确定图像降质的鲁棒性数字水印嵌入算法

A robust digital watermark embedding algorithm according to assigned image degradation

结合视觉系统的掩蔽特性提出了一种数字图像离散Radon变换(DRAT)域内嵌入水印的新算法。从不可见性和鲁棒性两个方面分析了在DRAT域中可嵌入水印的原因。以水印图像的峰值信噪比作为评价水印图像质量的标准,采用相互独立的Gaussian随机序列作为水印信号,导出了水印长度、嵌入深度、水印图像峰值信噪比三者之间的理论关系,给出了在给定图像质量条件下水印嵌入的自适应参数估计。利用Checkmark1.2系统对新水印算法进行了攻击测试。测试结果表明,新算法对常见的图像处理如滤波、边缘增强、对比度变化、剪切等攻击具有良好的鲁棒性。

Using the visual masking an adaptive digital watermark embedding algorithm is presented in discrete Radon transform domain. The feasibility of the algorithm is analyzed in regard to both invisibility and robustness. The independent Gausslan stochastic vector as the watermark sequence is selected, and peak signal to noise ratio （PSNR） as the criterion of image degradation grade of watermarked image is selected. The theoretical relationship between the scaling parameter and length of watermark and the PSNR is deduced. Thus the scaling parameter and the length of watermark can be estimated according to the assigned image quality gradeo Moreover, the new watermarking scheme is tested by using Checkmark 1.2 system. The experimental results show that the scheme is enough robust to some very important image processing attacks, such as filtering, edge enhancement, contrast adjust, cropping etc.