Algorithms for Efficient Segmentation of Non-ideal Iris Images: Optimization Based Segmentation Techniques For Iris Recognition System

Among the biometric technologies available, the iris biometric technology is the most accurate modality, because iris complex random patterns are unique and stable, they do not change throughout a person's lifetime. The Iris recognition is based on the fact that the human iris contains unique features and even genetically identical individuals have entirely independent iris textures. Iris segmentation is an essential step because the actual discriminating information will be present within the iris patterns. Therefore, it is plausible that the initial step in implementing an iris recognition system is separating the iris from irrelevant parts of an eye image, which are of no importance. A pre- segmentation using Otsu’s multilevel thresholding and variants of fuzzy c-means (FCM) based on IPSO (improved PSO) and IDSA (improved differential search algorithm) has not been investigated in the literature. The recognition accuracy is affected by noise artefacts that are included during the capturing of iris images. This encourages to effective implementation and accurate pre-segmentation in the recognition framework.