Online Signature Verification for Forgery Detection

The increasing trend of using e-versions document transmission and storage requires the electronic verification sender/author. This research presents an efficient robust online handwritten signature system targeting rates better than available state-of-the-art systems in presence skilled forgeries. Fourier analysis is employed on signatures to represent feature vectors higher dimensional space followed by Local Fisher Discriminant Analysis obtain compress representation while enhancing inter-class scatter between patterns. Signature modeling performed m-mediod-based approach where m-mediods are put data distribution each class. Connected component labeling applied binarized images Urdu text extract ligatures which separated into primary diacritics. Fast Euclidean Distance used as dis(similarity). A total 2414 samples including forgeries considered our study. evaluation proposed Japanese dataset provided SigWiComp2013 realized promising results competitors.