摘要
提出了一种新的抗几何攻击的鲁棒性数字水印嵌入算法。采用Harris算子提取最健壮的特征点来估计图像可能遭受的几何攻击的参数,水印的嵌入则在满足不可察觉性的条件下遍布整个空间域。为获得最健壮特征点(MRFP),文中采用了一种自适应的特征点提取算法,通过对数字图像的预攻击处理,自适应地调整Harris探测器的阈值,来获得特定数目的MRFP。仿真结果表明,本算法对旋转、缩放、平移等几何攻击都具有很好的鲁棒性,对一般的图像处理如JPEG压缩、低通滤波、噪声等攻击也具有较好的抵抗性能。
A novel feature-based digital image watermarking method which is robust against geometric attack is proposed in this paper. The watermark is embedded imperceptibly in the entire spatial domain by a direct mapping. To generate resistance to scaling and rotation attacks, a set of most robust feature point（MRFP） is extracted by an adaptive Harris comer detector. The experimental evaluation demonstrates that the proposed method is able to withstand a variety of attacks including JPEG compression, Noise addition and common geometric attacks such as rotation, scaling and shift.
出处
《信息技术》
2008年第1期55-58,62,共5页
Information Technology
关键词
数字水印
几何攻击
特征点
digital watermarking
geometric attack
feature point
