Publications
2018
1.
Apap, Adrian; Fernández-Robles, Laura; Azzopardi, George
Person Identification with Retinal Fundus Biometric Analysis Using COSFIRE Filters Book Section
En: Applications of Intelligent Systems, pp. 10–18, IOS Press, 2018.
Resumen | Enlaces | BibTeX | Etiquetas: Biometric Systems, COSFIRE Filters, Person Identification, Retinal Fundus Images
@incollection{apap_person_2018,
title = {Person Identification with Retinal Fundus Biometric Analysis Using COSFIRE Filters},
author = {Adrian Apap and Laura Fernández-Robles and George Azzopardi},
url = {https://ebooks.iospress.nl/volumearticle/50862},
year = {2018},
date = {2018-01-01},
booktitle = {Applications of Intelligent Systems},
pages = {10–18},
publisher = {IOS Press},
abstract = {This paper presents a method for person identification using retinal fundus images, leveraging the structure of the vessel tree as a biometric feature. Unlike conventional methods, which rely on vessel segmentation, the proposed approach matches the spatial arrangement of five bifurcations between two images. This method is more efficient and robust to noisy images, including those with pathologies. Using a hierarchical approach of trainable COSFIRE filters, the technique achieves an accuracy of 100% on both the Retinal Identification Database (RIDB) and VARIA datasets. The method is adaptable to other vision-based biometric systems like fingerprint and palmprint recognition.},
keywords = {Biometric Systems, COSFIRE Filters, Person Identification, Retinal Fundus Images},
pubstate = {published},
tppubtype = {incollection}
}
This paper presents a method for person identification using retinal fundus images, leveraging the structure of the vessel tree as a biometric feature. Unlike conventional methods, which rely on vessel segmentation, the proposed approach matches the spatial arrangement of five bifurcations between two images. This method is more efficient and robust to noisy images, including those with pathologies. Using a hierarchical approach of trainable COSFIRE filters, the technique achieves an accuracy of 100% on both the Retinal Identification Database (RIDB) and VARIA datasets. The method is adaptable to other vision-based biometric systems like fingerprint and palmprint recognition.