A robust-transparent watermarking scheme based on perceptual modeling

Abstract

In this paper, we present a generic watermarking framework that employs perceptual modeling to embed a watermark in such a way that it cannot be perceived. Our approach consists in building computational models which take into account the most common properties of the HVS (Human Visual System) that can be exploited for watermarking. We describe two ways a perceptual model can be incorporated into a watermarking system to control the imperceptibility of the watermark: the implicit and the explicit embedding scheme. Both of schemes are designed in the pyramid transform domain proposed by Burt and Adelson. In the implicit embedding, the watermark is only inserted into imperceptible regions defined using a visibility map. In the explicit embedding, the watermark is embedded in all Laplacian coefficients with strength determined by a JND (Just-Noticeable-Distortion) threshold. The obtained results are twofold. Firstly, the explicit embedding performs better than the implicit one. Secondly, watermarking schemes based on HVS modeling have advantages over schemes that do not take perceptual aspect into consideration. These advantages are demonstrated in terms of imperceptibility and robustness via various experiments.