CN103971096B - A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram - Google Patents

A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram Download PDF

Info

Publication number
CN103971096B
CN103971096B CN201410195919.0A CN201410195919A CN103971096B CN 103971096 B CN103971096 B CN 103971096B CN 201410195919 A CN201410195919 A CN 201410195919A CN 103971096 B CN103971096 B CN 103971096B
Authority
CN
China
Prior art keywords
face
image
msub
mrow
pose
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.)
Active
Application number
CN201410195919.0A
Other languages
Chinese (zh)
Other versions
CN103971096A (en
Inventor
王科俊
胡金裕
安晓童
邹国锋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Engineering University
Original Assignee
Harbin Engineering University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Harbin Engineering University filed Critical Harbin Engineering University
Priority to CN201410195919.0A priority Critical patent/CN103971096B/en
Publication of CN103971096A publication Critical patent/CN103971096A/en
Application granted granted Critical
Publication of CN103971096B publication Critical patent/CN103971096B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Image Analysis (AREA)
  • Image Processing (AREA)
  • Collating Specific Patterns (AREA)

Abstract

The present invention provides a kind of Pose-varied face recognition method based on MB LBP features and face energy diagram.The present invention trains storehouse by establishing Face Image with Pose Variations, after facial image is carried out into size normalized, the face average power figure and variance energy diagram in structure training storehouse；MB LBP feature extractions are carried out to resulting face average power figure and variance energy diagram again, and stored as matching library information；When carrying out Face datection, detect facial image and extract human face region, and size normalized is done to human face region image, obtain standard faces image；MB LBP feature extractions are carried out to standard faces image；The Classification and Identification of multi-pose Face is finally completed using nearest neighbor classifier.The present invention can preferably retain the intrinsic macroscopic features of multi-pose Face, and remain facial image pattern microstructure and macrostructure, can remove influences caused by single pixel noise, and required memory space is small, has excellent discrimination and recognition speed.

