Offline Signature Verification Using Supervised and Unsupervised Neural Networks?

Offline systems have only the static image containing the signature as an input, without having any knowledge on the signing process. Some difficulties that may arise in offline systems are related to the scanning process (noise on the image) and to the signature acquisition process where different pen tips and widths can produce different shapes, in this Dissertation work image preprocessing is applied to remove defects in captured images and extract the characteristics of preprocessed image. After that Supervised and Unsupervised techniques are applied as classifier for signature Verification. A comparison among the techniques is drawn and generalized on which of these techniques provide the better results i.e. which technique is more suitable for identifying the forged signatures. Percentage accuracy is calculated for each network and FAR and FRR is calculated for proposed Signature Verification System.