© 2021 Innovation Trail
background_fid.png
Tech

Fingerprint research points way to accuracy

fingerprint 3.jpg
Chad Miller
/
via Flickr
Nowadays, fingerprint analysis involves an expert comparing prints side by side, looking for matching attributes.

While each fingerprint is unique, lifting an intact print from a doorknob, window pane or shopping bag sometimes makes that fact hard to prove, especially in a court of law.

http://stream.publicbroadcasting.net/production/mp3/wned/local-wned-945965.mp3

Now research is moving fingerprint science closer to the sure thing many assume it already is. 

After all, it's had a while to develop.

“Over 100 years,” says Sargur Srihari, a University at Buffalo computer science professor, who is principal investigator on the research. 

Despite this century of use and development, mistakes are still made in fingerprint analysis and identification.

“I would not call it ironclad," Srihari says.  "People have always believed that fingerprints are infallible."

So a team at UB created a new method that looks at how rare particular features are, like certain kinds of swirls, ridges or shapes existing together in the same fingerprint.

“What our research is about is to try and quantify 'what is the degree of uncertainty?' however small it might be,” Srihari says.

The method spits out a number that indicates, for instance, that a given fingerprint likely belongs to one person in 25 million. That sort of statistic could eliminate a suspect from an investigation entirely, or do exactly the opposite. 

But don’t expect the technology to be in your local district attorney’s office anytime soon. Srihari is attempting to translate his research into a set-up that’s practical and affordable enough to get forensic units to sign on.

A decade ago, Srihari was the first researcher to determine each fingerprint is actually unique. But he admits this new research is preliminary and it could be many years before this method becomes the norm in forensic investigations.