Offline Signature Validation using Image Processing

Thesignatureiswidelyusedasameansofpersonalverificationemphasizestheneedforanautomaticver ificationsystembecauseoftheunfortunatesideeffectofbeingeasilyabusedbythosewhowouldfeignthe identification or intent of an individuals. Verification of signature can be performed either by Online or Offline based onthe type of application. Online system uses dynamic information of a signature captured when the time of signature hasbeen made. Offline systems work on the scanned images of a signature. In this paper, we are presenting an Offlinesignature verification where we are using both ANN and CNN algorithm. The geometric features like Area, Eccentricity,Kurtosis, Center of gravity, and Skewness are extracted from image in ANN and an augmentation technique in CNN. Before we extract the features, preprocessing of ascanned image are necessary to isolate the signature part and to remove any type of noises present in it. Thesystemisinitiallytrainedusingadatabaseofsignaturesobtainedfromindividualswhosesignature shavetobeauthenticatedbythesystem. Weareusinglessdatasetscomparedtoexistingmodel wewill dealingwithitinlatersections.