This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1. FACE RECOGNITION By Vaishali S. Bansal M.Tech Computer
C.G.P.I.T. Bardoli
2. LEARNING OBJECTIVES THE VERY BASICS : What is face
recognition? Difference between detection & recognition !!! The
origin and use of this technology ? What are the various approaches
to recognize a face? OUR SELECTED FACE RECOGNITION METHOD :
Introduction to PCA Based Eigen Face Recognition Method.
3. WHAT IS FACE RECOGNITION? Face Recognition is the task of
identifying an already detected face as a KNOWN or UNKNOWN face,
and in more advanced cases, TELLING EXACTLY WHOS IT IS !
FEATUREFACE DETECTION FACE RECOGNITION EXTRACTION
4. METHODS FOR FEATUREEXTRACTION/FACE RECOGNITION
5. FACE DETECTION V/S FRECOGNITION Face database Output:
Mr.Chan Face detection Face recognition Prof..Cheng5 face interface
v.2a
6. THE "PCA" ALGORITHM
7. STEP 0: Convert image of training set to image vectors A
training set consisting of total M images Each image is of size
NxN
8. STEP 1: Convert image of training set to image vectors A
training set consisting of total M image foreach (image in training
set) { 1 Image converted to vector NxN Image N Ti } Vector Free
vector space
9. STEP 2: Normalize the face vectors 1. Calculate the average
face vectors A training set consisting of total M image Image
converted to vector Calculate average face vector U U Ti Free
vector space
10. STEP 2: Normalize the face vectors 1. Calculate the average
face vectors 2. Subtract avg face vector from each face vector A
training set consisting of total M image Image converted to vector
Calculate average face vector U U Then subtract mean(average) face
Ti vector from EACH face vector to get to get normalized face
vector i=Ti-U Free vector space
11. STEP 2: Normalize the face vectors 1. Calculate the average
face vectors 2. Subtract avg face vector from each face vector A
training set consisting of total M image Image converted to vector
i=Ti-U U Eg. a1 m1 Ti a2 m2 1= . . . . Free vector space a3 m3
12. STEP 3: Calculate the Eigenvectors (Eigenvectors represent
the variations in the faces ) A training set consisting of total M
image Image converted to vector To calculate the eigenvectors , we
U need to calculate the covariance vector C Ti C=A.AT where A=[1,
2, 3, M] Free vector space N2 X M
13. STEP 3: Calculate the Eigenvectors A training set
consisting of total M image Image converted to vector U C=A.AT Ti
N2 X M M X N2 = N2 X N2 Very huge Free vector space matrix
14. STEP 3: Calculate the Eigenvectors A training set
consisting of total M image N2 eigenvectors Image converted to
vector U C=A.AT Ti N2 X M M X N2 = N2 X N2 Very huge Free vector
space matrix
15. STEP 3: Calculate the Eigenvectors A training set
consisting of total M image N2 eigenvectors Image converted to
vector But we need to find only K U eigenvectors from the above N2
eigenvectors, where K
16. STEP 3: Calculate the Eigenvectors A training set
consisting of total M image N2 eigenvectors Image converted to
vector SOLUTION U DIMENSIONALITY REDUCTION Ti i.e. Calculate
eigenvectors from a covariance of reduced Free vector space
dimensionality
17. STEP 4: Calculating eigenvectors from reduced covariance
matrix A training set consisting of total M image M2 eigenvectors
Image converted to vector New C=AT .A U M XN2 N2 X M = M XM Ti
matrix Free vector space
18. STEP 5: Select K best eigenfaces such that K threshold ?
UNKNOWN FACE w1 Calculate Distance between = w2 input weight vector
and all the : weight vector of training set wk =|i|2 i=1M Weight
vector of input image