This website requires JavaScript.
DOI: 10.1109/ETCM56276.2022.9935749

Artificial Pupil Dilation for Data Augmentation in Iris Semantic Segmentation

Daniel P. BenalcazarDavid A. BenalcazarAndres Valenzuela
Dec 2022
Biometrics is the science of identifying an individual based on theirintrinsic anatomical or behavioural characteristics, such as fingerprints,face, iris, gait, and voice. Iris recognition is one of the most successfulmethods because it exploits the rich texture of the human iris, which is uniqueeven for twins and does not degrade with age. Modern approaches to irisrecognition utilize deep learning to segment the valid portion of the iris fromthe rest of the eye, so it can then be encoded, stored and compared. This paperaims to improve the accuracy of iris semantic segmentation systems byintroducing a novel data augmentation technique. Our method can transform aniris image with a certain dilation level into any desired dilation level, thusaugmenting the variability and number of training examples from a smalldataset. The proposed method is fast and does not require training. The resultsindicate that our data augmentation method can improve segmentation accuracy upto 15% for images with high pupil dilation, which creates a more reliable irisrecognition pipeline, even under extreme dilation.