有限脊波变换在数字水印中的应用  

Finite Ridgelet Transform in Digital Watermarking

作　　者：祝瑞玲[1] 王欣[1] 

机构地区：[1]山东大学信息科学与工程学院,济南250062

出　　处：《计算机辅助设计与图形学学报》2009年第2期257-261,共5页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60872119)

摘　　要：已有的有限脊波域的水印算法通过调制位于最大能量方向的有限脊波变换系数实现,但是哪些系数更适合嵌入水印是一个有待解决的问题.采用Roberts算子提取边缘图像,然后用拉普拉斯分布模型分析边缘图像的有限脊波变换系数,依据最大后验概率准则估计出适合嵌入水印的变换系数,自适应地嵌入水印信息.实验结果表明,该算法具有很强的鲁棒性和不可见性.Existing watermarking algorithms in finite ridgelet transform domain usually work by modulating the transform coefficients in the direction of maximum energy. However, how to select suitable coefficients for watermark embedding is still an open question. In this work, a study on finite ridgelet transform in digital watermarking is carried out. At first, edges are detected by Roberts operator, followed by finite ridgelet transform on the obtained edge image. Then based on Laplacian probability distribution function （PDF）, the suitable finite ridgelet transform coefficients are obtained by adopting maximum a posterior （MAP） estimation principle. Finally the watermark is embedded adaptively in those coefficients. Experimental results demonstrate the good robustness and invisibility of our proposed approach.

关 键 词：数字水印 有限脊波变换 最大后验概率 拉普拉斯分布 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]