Annular Iris Recognition Using SURF Hunny Mehrotra 1 , Banshidhar Majhi 1 , and Phalguni Gupta 2 1 National Institute of Technology Rourkela, Rourkela 769008 2 Indian Institute of Technology Kanpur, Kanpur 208016 {hunny,bmajhi}@nitrkl.ac.in, pg@iitk.ac.in Abstract. This paper proposes an iris recognition system which can handle efficiently the problem of rotation, scaling, change in gaze of in- dividual and partial occlusions that are inherent to non-restrictive iris imaging system. In addition to this, traditional iris normalisation ap- proach deforms texture features linearly due to change in camera to eye distance or non-uniform illumination. To overcome the effect of aliasing features are extracted directly from annular region of iris using Speeded Up Robust Features (SURF). These features are invariant to transfor- mations and occlusion. The system is tested on BATH, CASIA and IITK databases and is showing an accuracy of more than 97%. From the re- sults it is inferred that local features from annular iris gives much better accuracy for poor quality images in comparison to normalised iris. Keywords: Annular Iris Region, SURF, Local Features, Occlusion, Transformation. 1 Introduction Iris is gaining added attention since last few decades due to accuracy, reliabil- ity and speed. Iris image acquisition is a highly restrictive process that requires cooperative and well trained audience. Acquired image is localised using pupil and iris boundary. Further, segmented iris region is normalised to form a rect- angular image for matching. However there are several issues to be taken into consideration prior to feature extraction and matching. Some of these issues are worth to mention. During image acquisition, there may be some tilt in head or change in gaze of an individual. Thus the features are transformed circularly in Cartesian plane. Again iris image may be occluded by lower and upper eyelids that makes images inappropriate for matching. Another issue is that texture pattern in iris undergoes linear deformation due to expansion and contraction of pupil under non-uniform illumination. Further mapping from Cartesian to polar plane creates the effect of aliasing that loses significant texture details that are relevant from recognition point of view. There exists several global feature extraction techniques in iris. In [1] Gaus- sian filter at multiple scales is used to extract features. Iris coding method based on differences of Discrete Cosine Transform (DCT) coefficients of overlapped an- gular patches from normalised iris images is presented in [2]. In [3] wavelet trans- form is applied on circular bands of iris and zero crossing representation is used S. Chaudhury et al. (Eds.): PReMI 2009, LNCS 5909, pp. 464–469, 2009. c  Springer-Verlag Berlin Heidelberg 2009