摘要
数字水印的一个重要研究内容是攻击方法和对策,目前考虑得较多的是JPEG等各种压缩编码、几何形变、滤波和附加噪声。结合分形IFS编码理论提出了一种新的水印攻击算法,将含有特定水印信息的图像块在压缩映射和Hausdorff测度下随机选择相似块替代,并通过评估和削弱攻击过程中水印的影响,达到去除水印的目的。在不同类型的图像上选择鲁棒的DCT域和小波域水印算法进行了实验,结果表明攻击能有效降低水印的相关度检测输出,结合其他信号处理操作还可以获得更好的攻击效果。这将为水印鲁棒性研究提供新的评价指标。
An important issue in watermarking refers to attack and its counteraction. So far, most existed attack algorithms are all kinds of compressions (e.g. the JPEG), geometric transformation, filtering, and noise imposing. A new algorithm of attack digital watermarking based on fractal IFS coding is presented. Image blocks with particular watermark information are substituted by other randomly selected Hausdorff-similar blocks under compression map, and removed by evaluating and impairing the watermark’s affection during the attacks. All experiments are done upon different types of images under DCT domain and wavelet domain and results show that these watermark detection score are degraded effectively after the attacks, and better results can be gotten when the attacks are combined with other signal processing operations.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2005年第4期965-967,981,共4页
Journal of System Simulation
基金
湖南省自然科学基金(02JJY3049)
关键词
数字水印
水印攻击
水印鲁棒性
分形IFS编码
digital watermark
watermark attack
watermark robustness
fractal IFS coding
