摘要
提出了一种基于广义Arnold变换与支持向量机模型的图像数字水印算法.选定的二值水印图像经过置乱以后,水印的嵌入位置通过广义Arnold变换随机产生,利用部分选定位置图像块中像素之间的关系来训练支持向量机,并采用水印嵌入增强处理技术,加强了水印的嵌入强度.在训练好的支持向量机辅助下,水印的检测过程可以看作是嵌入算法的一个逆过程.实验结果表明,该算法具有较好的鲁棒性,能有效抵抗Jpeg压缩、噪声等多种攻击.
An image watermarking scheme based on generalized Arnold transform and support vector ma- chine was proposed in this paper. Firstly, the watermark embedding positions and support vector machine training data selection were generated according to the given generalized Arnold transform, then the support vector machine was trained and the randomly permuted binary watermark was embedded into the host. The embedding strength was enhanced based on the watermark strengthen technique. Following the reverse step of the watermark embedding, the watermark could be extracted by the aids of the trained support vector ma- chine. The experimental results showed that the proposed scheme had good robustness against several attacks, such as Jpeg compression, noisy and shearing.
出处
《集美大学学报（自然科学版）》
CAS
2011年第1期65-70,共6页
Journal of Jimei University：Natural Science
基金
福建省自然科学基金项目(2007J0029
2009H0036)
厦门市科技项目(3502Z20103016
3502Z20075054)
集美大学优秀青年骨干教师基金项目(Z01108)
关键词
数字水印
鲁棒性
支持向量机
信息隐藏
广义Arnold变换
digital watermarking
robustness
support vector machine
information hiding
ized Arnold transform general
