A Two-Stage Robust Reversible Watermarking Algorithm

In this paper, a two-stage robust reversible watermarking algorithm is proposed. In the first stage, the robust watermark is inserted into the low-order Tchebichef moments (TMs) of the selected blocks by using a modified distortion-compensated quantized exponential modulation (DC-QIM) technique and the round-off error is computed by pre-recovering the image blocks. In the second stage, the compensation information is embedded as a reversible watermark into the unprocessed image blocks to achieve reversibility in case of no attacks. Resynchronization is used to restore the position of the image block before the watermark is extracted, which effectively improves the robustness against geometric attacks. Experiments show that this scheme is not only robust to common signal processing (CSP) but also to geometric attacks. Compared with the current state-of-the-art schemes the embedding capacity is increased and computational complexity is decreased.