Signature identification using neural networks

Abstract

Traditionally, the human signature has been used for the identification of a person. With the modernization of optical technology, it is now possible to electronically process human signatures and use neural network techniques to validate these signatures. The signature may involve visible parameters such as size, orientation, peaks and valleys. It may also involve invisible parameters such as depth, pressure variations, vibrations and consistency. This type of information can be suitably used for the training of feedforward neural networks for the purpose of signature identification. This article will present an approach to extract some of this information and demonstrate how this informati..