GMM for offline signature forgery detection

As signature continues to play a crucial part in personal identification for number of applications including financial transaction, an efficient signature authentication system becomes more and more important. Various researches in the field of signature authentication has been dynamically pursued for many years and its extent is still being explored. Signature verification is the process which is carried out to determine whether a given signature is genuine or forged. It can be distinguished into two types such as the Online and the Offline. In this paper we presented the Offline signature verification system and extracted some new local and geometric features like QuadSurface feature, Area ratio, Distance ratio etc. For this we have taken some genuine signatures from 5 different persons and extracted the features from all of the samples after proper preprocessing steps. The training phase uses Gaussian Mixture Model (GMM) technique to obtain a reference model for each signature sample of a particular user. By computing Euclidian distance between reference signature and all the training sets of signatures, acceptance range is defined. If the Euclidian distance of a query signature is within the acceptance range then it is detected as an authenticated signature else, a forged signature.