收稿日期: 2010-03-29
网络出版日期: 2010-12-21
基金资助
上海市自然科学基金资助项目(09ZR1412400);上海市教委创新基金资助项目(10YZ11,11YZ10)
Secure Reversible Watermarking Based on Subsample Prediction
Received date: 2010-03-29
Online published: 2010-12-21
Supported by
上海市自然科学基金资助项目(09ZR1412400);上海市教委创新基金资助项目(10YZ11,11YZ10)
针对可逆水印安全性问题,提出一套高容量低失真的实现方案.通过选择图像的参考子集,利用密钥控制下的自适应插值算法,有效预测参考子集的补集,形成相关性较高的估计图像.估计图像与真实图像的差形成差值图像,此差值图像直方图峰值点可以达到较大高度,通过移位操作能隐藏高容量水印,并保持较低的视觉失真.由于预测算法本身由密钥控制,能满足水印嵌入的Kerckhoffs安全准则,提高安全性能.
关键词: 可逆水印;子采样;密钥控制插值;安全性
冯国瑞,戴宁街 . 基于采样预测的安全可逆水印算法[J]. 上海大学学报(自然科学版), 2010 , 16(6) : 603 -607 . DOI: 10.3969/j.issn.10072861.2010.
We propose a scheme of reversible watermarking with high security, low distortion and large capacity. After setting the reference sets, we use a keyinformationbased adaptive interpolation algorithm to estimate a complementary image. Because the predicted images are close to the desired ones, the difference image has a high peak in the histogram. Histogram shifting is performed to achieve a large payload while keeping low embedding distortion. As the operation is controlled by a key, the method conforms to the Kerckhoffs’ criterion and is therefore secure.
/
| 〈 |
|
〉 |