This project developed algorithms for iris localization and iris feature extraction. The iris localization is a model-based approach. The proposed method utilized intensity and edges of image to localize an iris. Comparisons of fitting the detected iris using circle and ellipse are studied. From our experiments, the ellipsoid model yields better segmentation accuracy than the circular model. However, accuracy of the whole iris system is depending not only on the iris localization scheme but also the chosen feature extraction. A new iris feature extraction is also proposed in this report. The proposed method utilizes relative Gabor’s magnitude and phase to represent iris textures. The proposed method yields the best system accuracy (3.19%EER) when an iris is modeled using circle. When comparing our proposed iris feature extraction method with a wellknown Daugman method, the Daugman method yields better system performance. However, our proposed method is more robust. It is less sensitive to changes among iris databases