HMM BASED OFFLINE SIGNATURE FORGERY DETECTION

Abstract

The offline signature verification system finds several applications in monitory transaction systems like banks. Nevertheless, one of the major challenges in this instruction is the capacity of the organization to detect skilled and unskilled forgery. Many instances of bank check forgeries have been described. Most of the offline signature verification system adopts recognition based technique where the system sorts out a given signature sample as one of the samples from the database. However, detection of a forgery in a given sample is challenging as the input sample looks similar to ace of the samples in the database. In this report, we suggest an advanced approach for offline signature verification with a polar feature descriptor for a signature that contains Radon Transform and Zernike Moments. Confirmation is performed using Multiclass Support Vector Machine. At one time a signature is verified as being of a registered class, PLS Regression is applied to the sample against all samples in the database by the verified user to obtain a regression score. Log Likelihood of the sample against all samples of the user is calculated using Hidden Markov Model. Legitimacy of the classification is warranted if the regression score and Log Likelihood distance deviation are less than 5%. Results indicate that the system verifies signature with an accuracy of 98% with a false acceptance rate. 8%. Proposed system also detects skilled forgery with an accuracy of 71% and Random forgery with an accuracy of 76%.
Keywords: Offline Signature Verification, Skilled Forgery Detection, Hidden Markov Model, Partial Least Mean Square Regression, Support Vector Machine, Curvelet Transform, Radon Transform, Zernike Moments