Reversible Data Hiding with Selected Directional Context Based Prediction Using 8-Neighborhood
Ravi Uyyala, Rajarshi Pal
Abstract
In this paper, a novel reversible data hiding technique using a selected directional context based prediction has been suggested. In the proposed prediction strategy, the predicted values are obtained across vertical, horizontal, diagonal and anti-diagonal directions using pixel values in 8-neighborhood of a pixel. The final predicted value is obtained as the average of predicted values for a set of selected directional contexts. These directions are selected using the diversities in the neighborhood pixel values in the respective directions. Finally, the prediction error in an image pixel is used to insert the secret data in the pixel using a prediction error histogram bin shifting method. An adaptive histogram bin shifting method is introduced in this paper to adaptively embed one or two bits of data in a pixel. The proposed selected directional context based prediction with adaptive histogram bin shifting has outperformed several state- of-the-art reversible data hiding techniques.