一种基于神经网络的盲水印方法  

A Image Watermarking Scheme Based on Neural Networks

作　　者：姚志强[1] 林崧[1] 林晖[1] 陈荔聪[1] 

机构地区：[1]福建师范大学计算机科学系,福州350007

出　　处：《计算机工程与应用》2004年第26期52-55,共4页Computer Engineering and Applications

基　　金：国家自然科学基金资助项目(编号:60275023);福建省自然科学基金和教育厅资助项目(编号:A0210016;JA03028)

摘　　要：该文提出一种在图像小波变换的低频系数中嵌入水印的方法,水印的提取是基于神经网络方法,而且不依靠原始输入图像。实验结果表明,该方法所实现的水印图像视觉效果好,且对常见的图像处理是稳健的。A digital watermarki ng scheme based on wavelet transform for images is proposed in this paper.The sc heme is to hide the watermarking signals in the wavelet domain.The extract met hod is realized by the multilayer perceptron neural networks and the back-propa gation algorithm.The neural networks almost exactly recover the watermarking si gnals from the watermarked images after training and learning.The experimental results show the effectivity and robustness of the digital watermarks.

关 键 词：小波变换 神经网络 水印 稳健性 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]