Font Size: a A A

DWT-Domain Blind Watermarking Algorithm Based On Generalized Gaussian Distribution Model

Posted on:2009-02-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y C XuFull Text:PDF
GTID:2178360245474751Subject:Detection Technology and Automation
Abstract/Summary:PDF Full Text Request
Watermark detection plays a crucial role in digital watermarking. It has traditionally been tackled using correlation-based techniques. However, correlation-based detection is not the optimum choice either when the host media doesn't follow a Gaussian distribution or when the watermarking is not embedded in the host media in an additive way. This paper addresses the problem of DWT (discrete wavelet transform) domain multiplicative watermark detection for digital images. First, GGD(Generalized Gaussian distributions) are applied to statistically model the DWT coefficients of the original image. Then two parameters of GGD are modeled with the minimum of the relative entropy principle.Two digital watermark algorithms are also proposed. In the first embedding way, this paper presents a method of non-adaptive multiplicative watermark embedding, based on Cox's theory, one of which we can choose different embedded intensity factor and the different coefficient for different images; In the second way, taking account of human visual system model, an adaptive robust multiplicative watermark algorithm is proposed, which is decided by the location of watermark embedding and the intensity. Then the blind watermark detector of multiplicative hiding is introduced, based on the GGD model. The threshold of the blind watermark detector is also analysised, using the theory of least-squares method.Four standards of gray images are simulated on the robustness of digital watermarking, using the tools of matlab. In the first experiment, non-adaptive algorithm and adaptive algorithm is tested in the same detection algorithm and the same probability of false alarm; In the second experiment, two detection algorithm is tested, one of which is based on GGD model and the other one is based on Gaussian distribution. Experimental results indicate the superiority of the GGD-based detector in the case that the adapted multiplicative embedding strengths are unknown to the detector.
Keywords/Search Tags:digital watermarking, generalized Gaussian distributions, multiplicative embedding, least-squares method
PDF Full Text Request
Related items