Description

A kind of Pose-varied face recognition method based on MB-LBP features and face energy diagram
Technical field
The invention belongs to biological characteristics identity recognizing technology field, and MB-LBP features and face are based on more particularly to one kind The Pose-varied face recognition method of energy diagram.
Background technology
Automatic face recognition technology has collection convenient, non-relative to other biological feather recognition methods such as fingerprint, iris The peculiar advantages such as invasion property, thus there is very extensive application prospect and economic value.If recognition of face is entered by posture Row division, can be divided into forward sight recognition of face and Pose-varied face recognition.Wherein, the technology of forward sight recognition of face more into It is ripe.And Pose-varied face recognition method still has many technical problems such as amount of storage is big, calculating is complicated, discrimination is low.Multi-pose The research hysteresis of recognition of face, turn into restriction face recognition technology and really obtain one of major obstacle of practical application.Therefore, enter Popularization of the research of row Pose-varied face recognition to face recognition technology is significant.
Existing Pose-varied face recognition, typical document therein is patent《One kind is based on face average and variance energy The Pose-varied face recognition method of figure》(the Chinese invention patents such as Wang Kejun, Zou Guofeng：201310122161.3[P].2013- 07-24.), face energy diagram is built using the facial image of different postures to be used to realize Pose-varied face recognition.But face energy Figure does not possess the periodicity of gait energy diagram, can not represent the human face posture change of different luffing angles and different swing angles.
Liao et al. proposes (Liao S C, et al.Learning multi-scale block local binary patterns for face recognition.In Proceedings of the2007International Conference on Biometrics.Seoul,South Korea:Springer, 2007.828-837.) based on multiple dimensioned The face identification method of local binary patterns (Multi-scale Block Local Binary Pattens, MB-LBP).To the greatest extent Pipe MB-LBP feature extracting methods achieve good effect in texture analysis and face recognition application experiment, but in illumination In the case of the influences of complex factors such as change is violent, image-forming condition extreme variation, posture, expression, age, the table of MB-LBP features Sign ability and classification capacity are also restrained, and recognition performance drastically declines.
The present invention is by the way that the method for MB-LBP features and face energy diagram is combined, using MB-LBP feature extraction Method carries out Further Feature Extraction to face average power figure and variance energy diagram, while reducing computation complexity, goes again Except redundancy, then again by Classification and Identification of the image texture characteristic extracted for multi-pose Face.
The content of the invention
It is an object of the invention to provide one kind can effectively extract pitching change and situation of change human face of vacillating now to the left, now to the right Key message, while required memory space is small, and computation complexity is low, and discrimination and the high one kind of recognition speed are based on MB- The Pose-varied face recognition method of LBP features and face energy diagram.
The object of the present invention is achieved like this：
(1) Face Image with Pose Variations training storehouse is established, and Face Image with Pose Variations is trained into the face images in storehouse Carry out size normalization；
(2) according to the pitching excursion of human face region image luffing angle different demarcation face, model is changed with reference to pitching Enclose structure narrow sense face average power figure and narrow sense face variance energy diagram, the primary features as Pose-varied face recognition；
Involved narrow sense face average power figure FkThe expression formula of (x, y) is：
In formula, MkThe sum of image, I when representing same luffing angle scope, angle change of vacillating now to the left, now to the rightj(x, y) is colourful State Gray Face image, k represent different luffing angle scopes, and k=1 represents to look up, and k=2 represents to look squarely, and k=3 represents to bow Depending on j represents the image of j-th of angle change of vacillating now to the left, now to the right, and x, y represent two dimensional image plane coordinate；
Involved narrow sense face variance energy diagram DkThe expression formula of (x, y) is：
(3) using MB-LBP algorithms to the narrow sense face average power figure and narrow sense face variance energy that are obtained in step (2) Spirogram carries out Further Feature Extraction, stores the MB-LBP characteristic informations for Classification and Identification；
Involved MB-LBP features are represented by：
Wherein
gkRepresent the gray value of single pixel；B represents the average gray value of nth pixel block；
(4) Face Image with Pose Variations to be detected is read, human face region is detected and extracts face；
(5) human face region extracted is subjected to size normalized, obtains standard faces training image；
(6) MB-LBP feature extractions are carried out to standard faces training image, and store extracted MB-LBP characteristic informations；
(7) for being obtained in step (6) in the MB-LBP characteristic informations of the standard faces image to be detected of acquisition and step (3) The MB-LBP characteristic informations of the training storehouse face energy diagram taken, classification knowledge is carried out by the nearest neighbor classifier based on Euclidean distance Not, face recognition result is finally exported.
The beneficial effects of the present invention are：
