A High Payload Histogram-Based Reversible Wartermarking Using Linear Prediction

Reversible watermarking is an efficient method of embedding secret information into digital work for the purpose of copyright protection, certification, tracking, etc. This paper presents a high payload reversible watermarking algorithm for digital images using linear prediction. Prediction error and histogram shifting are used for embedding data. Different from previous methods, we utilize linear prediction to obtain the initial prediction, which is then modified by calculating the variance of adjacent pixels to get a more accurate prediction image. Since the prediction-error histogram is more concentrated, the proposed method can achieve larger capacity and less distortion by using histogram shifting. The experimental results indicate the validity of the proposed method compared with other schemes.