基于主成分分析的零水印算法

Zero-watermark Algorithm based on Principle Components Analysis

针对传统的数字水印方法,提出了一种基于主成分分量的零水印算法。该算法是对图像进行分块和主成分分析(PCA),得到图像的主要分量,然后将基于Logistic映射生成的混沌序列对图像的主要分量进行位置置乱,比较置乱后相邻主分量系数间的大小生成特征水印。当对待认证图像进行认证时,用同样的方法提取该认证图像的信息。实验结果表明,该算法具有无失真、鲁棒性好、简单易行及可操作性强等优点,并对一些常见的攻击,如JPEG压缩、剪裁、加噪、滤波等有很强的鲁棒性。

This paper presents a component based on principal component zero-watermarking algorithm. First, the image was subdivided block by block and principle components analysis （PCA） was used to decor relate the image pixel to obtain the principle components of an image. The chaotic sequence was generated based on Logistic mapping and the principle components were thrown into confusion. Then the coefficients of the confused principle components were compared to generate the characteristie watermark. The same process was done to draw out the characteristic watermark from the authentication image. Experimental results show that this method is no distortion, robustness, good, simple, and operability, etc, As well as some of the common attack such as JPEG compression, cropping, adding noise, filtering and so on. Compared with the image zero-watermarking method in discrete wavelet transfor domain, this algorithm was proved to be more robust.