Invariant features comparison in hidden markov model and SIFT for offline handwritten signature database

In Handwritten signatures analyzed for forgery have to undergo feature extraction process, due to varied samples in size rotation and intra-domain changes, invariance has to be achieved during feature extraction process; circular Hidden Markov Model with discrete radon transform approach of feature extraction provides invariance. On other hand Scale Invariant Feature Transform (SIFT) has inherent invariant feature extraction approach. This paper compares both approaches on common signature databases for False acceptance rate(FAR),False Rejection Rate(FRR) and Equal Error Rate(EER)