Computer Vision Research Group

IRIS Recognition under Noisy Conditions

Current research in Iris Recognition focuses on efficient iris localization and more discriminative feature descriptions for noisy iris images. Our research focuses on acheiving robust iris encoding and matching on highly noisy data of the UBIRIS.v2 database.

We have experimented with various new features for iris encoding and made use of different matching strategies to achieve maximum inter class dissimilarity and minimum intra-class dissimilarity. We performed our experiments on the set of training images provided for the contest. Segmented images were normalized according to Daugman's method. We evaluated use of features like LESH and Gabor Phase and Magnitude Patterns to encode irises and matched them using nearest neighbor classifier and Bayesian frameworks. To cater for the noise we assign weights to the noisy regions in the image.



Recent News
© 2010 Department of Electrical Engineering, CIIT Lahore  |  Terms and Conditions