|
 问题与解答
1. Q: What is fingerprint
verification?
A: Fingerprint
verification means comparing a fingerprint
from the sensor to a particular registered
fingerprint in a database, to confirm
that they are from the same finger.
2. Q: What is fingerprint
identification?
A: Fingerprint
identification means comparing a fingerprint
from the sensor to the fingerprint template(s)
in a database in order to determine a
person's identity.
3. Q: What is False
Accept Rate (FAR)?
A: The False Accept
Rate (FAR) is the probability that a finger
will be incorrectly verified when compared
to a different finger in the database.
The FAR is used as a threshold in the
verification process (see question 6).
The default value of FAR is 0.01% (0.0001),
but it can be set to any value greater
than 0% and less than 100%. The lower
the FAR is set, the smaller the chance
that someone will be incorrectly verified.
The higher the setting, the greater the
chance some people will have trouble being
correctly verified.
4. Q: What is False
Reject Rate (FRR)?
A: The False Reject
Rate (FRR) is the probability that a finger
will fail to be verified when compared
to the corresponding finger in the database.
It is an average value for the general
population and can vary significantly
for individual users. The FRR cannot be
directly controlled by the software, but
varies with the FAR: the smaller the FAR,
the larger the FRR and vice versa. Usually,
a person who is incorrectly rejected will
be verified if they just put their finger
on the sensor again (perhaps more carefully).
Some people with very dry skin may consistently
have difficulty if the FAR is set too
low.
5. Q: Can I do
identification with my database?
A: Identification
is done by comparing a fingerprint from
the sensor to registered fingerprints
in a database one at a time until it is
verified. The probability that someone
will be incorrectly identified depends
on the number of fingerprints (N) in the
database. If there is a relatively small
number, the formula is approximately equal
to N*FAR. "Small" means much
less than 1/FAR. For example, for the
default value of FAR = 0.0001 (i.e., 0.01%),
1/FAR = 10,000. Using this FAR and a database
with 100 registered fingerprints, the
probability of incorrectly identifying
someone is 0.01, or 1%. If N is not much
less than 1/FAR, then the probability
of error will be unacceptably high. If
this is the case and the FAR cannot be
reduced sufficiently without making it
unacceptably difficult for some people
to be verified, then your database is
too large to do identification with.
6. Q: How does
U.are.U compare two fingerprints?
A: The U.are.U
Fingerprint System uses three different
types of fingerprint templates: pre-registration,
registration and verification templates.
Pre-registration and verification templates
consist of minutiae (ridge endings and
bifurcations) and other information that
is extracted from a single image from
the sensor, using a proprietary algorithm.
During registration, the U.are.U Fingerprint
System collects four images of the same
finger and generates four temporary pre-registration
templates. These are used to create a
single registration template, which is
then stored in a database with other user
information and the pre-registration templates
are discarded. A verification template
is created when someone puts a finger
on the sensor in order to be verified
or identified. They fingerprint matching
process uses a proprietary algorithm to
calculate the probability that a verification
template and a registration template came
from the same finger. It is mathematically
impossible to say that two fingerprints
match (or don't match) with 100% accuracy,
so the current value of the FAR is used
as a threshold. If the probability that
two prints come from different fingers
is less than the FAR, then they are declared
a match (see question 3 for more information
about the FAR).
7. Q: How do I
exchange a fingerprint template between
SDK Gold and SDK Platinum editions?
A: With the Platinum
SDK, the authentication server signs templates.
This security feature ensures that a template
can only be imported if it was signed
by a trusted server. As a result, templates
created with the Gold SDK cannot be used
by a Platinum SDK authentication server.
It is, however, possible to extract a
Gold template from a Platinum template
and then use it with the GOLD SDK.
8. Q: Are the U.are.U
SDKs compatible with HA-API/BIOAPI/BAPI?
A: The U.are.U
SDKs are not compatible with HA-API, BIOAPI
or BAPI. There are several competing standards
that are emerging in biometrics. DigitalPersona
has been-and continues to be-an active
participant in the development of these
standards. DigitalPersona is committed
to support standards as they are endorsed
by the marketplace and our customers
|
[型号:YLC
4000 ]
