A Novel approach to transformed biometrics - using ANN

Biometrics is the technique that utilizes the distinct genetic features collected from face, thumb, retina, eye etc. as an of authentication code. The number of valid biometrics is finite for a human. Any person who enrolls in multiple organizations, would have to use his/her different biometrics as mentioned above, so that there is fool proof authenticity, without any possibility of tampering of the biometric data of one organization by the other. But an innovative idea would be to use a technique that utilizes the same biometric in different organizations. The technique must be such that it is difficult to invert (i.e.) the original biometric must not be retrievable by any inverse transformations. This paper proposes a novel approach to transform a biometric feature using artificial neural network in 'spatial domain'. By this method, different distorted biometric images by using different transformations can be created from the originally collected biometric and hence the same biometric data can be used by different organizations. The results of the proposed approach are very promising and have been validated by the 'image transformation' principles.