| Among the traditional watermarking technique, there are robust watermarking and fragile/semi-fragile watermarking. During the process of embedding watermark, it’s inevitable that the process will bring image distortion to some extent. Upon most occasions, it’s not easy to find the distortion for image users, and there is just a little impact on practical use. However, the requirement of image quality is really demanding in the field of medical and military. Thus, the distortion is intolerant in these fields. The problems hadn’t been solved until the innovative reversible watermarking technique was published. The coming of reversible watermarking makes it possible that the distortion will be completely eliminated and the image carrier will be recovered perfectly.In recent years, scientific research personnel proposed a new type of watermarking technique which can enhance the robustness of watermark, called robust reversible watermarking. As is well known, images are easy to be attacked by compressed, uncompressed and noise. When a watermarked image is not under attack, robust watermark in the image can be extracted and the image carrier can be recovered perfectly. However, when a watermarked image is suffered from malicious attacking or conventional editing, the robust watermark can still be extracted, but the reversibility is lost which means the image carrier can’t be recovered completely.In this article, on the basis of the previous research on robust reversible watermarking, we come up with two sorts of improved robust reversible watermarking algorithm. They upgrade the robustness of watermark and the visibility of images.1. Independent embedding domain based robust reversible watermarking method: the image carrier is divided into two independent parts, which are high frequency region and low frequency region, by wavelet transform. The classical robust watermarking method ’patchwork’ is used to embed watermark for ensuring highly robust in low frequency region and there will be residual error between the original image and the watermarked image after the watermark is embedded. The residual error should be embedded into the high frequency region to ensure the reversibility in the algorithm. At last, two independent regions are combined by the inverse wavelet transform as a watermarked image.2. The frequency domain JND based robust reversible watermarking method: the subjective assessment of Human vision is one of the key factors to evaluate image quality. Just noticeable difference in the human vision system can be used to measure the texture complexity of the different areas in an image, and also can be set to be the threshold value of modifying pixel values. Thus, a combination of robust reversible watermarking and JND is proposed for better robustness and better image vision. Firstly, the original image is divided into two independent parts, including low frequency region and high frequency region. And then, according to the embedding capacity, the low frequency region is divided into two regions of texture region and smooth region by JND. Secondly, we embed watermark by means of robust watermarking algorithm which concentrate on the textured area. This can not only improve the robustness of the watermark, but also can make the embedded watermark image more in line with the human visual evaluation criteria. Finally, a reversible watermarking technology will be used to restore the original information is embedded in the high frequency region.The two algorithms proposed in this paper are robust reversible algorithm for gray scale images. Through simulation experiments, the results show that the two algorithms are robust and reversible. Using the Experimental data and the previous performance of the robust reversible watermarking(MLW and GHS) for comparison, the robustness is greatly improved, and the visual subjective experience is better. |
