Robust detection of DCT-domain watermarks using alpha-stable models

This paper presents a robust watermark detector in the Discrete Cosine Transform (DCT) domain. First, the watermark detection problem is formulated as a binary hypothesis testing problem according to the statistical inference. Then, the alternative current DCT coefficients are statistically modeled by symmetric alpha-stable distributions. More specifically, the scale mixture of Gaussians as an analytical approximation of alpha-stable distribution is employed. Finally, based on the theory of statistical inference and weak signal detection in non-gaussian noise, a robust blind detection algorithm is derived. And the validity of the detector is also tested.