| 一种基于奇异值分解和反传神经网络的多小波域数字水印算法 |
| |
| 引用本文： | 李国良,谭月辉,玄克诚,李志勇,彭涛.一种基于奇异值分解和反传神经网络的多小波域数字水印算法[J].河北机电学院学报,2010(4):219-222. |
| |
| 作者姓名： | 李国良 谭月辉 玄克诚 李志勇 彭涛 |
| |
| 作者单位： | [1]军械工程学院计算机工程系,河北石家庄050003 [2]65301部队装备部,吉林长春130022 [3]军械工程学院训练部,河北石家庄050003 |
| |
| 摘 要： | 针对图像经过CARDBAL多小波变换后,各层系数的能量平均分配,且其同一尺度、同一方向上的4个子块系数相近这一特性,结合奇异值分解和BP神经网络将二值图案水印嵌入到多小波系数的高频分量中。实验表明,该算法有良好的不可感知性和强健的鲁棒性。
|
| 关 键 词： | 数字水印 多小波 奇异值分解 BP神经网络 |
Multiwavelet-based image watermarking using SVD and BP neural network |
| |
| Authors: | LI Guo-liang TAN Yue-hui XUAN Ke-cheng LI Zhi-yong PENG Tao |
| |
| Affiliation: | 1.Department of Computer Engineering,Ordnance Engineering College,Shijiazhuang Hebei 050003,China;2.No.65301 Unit of PLA,Changchun Jilin 130022,China;3.Training Department,Ordnance Engineering College,Shijiazhuang Hebei 050003,China) |
| |
| Abstract: | A blind watermarking algorithm based on CARDBAL multiwavelet transformation,SVD and BP neural network is proposed.In this scheme,the original image is firstly decomposed by multiwavelet transformation,then the relationship among singular values of subblocks in high frequency areas is learned by BP neural network.Finally a watermark signal is embedded into the multiwavelet domain partly by adjusting the relationship among singular values of these subblocks.The experimental results prove the robustness and imperceptibility of this method. |
| |
| Keywords: | digital watermarking multiwavelet SVD BP neural network |
| 本文献已被 维普 等数据库收录！ |
|
