The Research On Robust Technology For Digital Image Watermarking

With the development of computer and network technology, the multimedia informationis more vulnerable to malicious attacks and various external modifications. Copyright ownershave suffered severe violations. It causes the trust crisis of the network media. It has a seriousnegative effect on society. This paper focuses on the research of the robust technology. Tosolve the problem of how to better resist to the conventional attacks, especially againstgeometric attacks, we conduct a thorough study. This paper focuses the research work on thefollowing aspects:1. A robust digital image watermarking algorithm based on local exponent moments isproposed. The algorithm is based on the probability density and the exponential momenttheory. The feature point detector is constructed according to the probability theory. First, itextracts the image feature points. These points are not only uniform distribution but also goodstability. Then according to the autocorrelation matrix, the local feature region is constructed.Finally, combined with the excellent characteristics of the exponential moment, a robustdigital image watermarking based on local exponent moments is proposed. The experimentalresults show that the algorithm has good robustness.2. A robust digital image watermarking algorithm based on quaternion Harmonictransform is proposed. The algorithm combines invariant characteristics of the quaternionHarmonic transform. It selects a part of the stable moments of the quaternion Harmonictransform. Then it takes the method of the quantization index modulation for embedding thewatermark. And finally a robust digital image watermarking algorithm is proposed.Simulation results show that the algorithm not only can resist common attacks and geometricattacks, but also has a better imperceptibility.3. The Bessel K distribution of the nonsubsampled shearlet domain image watermarkingalgorithm based on statistical model is proposed. Combined with the excellent characteristicsof the shearlet and the Bessel K distribution model, it describes the coefficients of the heavytailed distribution optimally. It uses the Bessel K distribution model to describe the shearletcoefficients. The model is constructed based on the optimal detector. And then, the detectionof the digital watermarking algorithm is implemented.