Improving Histogram-based Reversible Information Hiding by an Optimal Weight-based Prediction Scheme

Information hiding has been an important research area for copyright protection in modern multimedia technologies. In this paper, a weight-based prediction scheme is presented to enhance the performance of several reversible histogram-based data hiding approaches. By computing the solution of the least-squares problem, we obtain the optimal set of weights for the neighboring pixels to improve the prediction accuracy of the target pixel across the whole image. The heights of the peak points in the histogram can then be raised to increase the embedding capacity. Experiments of our proposed algorithm have been conducted over several well-known test images. We will show that our proposed method significantly improves the embedding capacity upon several methods and still maintains the quality of stego-images.