Many watermarking algorithms in spatial domain lack of a mathematical derivation of the detector of digital watermarks. On this subject this paper proposes a statistical model which allows both blind digital watermark detection and blind threshold computing. This model with the assistance of other pre-processing tools such as pre-detection filtering and geometrical distortions compensation offers reliable fast and accurate watermark detection when strong attacks such as high power noise corruption, geometrical attacks and so on.
展开▼
