New predictor-based schemes for reversible data hiding

As techniques of Reversible Data Hiding based on Prediction-Error Expansion (PEE) have performed excellent, more attentions have been paid to the studies. Many PEE-based schemes increased the embedding capacity and controlled the distortion by employing different predictors. Since predictor is vital to PEE-based schemes, we propose a variable predictor which you can change whenever you find the optimized representation to the prediction for concrete image. We also propose three predictors derived from the variable predictors using Context-Based Seven-Star Map composed by seven neighboring pixels. Experiment results show that the proposed scheme performed as well as (sometimes even better than) Chang’s scheme in capacity-distortion control because of variable parameters and better accuracy of the predictors.