A NOVEL DOUBLE ENCODING SCHEME FOR LOW DISTORTION ADAPTIVE REVERSIBLE WATERMARKING

This paper proposes a low-distortion transform for prediction-error expansion reversible watermarking. The transform is derived by taking a simple linear predictor and by embedding the expanded prediction error not only into the present pixel as well as into its expectation setting. The inserting guarantees the minimization of the square error introduced by the watermarking. The proposed transform introduces less mutilation than the established forecast error development for complex indicators, for example, median edge detector or the gradient-adjusted predictor. Reversible watermarking algorithms dependent on the proposed transform is to be examined. Experimental results are given.