2020 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) A Deep Learning Based Approach to Iris Sensor Identification Ananya Zabin LCSEE West Virginia University Morgantown, WV 26505 Email: az0019@mix.wvu.edu Thirimachos Bourlai MILAB, School of Electrical & Computer Engineering University of Georgia Athens, GA 30602 Email: Thirimachos.Bourlai@uga.edu Abstract—An efficient iris sensor identification algorithm can be used in certain forensic applications, i.e. detecting mislabeled iris data at large scale iris datasets, and verifying the validity of the data origin of collected iris datasets that are available to be shared. Such knowledge can potentially increase the overall iris recognition system accuracy by offering the operator the option to match same-sensor or cross-sensor iris images. In either case the knowledge of the origin of the sensor used to collect these data, when not available, or the correction of mislabeled data, is expected to result in higher iris matching accuracy. Another benefit of iris sensor identification is that it can assist in improv- ing the detection of fake iris data, i.e. when knowing the iris sensor, we can apply more appropriate models for fake detection that are tuned for a specific iris sensor. In this paper we propose an efficient deep learning-based iris recognition algorithm that is sensor inter-operable. Our approach utilizes a moderate amount of data and is adaptable to learning rate variations as well as variations of the amount of data used for training per class. Our proposed approach uses a set of iris datasets that include iris images captured at different standoff distances. We are using the original captured, dual eye, or periocular images rather than the iris itself, after detecting, segmenting, and normalizing the iris. Thus, the algorithm is efficient, fast, and less depended on additional algorithmic processes that can add computational complexity. Our proposed process includes transfer learning using iris images of higher quality via the utilization of a set of image quality metrics and achieves close to a hundred percent accuracy after cross-validation. Index Terms—Iris Sensor Identification, CNN, Transfer Learn- ing, AlexNet, GoogLeNet, SGDM, Iris Sensor Inter-operability I. I NTRODUCTION Iris is among the most interesting biometric modalities. Iris recognition systems can be highly accurate as the iris features are unique from person to person. No two human iris patterns are same, not even in twins [1]. In many security related applications, where a highly efficient access control solution is needed, iris recognition systems are preferred to be used either independently or as part of a multi-modal biometric verification system. While large number of iris recognition solutions are avail- able, one of their challenges is that iris images can be captured using different imaging sensors. Thus, the issue of sensor interoperability can emerge, where some users may have (a) Iris Interval (b) Iris Lamp (c) Iris M1 S1 (d) Iris M1 S2 (e) Iris M1 S3 Fig. 1: Large scale iris datasets when captured with multiple sensors - can suffer from cases of iris sensor mislabeling. In this example we show how onerous it is to manually check and verify the validity of the data origin of collected iris datasets when capturing using multiple iris sensors. enrolled using one iris sensor, while matching is performed with their live iris counterpart captured by another sensor. In such a case, we are performing iris cross-sensor matching, which can lower the efficiency of the iris recognition system in terms of its accuracy. This cannot be avoided if we do not know for sure the correct origin and thus, label of the available iris images we have available before matching. Thus, a capability to automatically determine the identity of an iris sensor from labelled or unlabeled iris images can be beneficial in many ways. One of the challenges is that iris images are captured in the visible and near infrared bands via the same or different band-specific sensors. Thus, they differ due to the wavelength used, hardware related features and the illumination used while capturing the images [1], [2]. However, it is not always certain that we will have the correct label of each iris image captured. Another issue is that sensor features may differ even within IEEE/ACM ASONAM 2020, December 7-10, 2020 978-1-7281-1056-1/20/$31.00 © 2020 IEEE 827