ConsentEASE

Presentation on theme: “Facial Recognition Justin Kwong Megan Thompson Raymundo Vazquez-lugo.”— Presentation transcript:

1 Facial Recognition Justin Kwong Megan Thompson Raymundo Vazquez-lugo

2 Why Facial Recognition? Does not depend on consent Ease of use for security Contact-free recognition Able to use existing databases and technology

3 Applications Missing Children Identification Criminal Identification Passports/Driver’s License Voter identification Welfare fraud Logging on to computer Accessing files Surveillance Access to Building

4 Problem Statement Identification: Given a stored database of faces, identify an unknown input face through facial recognition techniques. Verification: Given a stored database of faces, confirm or reject the claimed identity of the input face through facial recognition techniques. Additional Information: Facial recognition techniques may be enhanced by taking into consideration other factors such as gender, race, and age.

5 Issues Lighting Angles Apparel Resolution Background Noise Movement Facial Expression Segmentation (in a scene)

6 Recognition Techniques Correlation Eigenfaces Linear Subspaces Fisherfaces 3-D

7 Correlation Nearest Neighbor classifier Normalize image mean and and make unit variance Correlate images Heavy computation Subject to all previous issues (lighting, angles, etc.)

8 Eigenfaces Uses a Principal Component Analysis (PCA) Transform Uses subspace that maximizes scatter Reduces information Retains illumination affects Does not necessarily separate classes

9 Linear Subspaces Assumes face is a Lambertian surface Solve for albedo and surface normal (photometric stereo) Insensitive to lighting conditions (but not shadows) Computationally expensive to measure variability in class

10 Fishfaces Linear Discriminate Analysis (LDA) Uses a learning set Maximizes scatter between classes Insensitive to illumination

11 3-D 2-D methods are sensitive to external factors such as illumination, head pose, and are also sensitive to the use of cosmetics 3D facial recognition uses geometry of the face for accurate identification of the subject Yet, the problem of facial expressions is a major issue in 3D face recognition, since the geometry of the face significantly changes as the result of facial expressions The main technological limitation of 3D face recognition methods is the acquisition of 3D images, which usually requires a range camera In [5] is developed an expression-invariant 3D face recognition approach based on the isometric model of facial expressions

12 Test Imagery Color FERET Database (controlled setup) Yale Face Database (numerous variations on single class) Harvard Database http://www.face-rec.org/databases/

13 Preliminary Method Choice – Fisherfaces LDA useful outside of face rec. problem Theoretically insensitive to many issues (e.g. expression, lighting, etc.) Previous papers show lowest %error Most databases not setup for 3-D image creation

14 Literature www.face-rec.org Belhumeur, Peter; Hespanha, Joao; Kriegman, David. Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol 19, No. 7. July 1997 A. Bronstein, M. Bronstein, and R. Kimmel. Expression-invariant 3D face recognition, Proc. Audio & Video-based Biometric Person Authentication (AVBPA), Lecture Notes in Comp. Science 2688, Springer, 2003, pp. 62-69 Duda, Richard O., Hart, Peter E., Stork, David G. Patter Classification. Second Edition. John Wiley & Sons, Inc. Gross, Ralph. Shi, Jianbo. Cohn, Jeff. Quo Vadis Face Recognition? Robotics Institute. Carnegie Mellon University Torres, Luis. Is there any hope for face recognition?. Technical University of Catalonia. Barcelona, Spain Zhao, W.; Face Recognition: A Literature Survey. National Institute of Standards and Technology.