| Abstract: | ABSTRACT Perceptual watermarking should always take full advantage of the results from human visual system (HVS) studies. Based upon visual saliency's modulation effect which incorporates visual attention's influence on visual sensitivity, this paper proposes an integrated visual saliency modulated HVS model guided least significant bit (LSB) watermarking approach for copyright protection. The proposed algorithm can embed more information into visually non-belligerent areas within the host image, determined by the visual attention model. The region of uninterested (ROU) indicates the most unimportant visible aspects of an image, so distortion within these areas will be less noticeable to any viewer. To resist common signal processing attacks, for each ROU, an improved quantization method is employed to embed the copyright information. Experimental results demonstrate that the proposed method is more robust to various image impairments while minimum effect on the image quality as verified by objective evaluations while compared with the state-of-the-art watermarking schemes. Moreover, we have evaluated the theory of Kurtosis testing to the steganalysis of the host image. |
