Somaliland: Iris Register to Eyedentify Voting Fraud


Left and right iris images from one Somaliland trial voter registration record

By:Lakshmi Sandhana
THE eyes have it. Somaliland’s election commission is trialling an iris-based biometric system that it hopes will put an end to duplicate registrations. This would make it one of the most advanced voter registration systems on the planet.
“Multiple registrations and multiple voting are a big issue in Somaliland,” says Mohamed Ahmed Hersi Geelleh, chairman of the National Electoral Commission (NEC) of Somaliland. “Most young people look for an opportunity to vote more than once to enhance their candidate’s chance of winning the election.”
Authorities had previously tried incorporating fingerprints and facial recognition when compiling voter lists, but still found a large number of duplicate registrations. So, on behalf of the NEC, election specialist Roy Dalle Vedove approached Kevin Bowyer and his team at the University of Notre Dame in Indiana for help designing a system using iris recognition, which is considered more accurate than fingerprint or face recognition.
“As the Somaliland experience indicates, fingerprints can be less powerful than iris for creating a national registry,” says Bowyer.
To test it out, 1062 trial voter records containing iris images from people in the Hargeisa and Baki areas were sent to Notre Dame for analysis. Bowyer’s team had to identify duplicate registrations intentionally introduced into these records.
They used algorithms to look for distinctive features in the iris patterns in each record, such as furrows, rings, ridges and freckles, to generate an iris “fingerprint”. Each iris fingerprint was checked against all the others to determine duplicates. Bowyer and his team were 100 per cent accurate in identifying the 457 duplicate registrations in a report to the NEC, which was made public last month.

If a full-scale version of the iris system is put in place for the national elections next year, “Somaliland would have the most technically sophisticated voting register in the world”, says Bowyer.
The team’s algorithms are able to identify people who have previously registered even if they are wearing different types of coloured or textured contact lenses. The team plans to present this at the UK Biometrics Working Group meeting in London in October.
This article appeared in print under the headline “Eyedentify duplicate voters with iris register”