Face energy diagram has merged the integrated information of several facial images, can not only save memory space well, drop Low computation complexity, and the noise jamming occurred in single-frame images can be weakened, face energy diagram has contained under many attitude Facial contour information, the recognition of face being advantageously implemented under wide-angle attitudes vibration.The present invention is by MB-LBP operators to people Face energy diagram and the standard faces image detected carry out feature extraction, for Classification and Identification.The MB-LBP feature energy extracted It is enough to retain the intrinsic macroscopic features of multi-pose Face well, and facial image pattern microstructure and macrostructure are contained, And can remove and influence caused by single pixel noise, also make discrimination and recognition speed has and is obviously improved, improve The combination property of Pose-varied face recognition.
Brief description of the drawings
Fig. 1 is the Pose-varied face recognition flow chart based on MB-LBP features and face energy diagram；
Fig. 2 is variation diagram of the face in three dimensions；
Fig. 3 is Face Image with Pose Variations and face average power figure；
Fig. 4 is Face Image with Pose Variations and face variance energy diagram；
Fig. 5 is the original image in face database；
Fig. 6 is the facial image after normalization；
Fig. 7 is part facial image in test facial image database；
Fig. 8 is the method for expressing figure of MB-LBP features；
Fig. 9 is the texture image after face average power figure and variance energy diagram MB-LBP；
Figure 10 is the texture image after standard faces image MB-LBP；
Figure 11 is to combine MB-LBP features and the method for face energy diagram and the recognition effect contrast table of other method.
Embodiment
Illustrate below in conjunction with the accompanying drawings and the present invention is described in more detail：
A kind of Pose-varied face recognition method based on MB-LBP features and face energy diagram, first by multi-pose Face figure As the face images progress size normalization in database, face average power figure and variance energy are further built into Figure；, then MB-LBP feature extractions are carried out to face average power figure and variance energy diagram；Then, multi-pose people to be detected is read Face image, and human face region detection is carried out to facial image based on Adaboost algorithm, do size normalizing after extracting human face region Change, obtain standard faces image；The standard faces image obtained again using MB-LBP feature extracting method to test library is carried out Feature extraction；Finally, recognition of face is completed by nearest neighbor classifier.
1. the size normalization of multi-pose Face training storehouse image
Present invention experiment has used institute of computing technology of Chinese Academy of Sciences CAS-PEAL-R1 to share face image data Storehouse, based on the non-frontal facial image subset in face database, including the image of 1040 people, it is chosen in experiment In 50 people image, everyone includes 21 kinds of different attitudes vibrations.First, from the face under everyone every kind of pitching situation of change 6 width are chosen, totally 18 width, structure obtains training face database, totally 50 × 18=900 width image.Then, size is carried out to all images Normalize, the size of human face region is uniformly normalized to 230 × 270 pixels in the present invention.
2. build face average power figure and variance energy diagram
Training storehouse image after normalization is respectively used to build the narrow sense face average energy under three kinds of pitching situations of change Spirogram and variance energy diagram.
2.1 narrow sense face average power figures
Angle is looked squarely as zero degree using face, the maximum luffing angle that face can occur is -45 ° and 60 °, ordinary circumstance The luffing angle of human face is distributed between [- 30 °, 30 °].The present invention is by luffing angle excursion between [- 5 °, 5 °] Facial image is defined as looking squarely scope facial image, and the facial image between [5 °, 30 °] is defined as looking up scope face figure Picture, the facial image between [- 30 °, -5 °] is defined as to overlook scope facial image.It is different according to luffing angle scope, have narrow The concept of adopted face average power figure, it is specific as follows：
Narrow sense face average power figure (Narrow face mean energy image, NFMEI)：Refer to same people same One luffing angle scope, the different imaging importing summations vacillated now to the left, now to the right under angle are averaging obtained average image again, according to people Different every 1 people of face luffing angle scope includes 3 width average power images, respectively looks up average power figure, looks squarely average energy Spirogram and vertical view average power figure.
Given multi-pose Gray Face image Ij(x, y), the calculation formula of narrow sense face average power figure is such as shown in (1)：
Wherein, MkThe sum of image when representing same luffing angle scope, angle change of vacillating now to the left, now to the right, k represent different Pitching
Angular range, k=1 represent to look up, and k=2 represents to look squarely, and k=3 represents to overlook, and j represents j-th of angle of vacillating now to the left, now to the right The image of change is spent, x, y represent two dimensional image plane coordinate.
2.2 narrow sense face variance energy diagrams
It is different according to luffing angle scope, there is the concept of narrow sense face variance energy diagram again, it is specific as follows：
Narrow sense face variance energy diagram (Narrow face variance energy image, NFVEI)：Refer to same people In same luffing angle scope, the different images vacillated now to the left, now to the right under angle square poor with corresponding narrow sense face average power figure Obtained image is averaging again.
