VASC Seminar Announcement
        =========================

Speaker: Marios Savvides 
Affiliation:  Carnegie Mellon University, Electrical and Computer Engineering
E-mail:  msavvid@ri.cmu.edu
Date: Monday, October 7th
Time: 3:30pm
Place: NSH 1507

Title:   Face Identification using Advanced Correlation Filter Methods.

Abstract: 

Verifying the identity of a user is critical for many security applications such as
e-commerce, access control and surveillance. Most current authentication systems
are password based, making them susceptible to problems such as forgetting the
password and passwords being stolen. One way to overcome these problems is to
employ biometrics (e.g., fingerprints, face, iris, voice, etc.) for authentication.

A major challenge in using biometrics is overcoming their normal variability. 
My research is focused on performing face verification robustly even in 
the presence of normal variations due to lighting, facial expressions and pose
changes.

Towards this task we have developed and evaluated advanced correlation 
filter designs that are capable of synthesizing a single filter template that capture the 
distortions from a set of training images.We report results based on CMU's AMP Labs 
facial  expression database  (15 subjects with 75 individual facial expressions) yielding 
0.15% equal error rates in certain circumstances using as few as 3 training images.

We also explore the task of using these advanced correlation filters to 
perform Illumination Tolerant Face Recognition. We observe excellent recognition 
rates (100% in certain circumstances) using the illumination subset of the PIE 
Database on 65 people. We achieve excellent performance even in the case where the  
filters templates are synthesized from near frontal illumination and testing is performed  
under unknown illuminations.  We will demonstrate that this method can also recognize 
partial face images (assuming fixed pose and scale) with high accuracy.