Beating signature fraud

student writing

PhD student Vu Nguyen has created the world’s best automated system for authenticating hand-written signatures.

Vu’s project won first place in the 2012 Competition on Forensic Signature Verification which was staged in conjunction with The International Conference on Frontiers in Handwriting Recognition held recently in Italy.

And Vu has a powerful motivation for wanting to stamp out signature fraud.

In 1997, Vu’s mother, a Chemistry lecturer at a University in Vietnam, was sacked amid claims of professional misconduct. In 2006, she discovered signatures had been forged on a range of documents. Unable to convince the authorities to investigate she went on a hunger strike. Eventually an agreement was reached with the University but the signature fraud was never examined.

Vu, then a Higher Degree Research student at Griffith University, made signature verification the focus of his IT research.

Traditionally, signatures under investigation have been referred to forensic handwriting experts who make visual checks for similarities or differences against known examples of the real signature.

Vu knew first-hand just how great the need was for a better system in which objective conclusions could be generated by a computer.

“Considering all the areas of daily life which involve signatures, such as credit card and banking transactions, and especially in court cases, I could see the need for a powerful and accurate automatic verification system,” Vu said.

The objective of the ICFHR competition was to compare the success of different automatic systems against the vision-based performance of forensic examiners. Vu’s system beat entries from across the globe.

Associate Professor Michael Blumenstein, Dean (Research) in the Science, Environment, Engineering, and Technology Group has applauded Vu and the international recognition his achievement has drawn.

“Vu’s PhD work has culminated in the publication of a number of influential papers in the area of automated signature verification using pattern recognition and artificial intelligence techniques,” Associate Professor Blumenstein said.

“His work is tremendous in that the papers we have co-authored report some of the highest accuracies for signature verification in the literature. And this has now been validated by Vu’s win against some of the world’s best in the field.”