An image self-recovery approach with variable payload

This paper proposes a self-embedding scheme with variable payload for image protection. We divide the original image into blocks and classify them into three types according to their contents. Different amount of bits are assigned for compressing each type of blocks into the reference bits. We further encode all the reference bits by Random Linear Fountain (RLF) codes and embed them into the original image. During data embedding, different distortions are allowed for the three kinds of blocks. On the receiving side, the hidden bits can be extracted to identify whether the watermarked image is tampered, and the principle content of the tampered blocks can be recovered. With the proposed method, both the quality of the watermarked image and the self-recovered image is improved than the previously published results.