1 results listed
In this paper, we propose an anti-spoofing method
that employs the fusion of various full-reference and no-reference
image quality assessment techniques to detect fake and real ear
images presented to biometrics systems under print attacks. In this
context, full-reference image quality assessment measures such as
Error Sensitivity Measures, Pixel Difference Measures,
Correlation-Based Measures, Edge-Based Measures, Spectral
Distance Measures, Gradient-Based Measures, Structural
Similarity Measures and Information Theoretic Measures are
used. Additionally, no-reference image quality assessment
measures such as Distortion Specific Measures, Training Based
Measures and Natural Scene Statistics Measures are implemented
to distinguish fake and real ear images. A comparative analysis of
the performance of these quality metrics and the proposed method
using decision-level fusion of all aforementioned measures are
performed. The experimental results are presented using AMI and
UBEAR ear databases by creating print attack counterparts of the
ear images used in these databases.
International Conference on Advanced Technologies, Computer Engineering and Science
ICATCES
İmren TOPRAK
Önsen Toygar