This paper presents a general framework for the reversible watermarking based on asymmetric histogram shifting of prediction error, which is inspired by reversible watermarking of prediction error. Different from the conventional algorithms using single-prediction scheme to create symmetric histogram, the proposed method employs a multi-prediction scheme, which calculates multiple prediction values for the pixels. Then, the suitable value would be selected by two dual asymmetric selection functions to construct two asymmetric error histograms. Finally, the watermark is embedded in the two error histograms separately utilizing a complementary embedding strategy. The proposed framework provides a new perspective for the research of reversible watermarking, which brings about many benefits for the information security.
General Framework of Reversible Watermarking Based on Asymmetric Histogram Shifting of Prediction Error