Automatic Signature Verification

While digital signatures are being tested and used for new applications, the old handwritten signature is still what most of us rely on for identity. But technology is also working to allow those signatures to be protected against fraud.

Manufacturers of image recognition software are finding ways to analyze those signatures beyond human eyes with increasingly efficient result. For example, at a recent banking conference, Parascript® LLC announced the next generation of its fraud detection and prevention software. The newest generation doesn't just pull up an electronic file for visual verification by a bank employee - it uses a number of advanced analysis tools that can find both random and skilled forgeries and can be easily integrated into existing document processing applications. The software is designed to work not only with original paper checks, but with images of checks - using tools such as pattern recognition, geometrical analysis, analytical analysis and neural networks to produce a "confidence value" about whether a check is legitimate. The system also can work with signature snippets cut from any type of document.

Earlier this year, Fiserv, Inc. introduced a product designed to allow community banks to take advantage of more sophisticated signature verification. Fiserv's product, which it is currently rolling out through check processing facilities across the country, works automatically to compare the signature on the image of a paid check to an electronic archive of previous signatures on the checking account. What the product looks for is based on a financial institution's customized criteria. The "digital interrogation," as Fiserv terms it, uses multiple algorithms to analyze signature characteristics and identify and present suspect checks for review by the financial institution.

Copyright © 2005 Bankers' Hotline. Originally appeared in Bankers' Hotline, Vol. 15, No. 6, 7/05