Effect and Uses of Synthetic Data Generation in Inverse Biometric Problem

Various biometric verification techniques are in use today like iris, fingerprint, signatures, hand vein structure, voice, face etc. The Direct problem states the feature extraction and matching whereas another branch which deals with the generation of synthetic biometric is an inverse biometric problem. This paper presents the effects and uses of such synthetic biometric. In some cases, it can be used to replace the original dataset so as to save time and expenses whereas on the other hand it can be used for spoofing and forgery. Therefore inverse biometric problem has a wide application in forensic investigation, systems with lack of available training sets, generation of anti-spoofing security systems and simulation of controlled and flexible training skills. This paper also suggests a probabilistic approach for an inverse biometric problem with respect to offline signature verification systems.