A new locally optimum watermark detection using vector-based hidden Markov model in wavelet domain

摘要: The vector-based HMM is used to model the wavelet coefficients of images.A new watermark detector using proposed.Closed-form expression for test statistic derived and experimentally validated.The proposed outperforms other existing detectors.The highly robust against various kinds attacks. Watermark detection a way verifying existence in watermarking scheme copyright protection digital data. Statistical modeling subband has been extensively detection. effectiveness depends directly on how are modeled. It known that hidden Markov (HMM) very powerful statistical describing distribution coefficients, since it capable capturing marginal as well inter-scale cross orientation dependencies coefficients. In this paper, shown gives better fit empirical data comparison with Cauchy, Bessel-K form (BKF) generalized Gaussian (GG) distributions. view this, we propose locally-optimum blind domain. Bayesian framework, closed-form expressions mean variance derived, validated evaluating performance detector. Using number images, evaluated. provides rate higher than provided by detectors designed based Gaussian, BKF or GG distributions also be