For multi-pose Face gray level image Ij(x, y), the definition of narrow sense face variance energy diagram is such as shown in (2)：
Wherein, MkThe sum of image when representing same luffing angle scope, angle change of vacillating now to the left, now to the right, k represent different Luffing angle scope, k=1 represent to look up, and k=2 represents to look squarely, and k=3 represents to overlook, Fk(x, y) represents a certain luffing angle model The narrow sense average power figure obtained during angle change of being vacillated now to the left, now to the right in enclosing, j represent the image of j-th of angle change of vacillating now to the left, now to the right, X, y represent two dimensional image plane coordinate.
With reference to Fig. 3 and Fig. 4, it show a certain people and overlooks, looks squarely, looking up and occur seven kinds in the case of three kinds of pitching and vacillate now to the left, now to the right Image and its corresponding narrow sense face average power figure and variance energy diagram during change.
3. pair face average power figure and variance energy diagram carry out MB-LBP feature extractions
Face energy diagram is the primary features obtained to the superposition of posture facial image, be can be directly used for point to identify.But It is bad due to still suffering from data redundancy, recognition effect in face energy diagram, so the present invention uses MB-LBP to face energy diagram Further Feature Extraction is done, for recognition of face.
The face identification method of multiple dimensioned local binary patterns, this method average ash between block of pixels (sub-block) The comparison of angle value come replace between traditional LBP operators pixel value relatively i.e. can be achieved MB-LBP calculate.Each block of pixels is bag Square block containing adjacent pixel.MB-LBP scales are represented as parameter, 9 × L × L according to the length of side L of block,
MB-LBP features are represented by：
Wherein
gkRepresent the gray value of single pixel.B represents the average gray value of nth pixel block.
Fig. 8 is the method for expressing of MB-LBP features.Fig. 9 is respectively by part face average power figure and face variance energy Scheme the texture image after MB-LBP feature extractions.
4. read Face Image with Pose Variations to detect with human face region
The change definition of 4.1 human face postures
With reference to Fig. 2, change of the face in 3-dimensional space is respectively the translation and rotation along axle, wherein translated along the left and right of axle, Facial image inclination caused by a certain angle is rotated along the upper and lower translation of axle, along the anterior-posterior translation of axle and centered on axle all may be used To be effectively overcome by the method for geometrical normalization.But for facial image with central shaft rotate caused by up and down pitching Change, the several normalization of change of being vacillated now to the left, now to the right centered on axle caused by rotation can not also overcome.The present invention by facial image with Change turns into pitching and changed caused by centered on axle, can be divided into according to the difference of the anglec of rotation and look up, looks squarely and overlook；Face The change that image is come with axle rotating band turns into change of vacillating now to the left, now to the right.
4.2 human face regions detect
With reference to Fig. 5 and Fig. 6, the present invention, which needs to read first from multi-pose Face storehouse, to be had pitching change and vacillates now to the left, now to the right The Face Image with Pose Variations of change.Then effective human face region is obtained by AdaBoost algorithms.
Adaboost graders are formed by the cascade of multilayer Weak Classifier, and the correct result obtained by first layer grader touches Second layer grader is sent out, third layer grader is triggered by the correct result of second layer output, by that analogy.On the contrary, from any one The result being denied of individual layer output can all cause detection to stop immediately.By the threshold value for setting every layer so that vast majority of people Face can be transferred through, it is non-face can not be by, so the layer close to cascade classifier rear end have rejected most non-face.Experiment Show, AdaBoost algorithms can effectively detect human face region.
5. human face region picture size normalizes
, it is necessary to carry out size normalization to all images after human face region image is obtained.By human face region in the present invention Size be normalized to 230 × 270 pixels.Fig. 7 is the human face region image detected.
6. pair standard faces image carries out MB-LBP feature extractions
Obtained standard faces image is subjected to MB-LBP feature extractions, the extracted characteristic value of storage, known for classifying Not.
Figure 10 is to the texture image after standard faces image MB-LBP.
7. Classification and Identification
Arest neighbors sorting technique based on Euclidean distance, calculate standard faces eigenmatrix and face energy diagram eigenmatrix Distance, realize classification, last output result.
8. test the Classification and Identification process of face
Test process：
(1) firstly the need of extracting facial image from test library, and face area is carried out to image based on Adaboost algorithm Domain is detected, and is obtained test face area image, is carried out size normalized to it, obtain 230 × 270 standard faces image T.
(2) MB-LBP feature extractions and then by standard faces image T are carried out.
(3) finally, the feature of standard faces image T and face average power figure and variance energy diagram is subjected to arest neighbors point Class draws classification results.

