Enhancements of Reversible Data Hiding by Interpolation-Based Prediction with Difference Alteration

Reversible data hiding has become an important research topic in recent years due to the characteristics of ´reversibility´. At the encoder, we embed the secret information into original image, and deliver the marked image to the decoder. At the decoder, with the aid of side information, both the secret information and original image should be separated from marked image to keep the reversibility. Among the techniques in reversible data hiding, by altering the difference values relating to original image seems effective with good performance. Researchers tend to obtain the difference values by producing the predicted image with interpolation, and then by subtracting the original image from the predicted one. By intentionally altering the difference histogram, secret information can be reversibly embedded. With proposed method, enhanced performances in most cases can be observed over relating methods in literature.