This paper aims at reducing the shifting distortion of histogram shifting reversible data hiding method. Instead of calculating symmetrically the prediction value as were done in other schemes, based on the gradient-adjusted predictor (GAP), a directed-prediction scheme, which includes two asymmetric predictors-the right and left GAPs, is designed to predict asymmetrically pixel value. Then two asymmetric error histograms, with right and left-skewness, are constructed by gathering the directed prediction errors, which effectively reduces the amount of pixels on the shifted side of the error histograms. Moreover, the optimal embedding points and thresholds are calculated by defining an evaluation index of the shifting distortion. Experimental results validate the effectiveness of the proposed method and demonstrate that it outperforms several previous methods in terms of payload and image quality.
- Asymmetric-histogram shifting
- Directed-prediction scheme
- Gradient-adjusted predictor (GAP)
- Prediction errors
- Reversible data hiding