曝光台 注意防骗
网曝天猫店富美金盛家居专营店坑蒙拐骗欺诈消费者
which combines accuracy and
comfort (it’s easier to place one’s
finger on a detector than to place
your eye in front of a camera).
But other approaches have
made breakthroughs in recent
years. In particular iris
recognition is at least as
accurate as fingerprints, while
facial recognition requires
a relatively simple detector
(a camera), but is not as
accurate.
Another key biological element
is of course our DNA, reputed
infallible, although entailing
certain legitimate ethics issues.
However, the DNA must be
physically sampled from the
person to be tested, and the
processing is long and costly.
A person can be identified from
his or her palm prints. Or even
from walk, or footfall patterns,
but both of these techniques are
still nascent and do not offer
sufficient precision.
safran magazine November 2007 _ safran magazine _ November 2007
18-19 Special report
technology. How does a system
differentiate between individual
fingerprints, irises or faces? That’s the
job of biometric processing software,
which calls on very different methods
depending on the body part in
question.
a technology
primer
W e usually see only the outside
aspect of a biometric
system, namely a fingerprint
reader or camera.
But it’s what behind the detector that
counts: the software that processes the
data gathered by the detector, and the
database used to compare the new fingerprint
with all previous ones. Therein
lies the added value of a biometric system,
which has to combine performance,
precision and reliability, since most large
systems of this type operate 24 hours a
day, 7 days a week.
But how exactly do they work? Every-
The system identifies as many of these
points as possible, assigning spatial and
angular coordinates to each minutiae.
Once this “map” has been established,
the system compares it to minutiae
already in its database. If two data sets
are identical, then it’s the same person.
This technology is already considered
mature, and currently accounts for more
than 65% of the biometrics market.
Filming faces
Face recognition, although less widespread,
is also a very successful technique,
largely because it is relatively easy to use.
The face is first filmed, in close-up or
from a distance, depending on the venue
(airport, street, etc.), the system used
(close-up or wider angle view) and the
degree of accuracy desired. The system
identifies the face on the screen, formats
it, if necessary adjusting the perspective,
and then compares it to a list of faces in
its memory, using very sophisticated algorithms.
One of the main advantages of
these systems is that they are comfortable
(people are simply “shot” by a camera),
and intuitive (anybody can check if the
system has made a mistake or not). On
the other hand, reliability depends on
the shooting conditions. If the subject is
poorly lit, too far away or out of focus,
the results will not be as accurate. Once
again, however, technical advances are
making these systems increasingly tolerant
of less than ideal shooting conditions.
This is still a relatively recent technology,
and currently accounts for a little over
10% of the market.
Coding eyes
Another identification technique now
being developed is for the iris, which
offers theoretical precision at least equal
to fingerprints, but is not as comfortable
(you have to position your eye in front of
a reader). When you take a closer look,
the iris comprises a seemingly chaotic
entanglement of micro-tubes, in a pattern
that is specific to each person. The
system divides this tangled mass into
eight concentric circles, like a target.
The pattern in each one looks something
like a barcode. Processing the image with
thing depends on the biometric marker
involved…
Zebra skin and
fingerprints
Let’s start with fingerprints. When you
place your finger on a detector the processing
software analyzes the ridge patterns,
which look a bit like a zebra skin. It
determines the points of interest, which
specialists call minutiae. There are two
main types: ridge endings and bifurcations.
These are the points that enable
analysts to tell the difference between
individual fingerprints.
1. Fabricatio n
of the prisms for
fingerprint detection.
2. ASSEMBLY
of fingerprint
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址：
航空资料39(48)
