Variance Based Extraction and Hidden Markov Model Based Verification of Signatures Present on Bank Cheques

In this paper a method is proposed for the extraction and verification of signatures present on images of bank documents, which include signature localization using variance, signature block extraction using entropy, normalization using desired mean and variance, and verification using a hidden Markov model (HMM) with fixed number of states. Horizontal projection, vertical projection, right envelope, left envelope, top envelope, bottom envelope, horizontal size variations and vertical size variations of the binary format of the genuine signatures are treated as the feature set of an individual. Six genuine signatures of each individual were used for training the HMM. Skilled forgeries of amateur type were used for testing the false acceptance rate of the system. It is observed that the extraction rate is 95% along with a verification rate of 85%.