The fingerprints are collected using a touch pad scanner and the model is then able to reverse any warping, creating an original image of the print.
‘Every time you present the same biometric data the result is different,’ said Dr Li Wang, a researcher at the university’s department of computer science. ‘Variation in pressure, angle and light conditions all cause a difference. Those variations can cause problems in identifying prints so we are trying to minimise them.
‘We have a mathematical model which can track and collect the movement of each part of the data and then unwarp and align the fingerprint.’
The computer model can also use the tightly crowded sweat pores to improve accuracy, something distortion previously prevented.
‘Each finger tip has hundreds of sweat pores that are so densely packed that the slightest distortion means you can’t analyse them,’ said Wang. ‘Once the distortion is removed we can use the pores to improve the accuracy of the system.’
The model lays the unwarped print onto a virtual ‘image space’ that includes all the fingerprints available to the database. This allows the system to search a database in a few seconds, regardless of whether it contains thousands or millions of fingerprints, as it doesn’t need to compare each feature of the fingerprint against each print in the database individually.
To take the product to market three researchers, Professor Roland Wilson, Dr Abhir Bhalerao and Wang, have founded ‘Warwick Warp’ a university spin out company.
According to Wang the technology could be be used in commercial access control systems, financial transaction authorisation systems, ID cards, passports, border control systems and forensics.
‘The biometric market is growing over 200 per cent a year and fingerprints are one of the most widely used forms of data,’ said Wang. ‘The market is huge and very technology hungry. I believe the bottleneck is the reliability and accuracy of recognition. So if you can improve them you have a very tangible selling point. Why would you choose something less accurate?’
The system could allow forensic teams to find matches with smaller fragments of prints, something which could prove exceptionally useful as crime scenes produce few, if any, ideal samples. Wang warned that it depends which parts of the print are left. ‘It’s actually what’s in the print that matters. You could have a large print with no features in it, or even a tiny print with a distinctive feature in it,’ he said.
Wang believes the speed of the system will also be an attraction to the potential commercial customers of the product. Large databases of the customers fingerprints will be able to be searched through in a matter of seconds.
‘We are at the stage of finalising the algorithm,’ added Wang. ‘The next phase is looking for equity. We are looking for £500,000 but we are close to finalising an investment. With these extra resources we can commercialise the product.
Wang believes the technology has potential at the high and low ends of the market from the supplying of systems integration down to the release of the software.
For more information visit Warwick Warp’s website , which is currently under construction here.