This paper proposes a new approach to solve iris recognition problem. In the segmentation step, the inner boundary of iris was extracted by applying Euclidean distance transform and Level Set method. Then the eyelid was removed by applying Hough transform. SIFT (Scale Invariant Feature Transform) was utilized to extract the iris features for use as the identified characteristics. Since SIFT keypoints are scale-invariant, the iris normalization process is not required. The recognition performances of the proposed system are 96.67% and 94.37% for HCMUS?IRITECH dataset and CASIA 3.0 Iris?Interval dataset, respectively. Also, our iris segmenting method shows better performance than the Integro-differential method and Hough transform method for CASIA dataset, but the proposed method is a little worse than those approaches for HCMUS-IRITECH dataset. The most important merit of the proposed method is that the calculation amount of the method is small compared with previous approaches with reasonal recognition accuracy.