IRIS IDENTIFICATION WITH ANN AND DAUGMAN'S ALGORITHM

Abstract

Among the array of biometric identification techniques, iris recognition stands out as one of the most advanced and efficient methods. It relies on pattern recognition to discern unique and easily identifiable patterns within the iris, facilitating precise identification of individuals. This approach offers superior accuracy and results compared to other methods. Given the increasing instances of security breaches and authentication fraud, implementing a robust biometric system is essential.The proposed research utilizes Daugman's method for iris localization, leveraging its integro-differential operator capability to effectively separate regular shapes and reduce noise. Daugman's algorithm is particularly well-suited for iris localization due to these features. Following iris localization, feature extraction identifies consistent and unique aspects of an iris image. Various statistical measures such as mean, standard deviation, entropy, root mean square, smoothness, kurtosis, energy, homogeneity, contrast, and variance are computed in the research. These features exhibit distinct behavior in response to different iris images, though some overlap in values may occur.Subsequently, these features are inputted into a neural network utilizing the Levenberg-Marquardt backpropagation training algorithm. The neural network is trained using feature values extracted from authorized photos and then tested for accuracy. The traditional MMU database is integrated into the system design. Compared to previous techniques utilizing the same database, the proposed method achieves a higher degree of accuracy, specifically reaching 99.7 percent accuracy.