[silk] foolproof iris recognition?

[Udhay Shankar N]

http://blogs.zdnet.com/emergingtech/?p=88

Foolproof iris recognition technology?

Posted by Roland Piquepaille @ 10:35 am

For almost twenty years, the iris recognition research field has been
hampered because of a broad patent covering it. As this patent
recently expired, many teams around the world are again working on
new technologies in this field. Iris recognition is in fact seen as
the most accurate biometric recognition technology because no two
irises are identical. And researchers at the University of Bath in
England have developed new computer algorithms which are 100 per cent
accurate in initial trials. Now the researchers are putting online a
database of 16,000 iris images collected mainly from students. The
source code is also available if you want to further improve the algorithms.

Before going further, let's go back in time to understand why this
research field was almost inactive for twenty years. Life Style Extra
tells us the story.

    Looking into a camera to confirm your identity would now be
routine and - were it not for the US firm's virtual monopoly of the
technology - it would already be in use at cashpoints and passport
control. Its backers say it could reduce fraud and illegal immigration.

    Iridian Technologies, based in New Jersey, patented the system
of identifying people using the coloured part of the human eye in the
mid 80s and other scientists have had to pay tens of thousands of
American dollars to do any research in the field, thus hampering competition.

    But the patent expired in the US earlier this year and expires
in the rest of the world in February 2006.

Now, it's time to return to 2005 at the University of Bath.

    Engineers are currently road-testing their technology using a
specially-constructed database containing thousands of iris images
collected from students and colleagues at the University.

    By making this database available to other research groups, the
researchers hope to encourage more advances in iris recognition and
overcome some of the restrictions caused by a generic patent
(recently expired) which has limited innovation for the last two decades.

    "Our new algorithm does the same job as the one used by almost
all of the commercially available iris recognition systems, it just
does it better," said Professor Don Monro from the University's
Department of Electronic and Electrical Engineering.

Below is a picture showing how an iris picture is shot and rendered
on a computer screen before being analyzed (Credit: Smart Sensors Ltd.).

The iris image acquisition process

And below is an illustration of the iris image normalization process
(Credit: Smart Sensors Ltd.).

The iris image normalization process

    First, the inner and outer iris boundaries are located to
eliminate the pupil, eyelid and other "clutter". Then the iris image
is transformed from polar coordinates to a 512x80 fixed size
rectangular image to reduce the effect of iris dilation and
contraction, of which 512x48 will be coded. The non-uniform
background illumination is finally homogenized.

Now that you know how this new technology works, why are these
researchers willing to share their database? Here are Monro's answers.

    Most of the databases that are available are held by commercial
interests, so it is difficult for independent researchers to make
headway in this field.

    We are making the database available online so that researchers
around the world can use it to develop their own products. So far,
more than 30 research groups have applied to use it.

If you want to know more about this project, here are two links at
the University of Bath about the Iris Image Database and the Iris
Capture Project.

And for even more information, one of the industrial partners of the
University of Bath for this project is a U.K. company named Smart
Sensors Ltd., which has published two interesting papers about this
iris recognition technology.

Here are the links to these documents, "Novel high performance iris
feature extraction techniques" (PDF format, 1 page, 119 KB) and
"Complexity low complexity human iris feature coding human iris
feature coding" (PDF format, 1 page, 222 KB). The above illustrations
were extracted from these documents.

Sources: University of Bath news release, November 15, 2005; Life
Style Extra, November 15, 2005; and various web sites


--
((Udhay Shankar N)) ((udhay @ pobox.com)) ((www.digeratus.com))


----- End forwarded message -----
--
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820            http://www.ativel.com
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE

[demime 1.01d removed an attachment of type application/pgp-signature which had a name of signature.asc]