鲁棒性图像水印算法研究.docVIP

武汉理工大学 Wuhan University of Technology 鲁棒性图像水印算法研究 学 院: 信息工程学院 专 业: 通信与信息系统 摘 要 随着多媒体技术与网络技术的飞速发展，人们获取信息也变得越来越方便快捷。然而科技的发展在给人们带来方便的同时，也出现了许多迫切需要解决的问题，如在网络上传输数据文件或作品时，恶意的个人或团体,有可能在没有得到作品所有者的许可下，十分容易地拷贝和传播有版权的内容。如何有效地防范对信息的非法复制和传播,成为版权保护中的一个重要问题。数字水印作为媒体知识产权保护的一种有效方法，近年来被广泛关注。本文针对数字图像水印，主要做了如下的研究工作：通过对SVM特性以及数字图像第二代水印算法的研究分析，提出了一种基于支持向量机的鲁棒盲水印算法。该算法首先用多尺度Harris-Laplace检测算子从载体图像中提取出稳定的特征点；然后根据特征自适应确定局部特征区域，在特征区域选择一些点作为嵌入水印的点，结合图像的邻域相关性，根据灰度图像特点，选取特征向量作为SVR训练模型，进而利用SVR进行预测，调节嵌入点的象素值进行水印的嵌入和提取。实验表明，用该技术嵌入水印后的图像具有很好的图像感知质量，对常规信号处理和去同步攻击特别是JPEG压缩具有较强的鲁棒性。 关键词：数字水印；Harris-Laplace角点检测；回归型支持向量机 Abstract With the rapid development of multimedia and network, we can get information more and more convenient. But when the development of technology brings convenience to us, it also brings in some difficulties to solve. For example, when we send data file or writings through network, illegal individual or community may copy or spread the copyright able works easily without permission of the author. How to availably prevent illegal copying and spreading is an important problem in copyright protection. Recently Digital watermarking is not iced broadly because of its behavior in the protection of multimedia copyright. This paper mainly researches into the image watermarking, as following. In the research of the behavior of SVM and the second generation watermarking, a new robust blind watermarking scheme based on SVM is proposed. Firstly, the Harris-Laplace detector is utilized to extract feature points, and then the local characteristic regions are adaptively constructed according to the characteristic scale theory, select some points from the regions as embedded points. The watermarking embedding and extraction achieve by using a preoperational technique based on support vector regression model to adjust the values of the selected pixels. The model is trained by learning the relationship between the selected pixel and statistical feathers of nei