Training course on in-depth concepts on Fingerprint Technology

1
Fingerprint Technology 001
Training course on in-depth concepts on
Fingerprint Technology
BIOENABLE TECHNOLOGIES PVT LTD C4/1, RAKSHA LEKHA
SOCIETY, KOREGOAN PARK, PUNE-1 WWW.BIOENABLETECH.C
OM INFO_at_BIOENABLETECH.COM
2
Minutiae, the discontinuities that interrupt the
otherwise smooth flow of ridges, are the basis
for most fingerprint authentication. Minutiae
points are local ridge characteristics that occur
at either a ridge bifurcation or a ridge ending.

• Algos can detect 30 or more minutiae points, new
  algos try to detect the max. no. of points
• One fingerprint can have more then 100 minutiae
  points
• Name of minutiae points, orientation, and
  distance from other points is stored in
  fingerprint template
• Fingerprint classifications like right/left
  loop, whorls, arches are not used by most
  minutiae algos
• Even if 15 minutiae points match in two
  templates they can be securely declared to be
  from same finger/person
• Security level define how many minutiae points
  must match. Security level is from 0-9, 1 may
  mean 5 3-5 minutia points must match depending on
  the algos

• Anomalies caused by scars, sweat, or dirt appear
  as false minutiae
• Not well developed algorithms can mis-judge
  minutiae points
• Pressure also can deform or merge minutiae
  points
• Water, oil, dust can hide, deform minutiae

3
Fingerprint Template
REGISTRATION
Registration/Enrolment Matching
ID2281
ID2281
Database ID Registered templates
Fingerprint Template 256-1000 bytes
ID2281
Optical Fingerprint scanner
Fingerprint Image 50KB-80KB
Fingerprint Minutiae extraction
MATCHING
11 matching (Verification, one to one) ID/Card
is required along with finger, ID can come from
keypad, card etc. extra cost for cards? Time
required to input ID on Keypad? More secure! More
useful for system more then 50 users! 1N
matching (Identification, one to many) ID/Card
is not required, software tries to match finger
scanned with indexed database of all registered
fingerprints. Can take longer time for system
with more then 50 users. No card reader or keypad
required. 1G matching (Group matching, one to
group) Modern software allow you to group users
using group ID. In case group ID is entered 1N
matching happen within the group only. Useful for
upto 500 users system. Supported by few product s
only.
ID2281
Fingerprint Template 256-1000 bytes
4
FAR/FMR False Acceptance rate, False match
Rate FRR/FNMR False Rejection Rate/False Non
match Rate ERR Equal Error rate

• Most figures are lab tested and not real
  environment figures. Most claims not verified by
  any body. Does not take dust, moist, dry
  conditions, Ways of testing different
• FAR/FRR are tested at default security levels.
  If security level change FAR/FRR also change, if
  FAR decrease, FRR increase, we want low FAR and
  low FRR. In 11 matching system lower security
  level is acceptable, In 1N FAR chances are high,
  to reduce FRR we increase security levels but
  then FRR increases
• Scanner as well as algos used can effect FAR/FRR
  along with people, behavior, awareness, place,
  environment, registration etc. (see next slide
  for details)
• In Time attendance systems low FRR is required,
  in high security Access control systems FAR must
  be almost zero and higher FRR may be
  acceptable.
• 1N is not common in the market so not necessary
  to always offer 1N. Focus on customer
  convenience and give choice to customer for 11,
  1N , pin/password, card readers etc.
• In India 1 in 100 people can have problem
  verifying or even registering fingers. This is
  prominent for very old people, due to finger skin
  type, manual labor, very deformed/uncommon
  finger ridge patterns, damage of finger skin,
  finger coloring, finger hardness.

5
Reasons for Fingerprint failures
Behavior and awareness also cause FAR/FRR
problems like higher pressure is bad for optical
scanners, very light pressure is also wrong,
finger must touch and cover the surface of the
scanner. Many people keep fingers tilted, some
scan the lower part of the finger, some scan the
upper part only, some peoples hand vibrate on
scanner, some place improperly in fear, some
think that they can get a electric shock, some
have privacy concerns and many other factors
related to person behavior can effect the
verification process. Education and awareness is
very important. Improper registration is cause
of 50 of the FRR problems. The person
registering finger must be able to judge if the
registration is perfect. Software normally do not
help. Since most people have first experience of
fingerprint system at the time of registration,
their fear and low awareness lead to low quality
registration and then FRR and sometimes FAR
problems. Registration must be repeated after 1-3
weeks of experience of employees with the system.
Also hamster based visual registration is better,
registering multiple fingers is better, I think
more work should be done to simplify registration
process and improve quality. Environmental
conditions effect the FRR significantly, in cold
season finger skin gets squeezed, in hot season
many people have sweat on their fingers, moist
and dry fingers due to work environment also
effect the FRR.
6
OPTICAL SENSOR WORKING
Live check Dual LEDs with prism are used to
avoid scanning of fake paper fingerprints.
Finger
Prism
LEDs
Lens 1
Lens 2
To PC(USB) or ARM controller
Optical Scanners
CMOS Image sensor
Issues

• Live check not secure enough
• Latent check increases verification time
• Larger Size, cost cannot be reduced after
  certain limit

7
Active/E-Field capacitive sensor
Capacitance is the ability to hold an electrical
charge. Fingerprint Cards' sensor contains tens
of thousands of small capacitive plates, each
with their own electrical circuit embedded in the
chip. When the finger is placed on the sensor
extremely weak electrical charges are created,
building a pattern between the finger's ridges or
valleys and the sensor's plates. Using these
charges the sensor measures the capacitance
pattern across the surface. The measured values
are digitized by the sensor then sent to the
neighboring microprocessor.

• ESD problem
• Surface degradation

The latter method, used by Fingerprint Cards, is
called the active or reflective method, and it
brings several advantages. Using the programmable
logic internal to the capacitive sensor
configuration it is possible to read off and
adjust the sensor reception to different skin
types and conditions. Another important benefit
is that the strengthened signal communications
between the fingerprint surface and the sensor
plates allows for the introduction of strong,
protective coating layers on the sensor surface.
This enables Fingerprint Cards, as the only
sensor vendor, to sustain the requirement for
15kV electrostatic discharge (ESD), as well as
wear cycle tests for one million touches to the
sensor.
8
Fingerprint sensor on card
Swipe sensors Latest in Fingerprint
technology Lower cost but not comfortable
enough, useful for mobile phones etc.

• Sites
• Findebiometrics.com
• Biodigest.com
• Biometricsgroup.com
• Biometrics.org
• Ibia.org
• Biometricsinfo.org
• Biometricscatalog.com

Thanks Pradeep Bhatia MD, BioEnable