Detecção de Ataque de Apresentação em Imagens de Íris utilizando Descritores de Zernike

Georgio S. Colares, Waldir Silva, Celso Carvalho

DOI: 10.14209/sbrt.2025.1571156556

Evento: XLIII Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2025)

Keywords: Íris sintética Classificadores Momentos de Zernike Descritores circulares

Abstract

In this paper, we propose a spoofing detection framework for synthetic iris images applied in security authentication systems. Zernike moments were applied due to some of their properties, such as rotation invariance. We investigated the performance of Zernike descriptors on three classifiers, varying their respective parameters. The experiments resulted in the magnitude of the moments of the grayscale representation providing good overall recognition results on a variety of classifiers. The performance of the MLP, CNN, and ResNet network with Zernike moments obtained, respectively, the accuracy of 79.0%, 90.0%, and 93.0%.

Download