Research On Robust Reversible Watermarking Based On Image Moment

Robust Reversible Watermarking(RRW)is a very novel and valuable research direction in the field of data hiding in recent years.Whether it is used as copyright certification of images or applied to the high-fidelity field,it has a high development prospect.However,the existing RRW schemes are less robust against geometric transformations,common attacks,and joint attacks,and the amount of watermark embedding is low for services with high-capacity requirements.In order to solve these problems,in this paper,two RRW schemes is proposed to improve the robustness and embedding capacity of the watermark,respectively:1.An efficient and robust reversible watermarking scheme based on Zernike moments and integer wavelet transform.First,the original image is transformed by integer wavelet to obtain the low-frequency region,and then the Zernike moment of this region is calculated.Next,the watermark information is embedded into the magnitude of Zernike moments.Errors generated when embedding the watermark are reversibly embedded into the pixels of the image.Afterwards,the receiver can use the errors extracted from the watermarked image to restore the original image.Experimental results prove that the watermarked image of the proposed scheme can not only completely restore the original image without attack,but also improve the robustness against geometric attacks,image processing operations and joint attacks.Compared with the state-of-the-art RRW methods,the proposed scheme improves the performance against noise and compression attacks,such as the Bit Error Rates(BERs)against JPEG2000 and Gaussian noise are reduced by about 4% on average.The BERs of resistance to rotation and JPEG are 0,and this scheme can resist joint geometric and compression attacks simultaneously.2.Adaptive high-capacity robust reversible watermarking scheme.RRW simultaneously improves watermark imperceptibility and embedding capacity under the condition of reversibility and robustness,which is still a big challenge nowadays.In this paper,the imperceptibility and embedding capacity are improved by an adaptive normalization method and an optimized embedding method.Firstly,the magnitude of the pseudo-Zernike moment is generated by calculating the low-frequency coefficients of the original image,and the qualified moment is manipulated by the adaptive normalization method,then the robust watermark is embedded into the appropriate pseudo-Zernike moment by the new embedding method.Finally the pseudo-Zernike moments of the watermark are reconstructed to generate the watermarked image.The errors generated by the reconstruction is retained,at the same time,the errors between the watermark image and the original image and the hash value form auxiliary information embedded in the watermark image,aiming to achieve reversibility without attack.Experimental results show that the proposed scheme is robust to common signal processing,including noise,JPEG,JPEG2000 and geometric transformations including rotation and scaling,which outperforms existing schemes.At the same time,compared with the existing RRW scheme,it also improves the embedding capacity while being better than the robustness.