{"title":"EfficientNet Based Iris Biometric Recognition Methods with Pupil Positioning by U-Net","authors":"Cheng-Shun Hsiao, Chih-Peng Fan","doi":"10.1109/ICCCI51764.2021.9486782","DOIUrl":null,"url":null,"abstract":"In this work, the deep-learning pupil location based iris recognition methods are studied for biometric authentication. First, by using U-Net, the developed design utilizes the semantic segmentation scheme to locate and extract the region of interest (ROI) of the pupil zone. Based on the located ROI of the pupil zone in the eye image, the iris region can be extracted effectively, and the entered eye image is cut to the small eye image with the ROI of the iris that has just been adjusted. Then the iris features of the cropped eye image are optionally enhanced by adaptive histogram equalization or the Gabor filter process. Finally, the cropped eye image with important iris region is classified by EfficientNet. By using the CASIA v3 database, the proposed deep learning based iris recognition scheme achieves recognition accuracies of up to 98.2%, and the Equal Error Rate (EER) of the proposed design can be close to near 0%.","PeriodicalId":180004,"journal":{"name":"2021 3rd International Conference on Computer Communication and the Internet (ICCCI)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2021-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"4","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 3rd International Conference on Computer Communication and the Internet (ICCCI)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICCCI51764.2021.9486782","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 4
Abstract
In this work, the deep-learning pupil location based iris recognition methods are studied for biometric authentication. First, by using U-Net, the developed design utilizes the semantic segmentation scheme to locate and extract the region of interest (ROI) of the pupil zone. Based on the located ROI of the pupil zone in the eye image, the iris region can be extracted effectively, and the entered eye image is cut to the small eye image with the ROI of the iris that has just been adjusted. Then the iris features of the cropped eye image are optionally enhanced by adaptive histogram equalization or the Gabor filter process. Finally, the cropped eye image with important iris region is classified by EfficientNet. By using the CASIA v3 database, the proposed deep learning based iris recognition scheme achieves recognition accuracies of up to 98.2%, and the Equal Error Rate (EER) of the proposed design can be close to near 0%.