Physical object authentication with correlated camera noise
In this paper, we address the problem of physical object authentication based on surface microstructure images. The authentication is based on digital content fingerprints computed from the microstructure images. We analyze the impact of camera distortions which follow additive correlated Gaussian noise. The optimal decision rule is derived and the performance is analyzed in the direct and transformed domain. The theoretical derivations are supported by experimental results obtained for the FAMOS dataset.
Full Text: PDF