基于广义猫映射与神经网络的图像空域水印算法  

Image spatial domain watermark algorithm based on general Arnold mapping and neural network

作　　者：黄大足[1] 陈志刚[1] 朱从旭[1] 

机构地区：[1]中南大学信息科学与工程学院

出　　处：《计算机应用研究》2008年第4期1144-1146,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60573127);湖南省自然科学基金资助项目(05JJ40131)

摘　　要：提出了基于广义Arnold变换和神经网络模型的图像空域水印算法。二值水印嵌入位置通过广义猫映射随机产生,训练一个神经网络记忆每个图像块中像素之间的关系,并采用水印嵌入增强处理技术,加强了水印的嵌入强度。实验结果表明,该算法具有较强的鲁棒性,能有效地抵抗剪切攻击、LSB攻击和椒盐噪声攻击等。This paper proposed an image robust spatial domain watermark algorithm based on general Arnold mapping and neural network model. Embedding position of the original binary watermark was generated by Arnold map in random, and then trained a neural network to memorize the relationship among pixels of each sub-block image. Moreover, used magnific information technology, and all these enhanced watermark embedded. Experimental results demonstrate that this algorithm has robustness against cutting attack, LSB attack, pepper＆sault, and so on.

关 键 词：水印 神经网络 ARNOLD映射 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]