摘要
针对现有遥感图像水印算法抵抗旋转、剪切、水平镜像及联合几何攻击能力差的问题,提出一种局部化遥感图像数字水印新方法。算法首先对水印图像进行Arnold置乱预处理,并将其行扫描形成一维向量作为待嵌入的水印信息;在宿主遥感图像中提取Harris-Laplace特征点,依据特征点确定水印嵌入特征区域;规范所选特征区域,对其进行归一化后实施Contourlet变换,利用奇偶量化方法将数字水印嵌入选定的变换系数内。实验证明算法对滤波、噪声、JEPG压缩等常规信号处理具备较好鲁棒性,并且能够更好地抵抗旋转、缩放、平移及其联合等几何攻击。
Against the problem of existing watermarking algorithms for remote sensing image＇s poor resistance to shear, rotate, horizontal mirror, and combined geometric attacks, a new localized watermarking algorithm is proposed for remote sensing image, the watermark image is scrambled by Arnold. Then the scrambled image is scanned into one-dimensional vector as watermark information. Harris-Laplace feature points are extracted in the host remote sensing image. And feature regions are determined for watermark embedding according to the points. Finally,the selected feature region is specified and normalized. And with contourlet transformation, watermark is embeded into the selected transform coefficients. Experiments show that the algorithm has good robustness against conventional signal processing such as filtering, noise, JPEG compression etc. , with better resistibility to geometric attacks such as rotation, scaling, moving and their combination.
出处
《世界科技研究与发展》
CSCD
2012年第2期256-260,共5页
World Sci-Tech R&D
基金
国家自然科学基金(41101452)
辽宁省教育厅基金(L2010158)
辽宁省百千万人才工程项目(201 1921063)资助
