Reversible Generating Information Hiding Method Through Quadratic Clustering Coding

Abstract

In texture generating-based information hiding and mosaic-based information hiding, there always exist stitching traces. To avoid these stitching traces and produce meaningful image and further improve attack abilities, a reversible generating information hiding method through quadratic clustering coding is proposed. The proposed method takes random two-value noise patches as candidate sample patches and selects coding sample patches for representing two-value secret information through quadratic clustering coding. The stego cover image is produced by placing coding sample patches according to a random two-value reference image and embedding positions. In extraction, the coding sample patches are generated by quadratic clustering coding, and then the two-value secret information is extracted by combining the two-value reference image and random embedding positions. The proposed method can generate meaningful images with fantastic visual quality without any splicing trace or modification embedding. The cover image can be completely restored to the original image without loss of quality. The proposed method has wonderful anti-attack abilities, and the embedding and extraction process is entirely dependent on the user key, therefore it has high security.

Keywords

• generating information hiding
• mosaic
• quadratic clustering coding
• reversible
• no modification