节点文献

数字音频水印技术的研究和实现

Study and Implementation on the Digital Audio Watermarking Technique

【作者】 李鑫

【导师】 马晓红;

【作者基本信息】 大连理工大学 , 信号与信息处理, 2008, 硕士

【摘要】 随着互联网和计算机技术的普及和飞速发展,数字多媒体产品得到了广泛应用。如何在网络环境下实施数字产品的版权保护、合法权益和信息安全,成为数字通信中亟待解决的问题。数字水印技术的发展提供了解决这一问题的新思路。随着以MP3为代表的音乐产品在网络上广泛传播,数字音频水印技术逐渐成为国内外学术界研究的热点。本论文重点研究数字音频水印技术,并在此基础之上,给出水印图像恢复的新思路。给出了两种基于独立分量分析(Independent Component Analysis,ICA)的数字音频水印方法和一种基于码本映射的分块离散Hopfield网络的水印图像恢复方法。首先,阐述了课题提出的意义和研究背景。其次,介绍了本论文中用到的基本理论知识。接着,给出了两种基于ICA的小波域数字音频水印方法。第一种方法针对数字水印算法对抗时间轴同步攻击能力较弱,以及版权保护和音频信号感兴趣区域(Region of Interest,ROI)内容保护的问题,给出了一种基于ICA的自同步小波域多目的音频水印方法,实现了对音频信号重要音频段和版权标识保护的双重目的。第二种方法针对多著作权版权保护、音频重要音频段内容保护和篡改定位的问题,给出了一种基于ICA的小波域多重和多目的音频水印方法,该方法把音频信号的ROI和两个图像标识嵌入到宿主信号中,实现了水印技术的多重性和多目的性这两个重要特性,并且只需嵌入鲁棒水印而不需要嵌入脆弱水印即可实现篡改定位的功能。最后,针对水印领域中经常遇到的版权标识遭受恶意攻击而无法有力保护版权的问题,给出了一种基于码本映射的分块Hopfield网络的水印图像恢复方法,该方法通过码本映射的方式解决了Hopfield网络权矩阵占用空间大的问题;采用对图像进行分块和降低相关性的方法,提高了网络对模糊图像的恢复性能。实验结果表明,给出的两种水印方法均满足音频水印技术对透明性、鲁棒性和安全性的要求,可以用于数字音频产品的版权保护、内容保护或篡改区域定位;给出的针对模糊水印图像的恢复方法,改善了传统水印系统的性能,进一步达到了版权保护的目的。

【Abstract】 As the prevalence and of rapid development computers and Internet, digital multimedia product has been applied widely. How to carry out copyright protection, copyright protection and information security on the Internet has become one of the urgent problems to be solved in digital communication. The development of digital watermark technique becomes a novel way to solve this problem. As music products like MP3 are widely spread on the Internet, digital audio watermarking technique becomes an internationally hot topic at present.In this paper, the main task focuses on digital audio watermarking scheme. And then, watermarking image restoration scheme is also provided, which is based on the audio watermarking scheme. Two kinds of digital audio watermarking schemes based on Independent Component Analysis (ICA) and one kind of watermarking image restoration scheme based on codebook mapping and blocked discrete Hopfield network are provided in this paper. Firstly, the meaning of the task and the research background are introduced. Secondly, basic background knowledge related is introduced. After that, two kinds of digital audio watermarking schemes based on ICA in wavelet transformed domain are proposed. Aiming at the problems that most existing digital audio watermarking schemes can weakly resist time-scale attack, and copyright protection and Region of Interest (ROI) of audio signal protection, a multipurpose digital audio watermarking scheme in the wavelet transformed domain based on ICA and synchronization is proposed, which realizes two purposes: ROI important audio segment protection and single copyright protection. Aiming at multiple copyrights protection, important segment audio content protection and tampered areas localization, a multiple and multipurpose audio watermarking scheme in the wavelet transformed domain based on ICA is proposed, which embeds one ROI audio signal and two image signals into the host signal. Owning two important characters: multiple and multipurpose, this scheme achieves the purpose of tampering localization only embedding robust watermarking signals, but not fragile watermarking signals. At last, aiming at the effective copyright protection even when the copyright symbols have undergone serious attacks, a watermarking image restoration scheme based on codebook mapping and blocked discrete Hopfield network is proposed, which solves the large space assuming problem of weight matrix of Hopfield network by codebook mapping, and enhances the network restoration performance by blocking scheme and reducing correlation of watermarking images.Experimental results show that both of the two audio watermarking schemes can satisfy the requirements of transparency, robustness and security, and can be applied in copyright protection, content protection or tampered areas localization; the blurred watermarking image restoration scheme can improve the performance of the traditional watermarking system and achieve the purpose of the copyright protection more effectively.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络