摘要
该文利用Hopfield网络的联想记忆功能,提出一种基于Hopfield神经网络的文本可恢复水印算法。该算法通过同义词替换嵌入水印,用可替换同义词的位置信息和水印信息完成Hopfield神经网络对原始文本特征信息和水印信息的记忆训练。仿真实验表明,该算法具有良好的隐蔽性和鲁棒性,不仅能够恢复由水印的嵌入造成的原始文本的改变,而且能够恢复由对文本内容的部分攻击造成的水印文本的改变。
Making using of associative memory function of Hopfield neural network, a text recoverable watermarking algorithm based on Hopfield neural network is proposed in this paper. Watermark is embedded by synonym replacement, and the Hopfield neural network is trained by the position information of replaceable words and watermark to remember the original text features and watermark. The experiment results show that the proposed algorithm has high concealment and robustness, not only the change of original text caused by embedding watermark can be recovered, but also some attacks on content watermark text can be recovered.
出处
《杭州电子科技大学学报（自然科学版）》
2013年第5期138-141,共4页
Journal of Hangzhou Dianzi University：Natural Sciences
关键词
神经网络
联想记忆
同义词
可恢复
neural network
associative memory
synonym
recoverability
