Multiwavelet-Based Digital Image Watermarking

In this paper, starting from the information hiding, and the depth of analysis and research to digital watermarking technology, a multibit, mul-tiplicative, spread spectrum watermarking using the discrete multiwavelet (including unbalanced and balanced multiwavelet) transform is presented. In this paper, the following work is completed:(1) A multibit, multiplicative, spread spectrum watermarking is de-signed using the discrete multiwavelet transform, and the watermark em-bedding mechanism of the system is adaptive. Correspondingly, an adap-tive optimum decoding is derived using maximum likelihood (ML) detec-tion rule, and assuming a statistic model based on generalized-Gaussian distribution (GGD) for multiwavelet coefficients of the detail subband. The wateramark bits can be extracted without reference to the original image, namely that is blind.(2) Performance improvement with respect to existing algorithm is obtained by means of a new just perceptual weighting (JPW) model. The new model incorporates various masking effects of human visual percep-tion by taking into account the eye's sensitivity to noise changes depend- ing on spatial frequency, luminance and texture of all the image subbands. In contrast to conventional JND (Just Noticeable Difference) threshold model, JPW describing minimum perceptual sensitivity weighting to noise changes, is fitter for nonadditive watermarking. Specifically, watermark-ing strength is adaptively adjusted to obtain minimum perceptual distor-tion by employing the JPW model.(3) Multiwavelet characteristics that is indicative of its merits play an important role in the field of image processing. In this paper, the impact of multiwavelet characteristics on proposed watermarking scheme is ana-lyzed, and the reference standard choosing multiwavelet filters is given in watermarking system.(4) The experimental results show that proposed JPW model can im-prove the quality of the watermarked image and give more robustness of the watermark as compared with a variety of state-of-the-art algorithms.