Retinal Identification System Using Fourier-mellin Transform and Fuzzy Clustering

The uniqueness of the retina blood vessels pattern is the most accurate pattern among other biometric systems. In this paper, we present a new approach for human identification using retina recognition and fuzzy C-means clustering algorithm. This method is insensitive to rotation, rescaling and transformation. The Fourier-Mellin transform coefficients and moments of the retinal image have been used as features extracted in our system. To compensate the rotational effects of the retinal scanner, a rotation compensator was designed. For optic disc localization, the Haar wavelet and snakes model have been used. The experimental results show an error rate close to zero for the propose method.