Claims (1)

  1. A kind of 1. Pose-varied face recognition method based on MB-LBP features and face energy diagram, it is characterised in that including following Step：
    (1) Face Image with Pose Variations training storehouse is established, and trains the face images in storehouse to carry out Face Image with Pose Variations Size is normalized, and the size of human face region is uniformly normalized into 230 × 270 pixels；
    (2) according to the pitching excursion of human face region image luffing angle different demarcation face, with reference to pitching excursion structure Build narrow sense face average power figure and narrow sense face variance energy diagram, the primary features as Pose-varied face recognition；
    Facial image of the luffing angle excursion between [- 5 °, 5 °] is defined as to look squarely scope facial image, will [5 °, 30 °] between facial image be defined as looking up scope facial image, the facial image between [- 30 °, -5 °] is defined as overlooking Scope facial image；
    Involved narrow sense face average power figure FkThe expression formula of (x, y) is：
    <mrow> <msub> <mi>F</mi> <mi>k</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <msub> <mi>M</mi> <mi>k</mi> </msub> </mfrac> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>M</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>I</mi> <mi>j</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>,</mo> <mi>k</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow>
    In formula, MkThe sum of image, I when representing same luffing angle scope, angle change of vacillating now to the left, now to the rightj(x, y) is multi-pose ash Facial image is spent, k represents different luffing angle scopes, and k=1 represents to look up, and k=2 represents to look squarely, and k=3 represents to overlook, j tables Show the image of j-th of angle change of vacillating now to the left, now to the right, x, y represent two dimensional image plane coordinate；
    Involved narrow sense face variance energy diagram DkThe expression formula of (x, y) is：
    <mrow> <msub> <mi>D</mi> <mi>k</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <msub> <mi>M</mi> <mi>k</mi> </msub> </mfrac> <munderover> <mo>&amp;Sigma;</mo> <mi>j</mi> <msub> <mi>M</mi> <mi>k</mi> </msub> </munderover> <msup> <mrow> <mo>(</mo> <msub> <mi>I</mi> <mi>j</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>F</mi> <mi>k</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>,</mo> <mi>y</mi> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> <mi>k</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> </mrow>
    (3) using MB-LBP algorithms to the narrow sense face average power figure and narrow sense face variance energy diagram that are obtained in step (2) Further Feature Extraction is carried out, stores the MB-LBP characteristic informations for Classification and Identification；
    Involved MB-LBP features are represented by：
    Wherein
    gkRepresent the gray value of single pixel；B represents the average gray value of nth pixel block；
    (4) Face Image with Pose Variations to be detected is read, human face region is detected and extracts face；By facial image with Change caused by centered on axle is referred to as pitching change, can be divided into according to the difference of the anglec of rotation and look up, looks squarely and overlook；Face The change that image is come with axle rotating band is referred to as change of vacillating now to the left, now to the right；Effective human face region is obtained by AdaBoost algorithms；
    (5) human face region extracted is subjected to size normalized, obtains standard faces training image；By human face region Size is normalized to 230 × 270 pixels
    (6) MB-LBP feature extractions are carried out to standard faces training image, and store extracted MB-LBP characteristic informations；
    (7) for acquisition in the MB-LBP characteristic informations of the standard faces image to be detected of acquisition in step (6) and step (3) The MB-LBP characteristic informations of storehouse face energy diagram are trained, Classification and Identification is carried out by the nearest neighbor classifier based on Euclidean distance, Finally export face recognition result；
    (8) the Classification and Identification process of face is tested：
    Facial image is extracted from test library, and human face region detection is carried out to image based on Adaboost algorithm, is tested Human face region image, size normalized is carried out to it, obtain 230 × 270 standard faces image T：By standard faces image T Carry out MB-LBP feature extractions：The feature of standard faces image T and face average power figure and variance energy diagram is carried out nearest Neighbour's classification draws classification results.
CN201410195919.0A 2014-05-09 2014-05-09 A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram Active CN103971096B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410195919.0A CN103971096B (en) 2014-05-09 2014-05-09 A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410195919.0A CN103971096B (en) 2014-05-09 2014-05-09 A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram

Publications (2)

Publication Number Publication Date
CN103971096A CN103971096A (en) 2014-08-06
CN103971096B true CN103971096B (en) 2017-11-21

Family

ID=51240567

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410195919.0A Active CN103971096B (en) 2014-05-09 2014-05-09 A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram

Country Status (1)

