Face Recognition TechnologyW.A.L.S.Wijesinghe
IntroductionBiometricsA biometric is a unique, measurable characteristic of a humanbeing that can be used to automatically recognize an individualor verify an individual’s identityFinger- ScanIris  ScanRetina ScanHand ScanFacial Recognition80 landmarks on a human face.Distance between eyes
Width of the nose
Depth of the eye socket
Cheekbones
Jaw lines
ChinWhy we choose face recognition over other biometric?It requires no physical interaction on behalf of the user.It does not require an expert to interpret the comparison result.  Identify a particular person from large crowdVerification of credit card, personal ID, passport
History of Face Recognition1. In 1960 , scientist (Bledsoe, Helen,Charles) began work on using the  computer           to recognize human faces.      2. Before the middle 90’s- single-face segmentation. 3. EBL-Example-based learning approach by Sung and Poggio (1994).4. The neural network approach by Rowley etal. (1998).5.FRVT-Face Recognition Vendor Test-(2002)6. FRGC-Face Recognition Grand Challenges-(2006)7. Polar Rose Technology-Text surrounding photo-(2007)					                  	3D image
Face recognition: ProcedureInput face image(Capture)Face feature extractionFeature MatchingDecision makerFacedatabaseOutput result
2.0 Face Feature Extraction Methods1. Eigen face  or  PCA (Principal Component Analysis)Other method;1.   EBGM -Elastic Bunch Graph Method.-2D Image2.   3D Face Recognition Method                                  3D Image
PCA-Principal Component Analysis(Eigen Face Method)1.Create training set of faces and calculate the eigen faces    ( Creating the Data Base)2. Project the new image onto the eigen faces.3. Check closeness to one of the known faces.4. Add unknown faces to the training set and re-calculate
1.0 Creating training set of imagesFace Image as  I(x,y)  be 2 dimensional N by N  array of    (8 bit) intensity values.Image may also be considered  as a vector of dimension  N2.    ( 256x256 image =  Vector of Dimension 65,536 )       y             Image  T1=I1(1,1),I1(1,2)…I1(1,N),I1(2,1)…….., I1(N,N)						   x
Training set of face images T1,T2,T3,……TM.- 1. Average Face of Image =Ψ =  1 ( ∑M  Ti )       ; M –no. of images				             M    i=1Ψ average face
2. Each Training face defer from average by vector ΦΦiEigen face   Each Image	              Average Image   Ti ΨΦi =Ti - Ψ
Uk Eigen vector ,λk  Eigen value of Covariance  Matrix  C Where  A is,λk  Eigen value  C= λkUk
Face Images  using as                                         Eigen Faces (Uk)     training images      (Ti)               U=( U11,…U1n,  U21,…U2n,…..,  Uk1,……Ukn, Um1,……Umn)    -Image must be in same size- Facedatabase
Using Eigen faces  Identify the New face image			             date base –eigen vectors U								ωk =   UkTΦNew Image(T)     Its  Eigen face (Φ)        U1                                                                    U2X	      .                                                             k Class                                                                      .UkΦ  = T – ΨΩ = ∑k=1m ωk=                minimum  ||Ω - Ωk ||
Mathematical equations-Identify new face image. 1. New face image T transform into it’s  eigen face component byΦ  = T – Ψ 2.   Find the Patten vector of new image  Ωωk =   UkTΦ   ;    where  Ukeigen vectors               Ω = ∑k=1m ωk   To determine  the which face  class provide the best input face image is to find the face class k by                              minimum  ||Ω - Ωk ||             Face  Image Detected  in k Face Class.

Face Recognition

  • 1.
    Face Recognition TechnologyW.A.L.S.Wijesinghe
  • 2.
    IntroductionBiometricsA biometric isa unique, measurable characteristic of a humanbeing that can be used to automatically recognize an individualor verify an individual’s identityFinger- ScanIris ScanRetina ScanHand ScanFacial Recognition80 landmarks on a human face.Distance between eyes
  • 3.
  • 4.
    Depth of theeye socket
  • 5.
  • 6.
  • 7.
    ChinWhy we chooseface recognition over other biometric?It requires no physical interaction on behalf of the user.It does not require an expert to interpret the comparison result. Identify a particular person from large crowdVerification of credit card, personal ID, passport
  • 8.
    History of FaceRecognition1. In 1960 , scientist (Bledsoe, Helen,Charles) began work on using the computer to recognize human faces. 2. Before the middle 90’s- single-face segmentation. 3. EBL-Example-based learning approach by Sung and Poggio (1994).4. The neural network approach by Rowley etal. (1998).5.FRVT-Face Recognition Vendor Test-(2002)6. FRGC-Face Recognition Grand Challenges-(2006)7. Polar Rose Technology-Text surrounding photo-(2007) 3D image
  • 9.
    Face recognition: ProcedureInputface image(Capture)Face feature extractionFeature MatchingDecision makerFacedatabaseOutput result
  • 10.
    2.0 Face FeatureExtraction Methods1. Eigen face or PCA (Principal Component Analysis)Other method;1. EBGM -Elastic Bunch Graph Method.-2D Image2. 3D Face Recognition Method 3D Image
  • 11.
    PCA-Principal Component Analysis(EigenFace Method)1.Create training set of faces and calculate the eigen faces ( Creating the Data Base)2. Project the new image onto the eigen faces.3. Check closeness to one of the known faces.4. Add unknown faces to the training set and re-calculate
  • 12.
    1.0 Creating trainingset of imagesFace Image as I(x,y) be 2 dimensional N by N array of (8 bit) intensity values.Image may also be considered as a vector of dimension N2. ( 256x256 image = Vector of Dimension 65,536 ) y Image T1=I1(1,1),I1(1,2)…I1(1,N),I1(2,1)…….., I1(N,N) x
  • 13.
    Training set offace images T1,T2,T3,……TM.- 1. Average Face of Image =Ψ = 1 ( ∑M Ti ) ; M –no. of images M i=1Ψ average face
  • 14.
    2. Each Trainingface defer from average by vector ΦΦiEigen face Each Image Average Image Ti ΨΦi =Ti - Ψ
  • 15.
    Uk Eigen vector,λk Eigen value of Covariance Matrix C Where A is,λk Eigen value C= λkUk
  • 16.
    Face Images using as Eigen Faces (Uk) training images (Ti) U=( U11,…U1n, U21,…U2n,….., Uk1,……Ukn, Um1,……Umn) -Image must be in same size- Facedatabase
  • 17.
    Using Eigen faces Identify the New face image date base –eigen vectors U ωk = UkTΦNew Image(T) Its Eigen face (Φ) U1 U2X . k Class .UkΦ = T – ΨΩ = ∑k=1m ωk= minimum ||Ω - Ωk ||
  • 18.
    Mathematical equations-Identify newface image. 1. New face image T transform into it’s eigen face component byΦ = T – Ψ 2. Find the Patten vector of new image Ωωk = UkTΦ ; where Ukeigen vectors Ω = ∑k=1m ωk To determine the which face class provide the best input face image is to find the face class k by minimum ||Ω - Ωk || Face Image Detected in k Face Class.