Symbolic Representation Model for Off-Line Signature Verification

Abstract

Nowadays, in our daily life, the importance of handwritten signatures is increasing, because people are more comfortable with pen and papers for all the legal transactions. So it is necessary to verify the identity of a person based on his handwritten signature. Identifying the human identity is more difficult work. So to overcome this problem offline signature verification method is proposed based on symbolic representation model. Writer–dependent technique is used to generate symbolic representation model. The system verifies the signature of two class i.e. genuine signature and skilled forgery. In the proposed work, Local Binary Pattern (LBP) features are used for feature extraction and symbolic features are then extracted for each feature in every signature class. As a result, the number of symbolic features is obtained for each individual's handwritten signature class. The k-Nearest Neighbor algorithm is used to classify the genuine signature and skilled forgeries. For analyzing the proposed approach Bangla offline signatures dataset is used. Results are discussed based on the writer dependent and writer independent techniques.