Research On Digital Watermarking Technology For Copyright Protection Of Digital Image Resources

With the rapid development of multimedia technology,digital resources have become more and more popular in contemporary teaching,but these resources are easily destroyed by means of forgery,illegal tampering,etc.As an active authentication method,digital image watermarking technology has become an important means for copyright protection of image resources.The research problems of digital image watermarking algorithm mainly focus on the following two aspects: the problem of resisting geometric attacks and the problem of good balance between invisibility and robustness.In this paper,we propose three effective digital image watermarking algorithms to deal with these problems,which can be summarized as follows:1.According to the Least Squares Support Vector Regression(LS-SVR)theory,a robust image watermarking algorithm based on BKF statistical parameter correction is proposed.Firstly,the Nonsubsampled Shearlet Transform(NSST)is performed on the resident image,and the watermark is embedded into the low-frequency coefficients by quantization.Then,the BKF statistical parameters of the two-scale high-frequency sub-bands are calculated and the geometric correction model is trained by LS-SVR.Finally,the geometric correction is performed on the watermarked image with LS-SVR model,and then the watermark information is extracted from the NSST low frequency sub-band of corrected watermarked image.The algorithm makes use of BKF statistical parameters to improve the characterization ability of the image feature,and makes the watermark algorithm have a strong ability to resist geometric attacks through LS-SVR correction model.Experimental results show that the algorithm is not only invisible,but also resistant to geometric and conventional attacks.2.Based on the theory of BKF vector Hidden Markov Tree(BKF-HMT),we propose a digital image watermarking detection method based on multi-correlation joint statistical modeling.Firstly,the watermark is embedded into the significant NSST coefficients by using a novel multiplicative method.Next,the BKF-HMT is utilized to model the NSST coefficients,in which the EM algorithm and multiple correlations,including adjacent scales,adjacent directions of the same scale and spatial neighborhood of the same scale and direction,are fully employed.Finally,based on the BKF-HMT model and the Maximum Likelihood test theory,the ML detector is constructed to extract the watermark,which can capture the important features of two-dimensional images by using the NSST coefficients with the advantages of multi-resolution,translation invariance and anisotropy.By making full use of the correlation of NSST coefficients and optimizing the BKF-HMT model,the detectionaccuracy of digital watermarking is improved.The comparison results show that the proposed method can achieve a good balance between imperceptibility and robustness.3.According to the correlation between color components and Cauchy-HMT theory,we propose a high performance color image watermarking method based on enhanced vector HMT model.The color channel of color image is transformed by Quaternion Wavelet Transform(QWT),and the watermark is embedded into the important sub-band amplitude in QWT domain.Then,the Cauchy-HMT is used to model the amplitude.Based on the Cauchy-HMT model and the Local Most Powerful(LMP)test theory,the Locally Optimum Detector(LOD)is constructed for watermark extraction.The model is optimized by the correlation between color components.The proposed LOD can significantly improve the accuracy of digital watermarking detection.Experimental results show that the proposed method has better performance in maintaining a good balance between imperceptibility and robustness.