High Capacity Reversible Watermarking

This paper proposes a high capacity low cost reversible watermarking scheme. Image is partitioned in pairs of pixels. In order to satisfy a congruence equation, the pairs obeying some simple constraints are transformed. By adding some correction data, the not transformed pixel pairs are enforced to obey the same equation, too. The payload and the correction data are embedded into the transformed pixels by simple additions. After watermark insertion, only the not transformed pixel pairs satisfy the congruence equation. This allows transformed pairs localization, payload and correction data extraction and, finally, exact recovery of the original image. The proposed scheme can provide, in a single pass, bit-rates greater than 1 bit per pixel without any additional data compression.