Research Of Digital Image Robust Watermarking Algorithm Based On Transform Domain

The convenience and rapidity of multimedia transmission on the open networks pose difficulties and challenges to the multimedia copyright and content protection.As one of the copyright protections,digital watermarking embeds the watermarks associated with the owner's identity and multimedia contents,and indicates the ownership,integrity,and historical operation of covers by verifying the similarity between the extracted watermarks and original watermarks.Compared with other multimedia protection methods,digital watermarking has the advantages of the relevance with covers,rich verifiable information,and convenient manipulation.To provide reliable copyright protection for images,the robust watermarking not only is resistant against common image processing and geometric attacks,but also distinguishes the true watermarks from faked watermarks.According to the above requirements and concerns of the imperceptibility,security,capacity etc.,this dissertation studies the robust watermarking for digital images from the following aspects:(1)In view of false positive problems occurred in the algorithms based on singular value robustness with gray-scale image watermarks,this dissertation digs out the reason that results in the false positive problems and gives a side information forgery method for the algorithm proposed by Makbol et al.in Information Sciences.The experimental results indicates that the algorithm offers an untrusted copyright proof.To solve the false positive problem of the watermarking algorithms based on singular value decomposition(SVD)and optimize the quantization step selection method of quantization index modulation(QIM),this dissertation utilizes the stability of the DWT low-frequency coefficients and singular vectors to propose an adaptive robust watermarking algorithm with ensured invisibility.Firstly,by analyzing the relationship among the modification of singular vector elements,modification of image pixels and peak signal-to-noise ratio(PSNR),an adaptive quantization step selection strategy is designed.Then the low-frequency coefficients of DWT are divided into non-overlapping blocks and SVD is applied on each block.The difference of the singular vector elements is quantized based on the proposed quantization step selection strategy.Finally,the right singular matrix is modified to achieve quality compensation.Compared with the fixed quantization steps or metaheuristic-based quantization step selection,the proposed quantization step selection strategy not only ensures the quality of the watermarked images,but also has significant advantages on the computational time.Experimental results show that the proposed algorithm has good performance on invisibility,robustness,and time cost.(2)To reduce the discretization errors of the continuous orthogonal moment calculation for digital images,a secondary grid division-based moment calculation method is proposed,then a new watermarking algorithm based on generic polar complex exponential transforms(GPCET)is designed.Based on the traditional moment calculation method,the proposed calculation method improves the accuracy of moment calculation by improving the number of sampling points in the kernel function calculation.The accuracy improvement of moment calculation is verified by comparing the performance of image reconstruction.Using the proposed accurate moment calculation method,the algorithm calculates the moment magnitudes of GPCET and applies the QIM on the moment magnitudes to embed the watermark.A difference image constructed by the moments before and after embedding is added to the original image to generate a watermarked image.GPCET has the advantages of geometric invariance,numerical stability and so on,which are adapted to watermarking,but as continuous moment,the integral error and geometrical error generated in the moment calculation will weaken these advantages.The proposed moment calculation method reduces these errors and improves the performance of watermarking.Subsequent experiments test the proposed watermarking algorithm according to its invisibility and robustness,and verify that the robustness of the proposed scheme outperforms that of other schemes when its level of invisibility is significantly higher.(3)To the blind selection of parameters in the watermarking algorithms based on the fractional-order moments,a watermarking algorithm based on generic radial harmonic Fourier moments(GRHFM)is proposed.First,by analyzing the relationship between the radial kernel function characteristics and the watermarking performance,the dissertation gives the rationality that GRHFM is selected as the embedding domain.Then,based on the similarity between the discrete Fourier transform(DFT)and the GRHFM moment calculation method in the polar coordinate system,a fast Fourier transform(FFT)-based moment calculation method is obtained.Ant colony optimization(ACO)is introduced to select the controlling parameters of GRHFM adaptively.Finally,a binary watermark is embedded in the GRHFM magnitudes based on the QIM strategy.The low time cost of the FFT-based calculation method is combined with the adaptive parameter selection method based on ACO to select the optimal parameter that maximize the invisibility and robustness.The experimental results indicate that the algorithm achieves higher robustness with the same payload and invisibility.(4)Moment-based zero-watermarking is robust against both image processing and geometric attacks,but the discrimination of watermarks is often ignored by researchers,resulting in the high possibility that host images and fake host images cannot be distinguished by verifier.To this end,this dissertation proposes a zero-watermarking algorithm based on the stability of the relationships between moment magnitudes of the same order and the stability of the relationships between moment magnitudes of the same repetition.The algorithm first groups multiple images into one image group and calculates the moment magnitudes for each image,Then the magnitudes of the same order and the magnitudes of the same repetition are compared to obtain the content-related features.All the image features are added together to obtain the features for the image group.Finally,the scheme extracts a robust feature vector with chaos system and takes the bitwise XOR of the robust feature and a scrambled watermark to generate a zero-watermark.The scheme produces robust features with both resistance to various attacks and low similarity among different images.In addition,the one-to-many mapping between magnitudes and robust feature bits reduces the number of moments involved,which not only reduces the computational time but also further improves the robustness.The experimental results show that the proposed scheme outperforms the state-of-the-art methods in terms of robustness,discrimination,and computational time under the same payloads.