Reversible Watermarking Algorithms Based On Image Interpolation

The distortion of the original carrier, which is caused by reversible watermarking, is repairable. That's to say, the process of embedding watermark is reversible and it is not only can extract the watermark but also can restore the original carrier without distortion in extraction side. Medical image, military media and legal evidence require the original carrier image cannot have any changes, because small changes can lead to the fault diagnosis of condition, error analysis of military and falsity of legal evidence. Reversible watermarking has realized the copyright protection and integrity authentication of these kinds of carriers.This thesis takes image interpolation as basis, points out the disadvantages of existing reversible watermarking algorithm, deeply analyzes the characteristics of image interpolation, the theory of histogram translation and the integer transform algorithm, reversible watermarking algorithm, which can realize the large capacity and low distortion is mainly researched.The main contents of the thesis include:(1) With the researching and analyzing of existing reversible watermarking algorithm in image interpolation, its reversible process of embedding watermark is designed with the characteristics of the currently used interpolation method, embed the watermark into interpolation pixel. These algorithms have following deficiencies, embedding process is poor in portability and only embed watermark in the interpolation pixel, which can't really play a role of copyright protection.(2) A reversible watermarking algorithm based on image interpolation and histogram translation is proposed. Aiming at the low embedding capacity of the existing histogram translation, using interpolation image, the pixels which are used in calculating the diamond predictive value, are calculated by the common original pixels, so these pixels having a strong correlation. Using this feature into the prediction error histogram can make the prediction error value distribution is more concentrated, and improve the improve the embedding capacity. The proposed method has increased the embedding capacity by 3 times, and it has better invisibility.(3) A reversible watermarking algorithm based on image interpolation and integer transform is proposed. Integer transform algorithm will lead serious distortion in image when having a large embedding capacity. Aiming at the problem, the sorted interpolation pixels are used into integer transform to decrease the changing of pixel, combining the characters such as the original pixel values do not change after image interpolation and the original pixel position are fixed whichever interpolation algorithm is used, the sort subscript can be gain according to the original pixel in the extract end, so it's no need to use additional location plan to record sorting result. The proposed method not only can ensure the feature of high embedding capacity, but also can decrease the degree of image distortion down to the visual range.