General Introduction to Biometrics
4 of 23
Reference Manual StarSign® Bio Token 3.0 M/Edition 10.2005
ID No. 30017639
1.1
General Introduction to Biometrics
Scope
Biometrics is the science of measuring physical or behavioral characteris-
tics unique to an individual such as face, voice or fingerprint to verify a
person's identity. Biometric characteristics can be described as some-
thing we are.
Biometrics and other
Types of User
Authentication
Unlike user authentication based on something the user knows, such as
a PIN or password, or something he or she has, e.g. a smart card or other
token, biometric systems work by relying on a biometric characteristic -
something that is both unique and inseparably tied to the person. While
PINs, passwords and keys can be forgotten, lost, lent or stolen, biomet-
rics cannot. The user himself becomes the means of identification, the bi-
ological password.
Biometric user authentication can elevate overall system security and en-
hance ease of use, as users no longer have to remember PINs and pass-
words.
Enrollment and
Verification
Before biometric authentication can be used to verify the identity of a
user, a biometric enrollment has to be performed beforehand. This
means that the characteristic data of the biometric trait has to be cap-
tured and saved as a reference in a separate process in advance to verifi-
cation. During verification, the characteristic data of the biometric trait is
captured again and compared to the previously stored reference data. If
both data sets coincide to a sufficient level, access is granted.
Biometric Error Rates
In contrast to a PIN or password comparison, two different photos or
characteristic data sets captured of the same biometric trait will always
differ a bit due to positioning, background lighting, etc. Thus, biometric
comparison returns a figure which represents a level of coincidence, i.e.
the probability that two presented data sets belong to the same person.
Depending on a threshold value, access is granted or denied. As a con-
sequence, a slight possibility remains that an unauthorized user be
granted access to a protected system or that a legitimate user will be de-
nied access. The threshold value responsible for the error rates can be set
by the system administrator. These error rates are characteristic for all bi-
ometric systems and are called false acceptance rates (FAR) and false re-
jection rates (FRR).
Fingerprint
Verification
Fingerprint verification is not only the most prominent but also one of
the most secure and well-understood biometric measures. Software con-
verts the image of a fingerprint into digital form and extracts a set of
characteristics, i.e. a template, unique to the user's fingerprint. The char-
acteristic information from one fingerprint contains up to 60 key points.
Crucial key points where finger-ridges end or split up are local features
called minutiae. They provide unique, identifiable information.