Country Link
CN (1) CN103971096B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105825243A (en) * 2015-01-07 2016-08-03 阿里巴巴集团控股有限公司 Method and device for certificate image detection
CN105590107B (en) * 2016-02-04 2019-07-02 山东理工大学 A kind of face low-level image feature construction method
CN108388862B (en) * 2018-02-08 2021-09-14 西北农林科技大学 Face recognition method based on LBP (local binary pattern) characteristics and nearest neighbor classifier
CN108932587B (en) * 2018-06-29 2021-09-21 大连民族大学 Overlooking pedestrian risk quantification system of two-dimensional world coordinate system
CN112528777A (en) * 2020-11-27 2021-03-19 富盛科技股份有限公司 Student facial expression recognition method and system used in classroom environment
CN112580530A (en) * 2020-12-22 2021-03-30 泉州装备制造研究所 Identity recognition method based on fundus images
CN114821773A (en) * 2022-04-14 2022-07-29 杭州华橙软件技术有限公司 Gait data processing method, gait recognition method and related equipment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102855496A (en) * 2012-08-24 2013-01-02 苏州大学 Method and system for authenticating shielded face
CN103218606A (en) * 2013-04-10 2013-07-24 哈尔滨工程大学 Multi-pose face recognition method based on face mean and variance energy images
CN103661102A (en) * 2012-08-31 2014-03-26 北京旅行者科技有限公司 Method and device for reminding passersby around vehicles in real time

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7043079B2 (en) * 2002-04-25 2006-05-09 Microsoft Corporation “Don't care” pixel interpolation
US7920751B2 (en) * 2007-03-16 2011-04-05 General Electric Company Adaptive gradient weighting technique for detector bad cell correction

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102855496A (en) * 2012-08-24 2013-01-02 苏州大学 Method and system for authenticating shielded face
CN103661102A (en) * 2012-08-31 2014-03-26 北京旅行者科技有限公司 Method and device for reminding passersby around vehicles in real time
CN103218606A (en) * 2013-04-10 2013-07-24 哈尔滨工程大学 Multi-pose face recognition method based on face mean and variance energy images

Also Published As

Publication number Publication date
CN103971096A (en) 2014-08-06

Similar Documents

Publication Publication Date Title
CN103971096B (en) A kind of Pose-varied face recognition method based on MB LBP features and face energy diagram
Sheela et al. Iris recognition methods-survey
Zhang et al. Multi-phase offline signature verification system using deep convolutional generative adversarial networks
Suruliandi et al. Local binary pattern and its derivatives for face recognition
CN108614999B (en) Eye opening and closing state detection method based on deep learning
Shams et al. Iris recognition based on LBP and combined LVQ classifier
Wang et al. Review of ear biometrics
CN105825176A (en) Identification method based on multi-mode non-contact identity characteristics
CN105956579A (en) Rapid finger vein identification method integrating fuzzy template and point characteristics
CN109344856B (en) Offline signature identification method based on multilayer discriminant feature learning
CN107169479A (en) Intelligent mobile equipment sensitive data means of defence based on fingerprint authentication
Savithiri et al. Performance analysis on half iris feature extraction using GW, LBP and HOG
CN103020602A (en) Face recognition method based on neural network
Li et al. Multi-view vehicle detection based on fusion part model with active learning
CN103034840B (en) A kind of gender identification method
Xie et al. Review of personal identification based on near infrared vein imaging of finger
Wang et al. Local feature fusion and SRC-based decision fusion for ear recognition
Gona et al. Convolutional neural network with improved feature ranking for robust multi-modal biometric system
Chinnappan et al. Fingerprint recognition technology using deep learning: a review
Sudhamani et al. Fusion at decision level in multimodal biometric authentication system using Iris and Finger Vein with novel feature extraction
Méndez-Llanes et al. On the use of local fixations and quality measures for deep face recognition
Rajithkumar et al. Template matching method for recognition of stone inscripted Kannada characters of different time frames based on correlation analysis
Rao et al. Amalgamation Biometric Deep Features in Smart City-ITS Authentication
Zou et al. Deep learning-based pavement cracks detection via wireless visible light camera-based network
Cai et al. SVM-KNN algorithm for image classification based on enhanced HOG feature

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant