一种基于奇异值分解的改进图像水印算法

数字水印作为一种新型的信息隐藏技术而成为目前业界研究的热点。本文提出了一种基于奇异值分解的图像水印改进算法。该改进算法具有如下特点：（1）在水印嵌入时，对原始图像以块为单位进行奇异值分解，在得到的奇异值中直接嵌入水印；（2）在水印提取时，不需要原始图像数据或原始水印数据。改进后的算法更加简便、快捷。实验
表明，与DCT算法相比较，本文所提出的方法在抵抗各种常规数据处理和攻击（如高斯噪声、JPEG压缩、滤波、裁剪等）方面具有良好的性能。

An Improved Image Watermarking Algorithm Based on Singular Value Decomposition

As a new type of information-concealing technology, digital watermarking has become a focus of research of the industry. This thesis puts forward an improved image watermarking algorithm based on singular value decomposition. It has the following characteristics, firstly, when watermarking data are being embedded, it exercises singular value decomposition by handling block as a unit,and embeds watermarks in the singular value directly; and secondly, when watermarking data are being extracted, no original image data or original watermarking data are needed, so the improved algorithm is much easier and faster. Experiments show that, compared with the DCT algorithm, the method put forward in this thesis works well in dealing with various kinds of regular data processing and attacks, such as the Gaussian noise, JPEG-eompression, filtering, and cropping.