Reversible Data Hiding of Subsampled Model for Edge-Information Prediction

This paper tries to analyze a new framework of lossless information hiding research. Its key is how to get adaptively better difference image architectures for given applications. A unique sampled pattern is introduced and described in term of high-similar interpolation image. Seeking the higher peak value in the difference-image is also our concerns. As a whole algorithm, histogram-difference-expanded based structures are reported. Simulations results demonstrate and verify that our new approach is much effective than the nearest difference expansion method with good generalization performance.