Effects of Segmentation Routine and Acquisition Environment on Iris Recognition

by Tanya H. Peters Every year we see a growing use of iris recognition, with it now utilized as a means of border control in a number of countries, including the United Kingdom, Canada, and the United Arab Emirates. As this technology becomes more common and more relied upon, the importance of algorithms that can identify subjects in a robust, consistent, and accurate manner becomes all-important. Working with a collection of over 20,000 iris images captured in 2008, we determine optimal parameters for different elements of the texture encoding process. Additional work was done to improve the segmentation process, both to handle the introduction of images captured with the LG 4000 and to improve iris segmentation and eyelid masking. Furthermore, we study the relative biometric performance of images captured with the LG 2200 based on which of three illuminants were used to light the eye in each image, and determine the sameand cross-sensor performance of the LG 2200 compared with the LG 4000.