摘要
几何攻击被公认为是数字图像水印技术走上商用的瓶颈之一。本文根据图像几何变换过程中灰度值统计特征的稳定性原理,提出了一种基于直方图调整的抗几何攻击数字图像水印算法。首先根据图像灰度均值选取一个灰度区间,然后通过调整位于该灰度区间内的图像直方图来嵌入水印信息。在提取水印时无需提供原始图像,实现盲检测。实验结果表明本算法不仅具有很好的隐蔽性,而且对诸如旋转、缩放、平移、仿射变换、剪切等几何攻击和常规信号处理均具有很强的鲁棒性。
Geometric attacks are regarded as one of the main bottlenecks for digital image watermarking techniques on the way to commercial application.According to the invariant property of the statistic characteristic of an image during geometric transformations,a digital image watermarking robust to geometric attacks based on histogram modification is proposed in this paper.First,a gray range is selected by reference to the mean of the cover image.And then the watermark is embedded by adjusting the local histogram,which is located in the selected gray range.The watermark can be retrieved without any knowledge about the original image.So the detection is blind.Experimental results demonstrate that the proposed algorithm not only keep a satisfactory imperceptibility,but also provide sufficient robustness against common signal processing operations and various geometric attack,such as rotation,scaling,translation,affine transformations and cropping,etc.
出处
《网络安全技术与应用》
2011年第4期46-49,共4页
Network Security Technology & Application
基金
合肥工业大学2010-2011学年大学生创新基金资助项目(XS2010066)资助
关键词
数字水印
几何攻击
直方图
统计特征
鲁棒性
Digital watermarking
geometric attacks
histogram
statistical characteristics
robustness