Robust Reversible Watermarking Algorithm Against Geometric Attack

As an indispensable embranchment of information security profession,digital watermarking technology can be used for the integrity authentication and copyright protection of digital images,and play a critical role in medical,military,judicial authentication and other fields.Robust watermarking technology uses robust characteristics of digital multimedia to embed watermark information into multimedia to realize copyright protection,but the embedding of watermark will cause permanent distortion of the original image.Reversible watermarking techniques solves the problem of permanent distortion.The watermarking image is reversible when it is not attacked,but its robustness is low and it cannot resist attack.Robust reversible watermarking technology is both robust and reversible.If the watermark image is not attacked,it is reversible.If the watermark image is attacked,the watermark information can be extracted to protect the copyright.Most of the current robust reversible watermarking technologies are resistant to signal attacks,but have no resistance to common geometric attacks that are more common in practical applications.Through the research of the previous robust reversible watermarking schemes,this thesis proposes two robust reversible watermarking schemes that are robust to geometric distortion:(1)A robust reversible watermarking technique based on polar harmonic fourier moments(PHFM)is proposed.Firstly,the properties of geometric invariant moments are analyzed and the characteristics of PHFM are studied.Secondly,the PHFM is applied to the carrier image,and the robust watermark is embedded in the PHFM.Then the compensation information needed to restore the carrier image is calculated and embedded in the independent embedding region of the robust watermark.Finally,the aspect ratio change coefficient is introduced to resist mirror image flip attack and aspect ratio change attack.Experimental results show that compared with the existing algorithms,the watermark image quality is improved from 38.24 d B to 40.52 d B when the embedding capacity is 128 bit and the quantization step size is optimal.And it has strong robustness to common signal attack and geometric attack.(2)A robust reversible watermarking algorithm based on local region is proposed to resist cropping attack.Firstly,the robust and reversible watermarking regions are obtained by using the feature points and texture information of the carrier image.Secondly,using the polar harmonic fourier transform and two-stage watermarking algorithm,the watermark is embedded into the robust watermark region to obtain the intermediate watermark image block,and the quantization error and watermark error are reversibly embedded into the independent embedding region of the robust watermark to obtain the robust watermark image block.Then,the quantization error and watermark error are used to pre-restore the intermediate watermark image block.Finally,the difference between the original image and the pre-restored image and the hash value used for image authentication are embedded into the reversible watermark region as compensation information to obtain the final watermark image.Experimental results show that compared with existing algorithms,this scheme can resist signal attack and small angle rotation attack,and has strong robustness to cropping attack.In order to verify the effectiveness of the proposed robust reversible watermarking algorithm based on extreme harmonic Fourier moments,a Graphical User Interface(GUI)simulation system was designed to show the performance of the algorithm more intuitively.