|
Bioscrypt's verification library, Bioscrypt
Core, is a sophisticated, state-of-the-art algorithm for
comparing fingerprint images. It is at the heart of all Bioscrypt
product offerings and is also available for license.
Bioscrypt
Core is proven technology. It is the engine behind the over
50,000 Bioscrypt physical access products that have been deployed
to organizations such as NASA, E*Trade, and Continental Airlines.
It is at the heart of over 100,000 Network Access stations, including
Targus fingerprint capture devices. Additionally, Bioscrypt
Core has been selected by various fingerprint sensor manufacturers
such as Atmel, Authentec and Fidelica, and application developers
such as Indivos and Sense Technologies.
The Bioscrypt Core algorithm was originally
developed along with Areté Associates, a defence research company
with over 25 years of experience in image processing, noise reduction,
and pattern recognition associated with advanced sensor systems
in use by the U.S. Department of Defence. The fundamental approach
applied to developing this technology was to make the most accurate
and complete comparison possible.
Bioscrypt Core has been specifically
designed to be portable and has been successfully deployed on numerous
operating systems and processing architectures. The library is available
under Windows, Windows CE, and Linux, as well as for many popular
Digital Signal Processors. The Bioscrypt team has extensive experience
with both PC and embedded software development. Over the past six
years Bioscrypt has developed products using Texas Instruments,
Analog Devices, and Motorola DSPs.
Bioscrypt Core has three primary functions:
Enroll, Verify, and Identify.
Enrollment
is the process of registering a user. The Enroll function accepts
a raw image delivered directly from a sensor. The image is processed,
enhanced, and compressed to create a fingerprint template. The template
is returned, along with image statistics reflecting the quality
of the enrollment. Subsequent verifications and identifications
compare live-scan images to previously enrolled templates.
Verification is performed in order
to validate a user¹s identity. It consists of
comparing a raw candidate image to a previously enrolled
template. A score is returned indicating the similarity
of the candidate and template. This score can be compared
to a threshold to make a yes/no decision.
Identification consists of comparing
a raw candidate image to a list of previously enrolled
templates. Through a series of screening processes,
the algorithm narrows the list of templates to a manageable
size. Those templates surviving screening are compared
to the candidate and verification scores are provided.
A score exceeding a pre-set threshold indicates a
positive identification.
|