摘要
对神经网络应用于数字水印技术进行了改进,选择了一种DWT变换嵌入算法的BP四层模型作为神经网络的训练模型,它能更好地实现了水印的盲提取。实验结果表明,该算法对常用的图像处理如压缩、低通滤波、加噪、剪切、旋转和缩放等攻击具有较好的鲁棒性和不可见性。
The authors improved the application method of four-layer neural network to digital watermarking technology and chose BP for neural network of DWT transformation as the training mode,which could better realize blind withdraw of watermark.The results show that this algorithm provides high robustness and invisibleness against image processing attacks such as compression,low-pass filtering,adding noise,cropping,rotating and zooming.
出处
《教育技术导刊》
2008年第2期106-108,共3页
Introduction of Educational Technology
关键词
神经网络
盲水印
鲁棒性
不可见性
neural network
blind watermark
robust
invisibleness
