Face Recognition

1
Face Recognition

• By Sunny Tang

2
Outline

• Introduction
• Requirements
• Eigenface
• Fisherface
• Elastic bunch graph
• Comparison

3
Introduction

• What is face recognition?
• Applications
• Security applications
• Image search engine

4
Requirements

• Accurate
• Efficient
• Light invariant
• Rotation invariant

5
Eigenface

• Euclidean distance between images
• Principal component analysis (PCA)
• For training set T1, T2, TM
• Average face ? 1/MS TM
• Difference vector fi Ti ?
• Covariance matrix C 1/MS fn fTn

6
PCA
7
Recognition

• Projection in Eigenface
• Projection ?i W (T ?)
• W eigenvectors
• Compare projections

8
Fisherface

• Similar approach to Eigerface
• Fishers Linear Discriminant (FLD)
• PCA
• Scatter Matrix
• Projection Matrix

9
Fisherface

• FLD
• Between-class scatter matrix
• Within-class scatter matrix
• Projection Matrix

10
FLD
11
Elastic Bunch Graph

• Gabor wavelet decomposition
• Gabor kernels

12
Gabor Filters
13
Jets

• Small patch gray values
• Wavelet transform

14
Comparing Jets

• Amplitude similarity
• Phase similarity

15
Comparing Jets
16
Face Bunch Graphs (FBG)

• Stack like general representation
• Two types of FBG
• Normalization stage
• Graph extraction stage
• Graph similarity function

17
Graph Extraction

• Step 1 find approximate face position
• Step 2 refine position and size
• Step 3 refine size and find aspect ratio
• Step 4 local distortion

18
Recognition

• Comparing image graph
• Recognized for highest similarity

19
Comparison

• Eigenface
• Fast, easy implementation
• Fisherface
• Light invariant, better classification
• Elastic bunch graph
• Rotation, light, scale invariant

20
Q A Section