Co-occurrence features and neural network classification approach for iris recognition

Biometrics provide ways of person authentication using physiological and behavioral characteristics of the subject. Among all characteristics, iris is best suited for biometric applications. In this paper, a new iris recognition method is presented which is based on co-occurrence features and neural network classification. The gray level co-occurrence matrices are utilized to extract the Haralick features which are further used to train the neural network classifier. Different parameters, like number of hidden layer neurons, training functions and performing functions, are also investigated. Effectiveness of the discussed approach has been revealed by the experiments performed on IITD iris database. The best accuracy achieved with the proposed scheme is 97.83% which is at par with state-of-the-art approaches.