iris recognition by john daugman
TRANSCRIPT
-
7/30/2019 iris recognition by john daugman
1/13
Iris Recognition
Following the work of
John DaugmanUniversity of Cambridge
-
7/30/2019 iris recognition by john daugman
2/13
Biometricsfingerprint face iris DNA
Convenience incapturing fair good good poor
Low intra-classvariance
good fair excellent excellent
High interclassvariance
good good excellent excellent
Difficult to fool good good excellent excellent
Cost effective fair good fair poor
-
7/30/2019 iris recognition by john daugman
3/13
Properties of the iris Has highly distinguishing texture
Right eye differs from left eye
Twins have different iris texture
Not trivial to capture quality image
+ Works well with cooperative subjects + Used in many airports in the world
-
7/30/2019 iris recognition by john daugman
4/13
Represent iris texture as a
binary vector of 2048 bits
Representation ofiris and also of aperson
Textured region isunique for aperson
-
7/30/2019 iris recognition by john daugman
5/13
Find (nearly circular) iris and
create 8 bands or zonesNeed to locate theoverall region of
the iris. Then needto measuretexture in 1024smallneighborhoods;perhaps 128
around each of 8bands.
-
7/30/2019 iris recognition by john daugman
6/13
Cross correlate 1024 local
areas with a Gabor wavelet
Get 2 bits at eachlocation/orientation.
Threshold the dotproduct of 2 filterswith the iris area.
Polar coordinates
locate the texturepatch.
Filter (mask)has 2 width
parameters
-
7/30/2019 iris recognition by john daugman
7/13
Use 2nd directional derivative
and 1
st
directional derivative
LOG wave in alphadirection
Gaussian
smoothing inbeta direction
-
7/30/2019 iris recognition by john daugman
8/13
The directional filters defined
mathematically
sinusoidtaper downin radialdirection
Taper downin tangentialdirection
Image intensity
in polar coords
-
7/30/2019 iris recognition by john daugman
9/13
Summary of feature extraction Obtain quality image of certain (left) eye
Find boundary of pupil and outside of iris
Normalize radii to range, say, 0.5 to 1.0
Define the 8 bands by radii ranges
Perform 2 dot products at each of 1024locations defined around the bands byradius rho and angle phi
-
7/30/2019 iris recognition by john daugman
10/13
How is the matching done to
templates of enrolled persons? Person scanned under controlled
environment and iris pattern is stored
with ID (say address, SS#, etc.) Might be several million such templates
for frequent flyers (6B for all world)
At airport or ATM, scan unknownpersons left eye; then computeHamming distance to ALL templates.
-
7/30/2019 iris recognition by john daugman
11/13
Distributions of true matches
versus non matches
Hammingdistancesof true
matches
Hammingdistancesof falsematches
-
7/30/2019 iris recognition by john daugman
12/13
Design of former SENSAR ATM
iris scanner
-
7/30/2019 iris recognition by john daugman
13/13
ecogn on s poss e ycomparing unknown scan to
MILLIONS of stored templates Less than 32% unmatched bits means
MATCH
Only need to count unmatched bits
use exclusive OR with machine words
Mask off bad patches due to eyelid or
eyelash interference (have to detectthat)