Signature Forgery Detection

By: Nicholas Gaudio, Clayton Keating, and Travis Parr

Advisor: Robert Morley

Abstract

The Signature Forgery Detection system utilizes image processing and neural network techniques to determine whether a signature is forged or not. Forgery and security are vital to an individual’s financial security and identity, and requires a comprehensive and forward thinking solution. Our neural network is programmed and trained through the use of Matlab’s Neural Network Toolbox involving novel preprocessing steps. The product is a physical device in which a signature can be captured by a stylus touch pad or video camera to then be processed and sent through a classification neural network. This system is able to tell the user if the processed signature is forged or not using a determined threshold of percent similarity to the trained network.