Classification of Genetically Identical Left and Right Irises Using a Convolutional Neural Network

• DOI: 10.3390/electronics8101109
• Publication Date: 2019-10-01T15:11:16Z
• Authors: Beihua Fang; Yuanfu Lu; Zhisheng Zhou; Zhihui Li; Yuwen Yan; Linfeng Yang; Guohua Jiao; Guangyuan Li
• Keywords: Iris Recognition; IRIS (biosensor); Identification

As one of the most reliable biometric identification techniques, iris recognition has focused on differences in textures without considering similarities. In this work, we investigate correlation between left and right irises an individual using a VGG16 convolutional neural network. Experimental results with two independent datasets show that remarkably high classification accuracy larger than 94% can be achieved when identifying if (left right) are from same or different individuals. This exciting finding suggests similarities genetically identical indistinguishable traditional Daugman’s approaches detected by deep learning. We expect work will shed light further studies and/or other identifiers related individuals, which would find potential applications criminal investigations.