Iris Recognition Using Low-Level CNN Layers Without Training and Single Matching
Published in IEEE Access, 2022
This work proposes an iris recognition method that leverages features extracted from the low-level layers of pretrained convolutional neural networks, without requiring additional training on iris data. The approach incorporates a bit-shifting strategy to improve robustness against pupil dilation, a known challenge in iris biometrics. Single-matching evaluation on standard iris databases demonstrates competitive recognition performance with reduced computational cost.
How to cite
@article{zambrano2022irislowlevel,
author = {Zambrano, Jorge E. and Benalcazar, Daniel P. and Perez, Claudio A. and Bowyer, Kevin W.},
journal = {IEEE Access},
title = {Iris Recognition Using Low-Level CNN Layers Without Training and Single Matching},
year = {2022},
volume = {10},
pages = {41276--41286},
doi = {10.1109/ACCESS.2022.3166910}
}
Recommended citation: J. E. Zambrano, D. P. Benalcazar, C. A. Perez and K. W. Bowyer, "Iris Recognition Using Low-Level CNN Layers Without Training and Single Matching," in IEEE Access, vol. 10, pp. 41276-41286, 2022, doi: 10.1109/ACCESS.2022.3166910.
Download Paper | Download Bibtex