基于奇异值分解的鲁棒零水印算法

Robust zero watermarking algorithm based on singular value decomposition

针对当今数字水印技术中透明性不好、鲁棒性差、内容完整性差等问题,提出基于小波包分解和奇异值分解相结合的零水印算法,该算法利用小波包对高频信息进一步分解和奇异值分解的稳定性好,健壮性特点和数字水印隐秘性强的优势,和零水印相结合,解决鲁棒性差,内容不完整等的问题.首先,充分了解小波变换的原理,在此基础上对小波包分解和奇异值分解充分研究,然后把载体图像进行置乱、奇异值分解和提取特征矩阵;生成零水印最后,将零水印保存至版权保护中心,在后期认证中用于鉴定版权所有.小波包分解克服了小波变换不能分解低频的缺点,奇异值分解在图像分解过程中具有很好的稳定性从而提高了水印的鲁棒性.零水印不需要把水印信息嵌入载体中,进而保证了内容的完整性,仿真实验表明,零水印算法具有较好综合性能,与同类零水印算法相比,鲁棒性稳步提升,所容纳的数据信息更多,并实现了更方便的认证.

Aiming at the problems of poor transparency,poor robustness and poor content integrity in today’s digital watermarking technology,a zero watermarking algorithm based on wavelet packet decomposition and singular value decomposition is proposed.The algorithm uses wavelet packet to further decompose and singular high frequency information.The stability of value decomposition is good,the characteristics of robustness and the advantage of strong digital watermarking,combined with zero watermark,solve the problems of poor robustness and incomplete content.First,fully understand the principle of wavelet transform based on wavelet packet decomposition and singular value decomposition of the full study,then the vector image scrambling,singular value decomposition and feature extraction of zero watermark generation matrix;finally,the zero watermark is saved to the copyright protection center,for the identification of copyright in post certification.The wavelet packet decomposition overcomes the shortcoming that the wavelet transform can not decompose low frequency,and the singular value decomposition has a good stability in the process of image decomposition,so as to improve the robustness of the watermark.Zero watermark without the watermark is embedded in the carrier,so as to ensure the integrity of the content.The simulation results show that the zero watermarking algorithm in this paper has good comprehensive performance,compared with the same kind of zero watermarking algorithm,robustness steadily,data contain more information,and achieve a more convenient authentication.