Transcript ECE 533 Final Project - Computer

ECE 533 Final Project
SIMPLE FACE RECOGNITION
IMPLEMENTATION FOR
COMPUTER AUTHENTICATION
Josh Easton
-
Tin-Yau Lo
Goal
 Demonstrate the feasibility of
computer authentication using facial
recognition algorithms
What is facial recognition?
 Every person’s face has a set of unique
characteristics
 Some examples are:
 Distance between eyes
 Location and size of nose
 Distance from forehead to chin
 Humans are able to easily recognize a face
What is computer-based facial
recognition?
 Programming a computer to use an
algorithm to detect if two faces match
Facial recognition algorithms
 Various computer algorithms exist that
can be used to recognize faces
 Eigenface analysis (AKA Principal
Component Analysis)
 Hidden Markov Models
Eigenfaces
 Computer is trained with several pictures of the
same face
 Eyes are used as reference point between
pictures
 Various Eigenvectors are calculated to create a
signature of the face
Eigenfaces
Embedded HMM
for Face
Recognition
Model-
- Face ROI partition
Face recognition
using Hidden Markov Models
One person – one HMM
 Stage 1 – Train every HMM

1
…

Stage 2 – Recognition
i
n
Pi - probability
Choose max(Pi)
Running the Programs
 The distribution came with the directory
“FaceRecognitionCap” and
“FaceRecognition”.
FaceRecognitionCap
 Quicktime Java program, that requires
Quicktime 6.1 and a compatible camera that
support Quicktime on Windows with a simple
recompilation.
 It runs out of the box on Mac OS X by doubleclicking the “FaceRecognitionCap” Icon. Push
“Power” to initialize the Firewire bus, and click
“Take Snapshot” to produce a 320x240
greyscale image suitable for “FaceRecognition”.
The resultant capture file is “test.jpg”
FaceRecognition
 FaceRecognition is the actual face recognition
engine. Type the following at the
“FaceRecognition” directory :
java FaceRecognition trainedimages testing.jpg
 A sample running such as the following will be
produced :
kenneth% java FaceRecognition trainedimages testing.jpg
Constructing face-spaces from trainedimages ...
Comparing testing.jpg ...
Most closly reseambling: 15.jpg with 2.108734631580217 distance.
kenneth%
Conclusion
 Facial recognition software is a new,
advanced replacement for text passwords
 We can look forward to seeing more facial
authentication systems in the future