8. What is it?
Facial recognition systems are built on
computer programs that analyze images of
human faces for the purpose of identifying
them.
9. How does it work?
Measure specific facial characteristics to
create unique file called “template”
Using templates, compare image to another
image
Produces a score on similarity
Video camera signals
Pre-existing photos
i.e. drivers license databases
10. 2D Facial Recognition
2D Recognition
Maximum angle: 35 degrees
Must be similar to program in database
Sometimes ineffective due to lighting changes
and other uncontrolled variables
11. 3D Recognition
3D Recognition
Can create template from face at 90 degree angle
More accurate
Uses depth and an axis of measurement not
affected by lighting
Example: Identix® - FaceIt®
Landmarks or nodal points
And now: FaceIt®Argus, skin biometrics
15. Security
Casinos
Super Bowl
Tampa, Fl. (2001): 19 people identified
London 2012 Olympics
MORIS
16. Transportation
Germany: Fully automated border controls
Australia: SmartGate
Compares the face of the individual with image in
the e-passport microchip
18. New Developments
ATM’s
Advertising & marketing
“Adidas is working with Intel to install and test digital walls with
facial recognition in a handful of stores either in the U.S. or Britain.
If a woman in her 50s walks by and stops, 60% of the shoes
displayed will be for females in her age bracket, while the other
40% will be a random sprinkling of other goods.
‘If a retailer can offer the right products quickly, people are more
likely to buy something,’ said Chris Aubrey, vice president of global
retail marketing for Adidas.”
20. Limitations
Not 100% accurate.
Accuracy can fluctuate because of:
picture quality
Lighting
camera positions
facial expressions
and more
21. Security Issues
Mistaken identity cases
Images cannot be used to convict suspects
The CCTV cameras in London
1 crime solved per 1000 cameras
23. MORIS
Mobile Offender Recognition and
Information System
Illegal search without a warrant
No information is stored
24. Facebook
Has roughly 600 million users
that means that Facebook has a database of 600
million faces.
Each time you “tag” a photo, Facebook learns
more about your face.
25. Google
Picasa uses the same tagging techniques
People fear a face recognition update to the
app Google Goggles.
the app may even be able to identify peoples SSNs
just from the photo.
-Age is an issue: a study by government’s National Insitute of Standards and Technology found false negative rates for face-recognition verification of 43 percent using photos from only 18 months earlier
Skin biometrics: surface texture analysis of texture of skin, uses algorithms to turn patch of skin into mathematical, measurable space Can define the differences between twins Many problems wouldn’t work if: significant glare on sunglasses, long hair in center of face, poor lighting, lack of resolution
detection: scans already existing 2d photograph or a shot from a 3D video Alignment: up to 90 degrees turned hereMeasurement: measures curves of face on submillimeter scale and creates a templateRepresentation: translates into unique code to create template with set of numbers to represent a person’s faceMatching: new technology converts 3D image into 2D image using algorithm to compare to 2D image in database