519A (en) | 2017-05-31 |
Family
ID=58910746
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611054308.XA Withdrawn CN106778519A (en) | 2016-11-25 | 2016-11-25 | A kind of face verification method by matching user identity certificate and take pictures certainly |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106778519A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110020519A (en) * | 2019-01-08 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of identity checking method, device and electronic equipment |
| CN111291780A (en) * | 2018-12-07 | 2020-06-16 | 深圳光启空间技术有限公司 | Cross-domain network training and image recognition method |
| CN112418009A (en) * | 2020-11-06 | 2021-02-26 | 中保车服科技服务股份有限公司 | Image quality detection method, terminal device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680131A (en) * | 2015-01-29 | 2015-06-03 | 深圳云天励飞技术有限公司 | Identity authentication method based on identity certificate information and human face multi-feature recognition |
| CN104881648A (en) * | 2015-05-28 | 2015-09-02 | 李宏森 | Fingerprint verification system arranged in valid identity card |
| CN105138980A (en) * | 2015-08-20 | 2015-12-09 | 深圳云天励飞技术有限公司 | Identify authentication method and system based on identity card information and face identification |
| CN105956578A (en) * | 2016-05-23 | 2016-09-21 | 深圳华中科技大学研究院 | Face verification method based on identity document information |
-
2016
- 2016-11-25 CN CN201611054308.XA patent/CN106778519A/en not_active Withdrawn
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104680131A (en) * | 2015-01-29 | 2015-06-03 | 深圳云天励飞技术有限公司 | Identity authentication method based on identity certificate information and human face multi-feature recognition |
| CN104881648A (en) * | 2015-05-28 | 2015-09-02 | 李宏森 | Fingerprint verification system arranged in valid identity card |
| CN105138980A (en) * | 2015-08-20 | 2015-12-09 | 深圳云天励飞技术有限公司 | Identify authentication method and system based on identity card information and face identification |
| CN105956578A (en) * | 2016-05-23 | 2016-09-21 | 深圳华中科技大学研究院 | Face verification method based on identity document information |
Non-Patent Citations (1)
| Title |
|---|
| GUILHERME FOLEGO等: ""Cross-Domain Face Verification: Matching ID Document and Self-Portrait Photographs"", 《网页在线公开:HTTPS://ARXIV.ORG/ABS/1611.05755》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111291780A (en) * | 2018-12-07 | 2020-06-16 | 深圳光启空间技术有限公司 | Cross-domain network training and image recognition method |
| CN110020519A (en) * | 2019-01-08 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of identity checking method, device and electronic equipment |
| CN112418009A (en) * | 2020-11-06 | 2021-02-26 | 中保车服科技服务股份有限公司 | Image quality detection method, terminal device and storage medium |
| CN112418009B (en) * | 2020-11-06 | 2024-03-22 | 中保车服科技服务股份有限公司 | Image quality detection method, terminal equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11354917B2 (en) | Detection of fraudulently generated and photocopied credential documents | |
| JP7165746B2 (en) | ID authentication method and device, electronic device and storage medium | |
| KR102324706B1 (en) | Face recognition unlock method and device, device, medium | |
| Nabatchian et al. | Illumination invariant feature extraction and mutual-information-based local matching for face recognition under illumination variation and occlusion | |
| US20180034852A1 (en) | Anti-spoofing system and methods useful in conjunction therewith | |
| Gill et al. | A review paper on digital image forgery detection techniques | |
| CN112052831B (en) | Method, device and computer storage medium for face detection | |
| WO2022089124A1 (en) | Certificate authenticity identification method and apparatus, computer-readable medium, and electronic device | |
| CN112052830B (en) | Method, device and computer storage medium for face detection | |
| Liu et al. | Detect image splicing with artificial blurred boundary | |
| Deep Kaur et al. | An analysis of image forgery detection techniques | |
| Niyishaka et al. | Image splicing detection technique based on Illumination-Reflectance model and LBP | |
| CN106778519A (en) | A kind of face verification method by matching user identity certificate and take pictures certainly | |
| CN112052832A (en) | Face detection method, device and computer storage medium | |
| Mushtaq et al. | Image copy move forgery detection: a review | |
| Galdi et al. | Secure user authentication on smartphones via sensor and face recognition on short video clips | |
| Kaur et al. | Illumination invariant face recognition | |
| Teja | Real-time live face detection using face template matching and DCT energy analysis | |
| CN111191549A (en) | Two-stage face anti-counterfeiting detection method | |
| Mr et al. | Developing a novel technique to match composite sketches with images captured by unmanned aerial vehicle | |
| Bala et al. | An Illumination Insensitive Normalization Approach to Face Recognition Using Locality Sensitive Discriminant Analysis. | |
| Cheng et al. | Illumination normalization based on different smoothing filters quotient image | |
| Berriche | Comparative Study of Fingerprint‐Based Gender Identification | |
| Vidyadharan et al. | Brightness distribution based image tampering detection | |
| Rani et al. | Digital image forensics-image verification techniques |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170531 |
|
| WW01 | Invention patent application withdrawn after publication |