High Fidelity Reversible information Steganography in Images using Difference Expansion and Appropriate Region Selection

Nowadays, reversible embedding of data in images has received a lot of attention. With reversibility, the original digital content can be fully retrieved. In this paper, a reversible data embedding method for digital images is presented, which seeks to achieve high embedding capacity and reduce distortion in digital images. In this paper, a reversible watermarking method is proposed based on the selection of the appropriate watermarking region in which the data is embedded in the host image with a slight change in brightness in the pixel domain. It has been shown that the modified pixels can return to their original values during the reconstruction process. The effect of compensation in this method can significantly improve the quality of the specified image. To improve image quality, we preserve non-embedded pixels and use flag bits to distinguish between embedded pixels and non-embedded pixels. Using this pixel selection method, a higher quality watermarked image that the capabilities was confirmed by extensive experiments. In addition, in this paper a method of selecting the watermark region based on entropy criteria is proposed. In the performed experiments to a sample image (baboon), the criterion for measuring Watson distance before using the entropy model was 27.8 and after using the entropy model was 7.9, which shows that the proposed method can preserve the quality of the image, better than the competitor's reversible watermarking methods.