Biometric knee identification system developed
30 January 2013
Posted by Satvir Bhullar
A scientist from Lawrence Technological University has developed a biometric identification system based on knees.
The system, which was created by Computer Scientist, Lior Shamir, uses magnetic resonance imaging (MRI) to scan the knees of people in a crowd and identify them.
Individuals are recognised by mapping the bone structure inside their knee and then matching this to a biometric record, using software developed by Mr Shamir at the Michigan-based university.
The system was tested by attempting to recognise 2,686 individuals based on their knee scans. It proved able to identify people with an accuracy of 93 per cent.
This is not sufficiently accurate for it to be used as a stand-alone biometric system, but it could be used alongside other methods such as iris recognition and fingerprint scans.
The technology could potentially be fixed to walls in places such as airports, where it would identify individuals as they walked past. This would reduce the likelihood of false readings occurring.
Mr Shamir claims the system could bring added security to biometric recognition, as it is more difficult to fake internal body parts like knee bones than external features.
He said: "Deceptive manipulation requires an invasive and complicated medical procedure, and therefore it is more resistant to spoofing compared to methods such as face, fingerprints or iris."
A further advantage is that MRI does not expose people to potentially harmful radiation, such as that used in X-rays.
However, the system faces one major drawback, as MRI technology is not yet advanced enough to be effectively deployed in a busy environment like an airport. It requires large machinery and takes a long time to produce an image, which limits its short-term usefulness.
This means the system is dependant on developments in MRI technology before it can be widely used.
Mr Shamir believes that such progress is likely. He said: "Further studies will develop the concept of internal biometrics and will lead to automatic identification methods that are highly resistant to spoofing."