【作者】 江涛；

【导师】 周祖德；

【作者基本信息】 武汉理工大学 ， 通信与信息系统， 2007， 硕士

【摘要】 随着计算机通信和信号处理技术的迅猛发展，数字图像、音频和视频等多媒体数字作品得到日益广泛的普及。但由于数字作品很容易被非法复制、纂改和传播，使得如何实施网络环境下的信息安全和版权保护成为数字通信中亟待解决的问题。数字水印技术为数字化产品的版权保护提供了一种有效的途径。随着数字音像作品的大量制作和发行，音频数据的版权保护也显得越来越重要。通过在音频载体中嵌入水印信息，可以实现拷贝限制、使用跟踪、盗用确认等功能。本文详细分析了音频数字水印技术的原理和音频水印系统的特点，从提高数字水印的不可感知性和鲁棒性出发，对现有音频水印算法进行研究分析，提出了改进的音频数字水印算法。本文主要研究工作和取得的主要成果如下：(1)在离散余弦变换域，提出了基于均值量化的音频数字水印算法。本算法中对二值水印进行了置乱处理和纠错编码，在一定程度上增加了水印信息的保密性和水印提取的正确率；将水印嵌入到DCT变换域的低中频系数上，保证了水印的不可感知性；使用均值量化原理嵌入水印，提高了算法的鲁棒性。(2)利用扩频原理，提出了一种改进的基于DCT域的音频数字水印算法。该算法首先将视觉上可辨的二值水印图像降维为一维序列，再将一维序列伪随机排序并与m序列作扩频调制，最后对数字音频信号进行分段离散余弦变换，通过均值量化的方法嵌入经过调制的水印信息。采用扩频原理和均值量化方法，能大大改善水印嵌入的不可感知性和鲁棒性，但由于采用扩频调制后，一个水印信息要用一个周期的伪随机序列调制，因此会增加嵌入的水印量。最后对算法的鲁棒性和不可感知性进行了仿真实验，实验结果表明所提出的算法对MP3压缩，低通滤波，重采样等攻击有很好的鲁棒性。(3)提出了一种新的基于奇异值分解和Arnold变换的音频数字水印算法。该算法首先原始音频信号进行分块SVD分解，选取一组奇异值通过均值量化的方法嵌入水印，并进行了仿真实验。本算法对作为水印的二值图象进行了Arnold置乱加密处理，可以增加水印的安全性。在水印检测过程中不需要原始音频信号，是一种盲水印算法。更多还原

【Abstract】 With the development and applications of computer communications and signal processing technologies, multimedia products, such as digital image, audio and video, have been popularized and applied increasingly and widely. But the digital products can easily be copied, edit and transmit, so how to take measures to enable copyright protection and information safety in the network, has became an important problem that shoule be solved in digital communication field immediately. Digital watermarking technique is the most effective method to solve copyright protection. With the products of digital audio products, the copyright protection of audio products becomes more and more important. Recently, audio digital watermarking technique has been recognized and researched, the frictions such as copy control, using tracing, can be realized by embedding the watermarking in the audio signal.The paper has been researched mainly on the algorithms of audio digital watermarking. It anlalyzes the theory of audio digital watermarking and the characters of audio digital watermarking system, from the point of improve the watermarking imperceptibility and robust, based on the analysis, comparison and test of existing digital audio watermarking algorithms, some digital audio watermarking algorithms are proposed and improved. The main research work and contributions of the dissertation are as follows:(1) In the discrete cosine transform domain, put forward an audio watermarking algorithm based on mean_quantization. The algorithm increase the watermarking security by disordering the watermarking. Applying the BCH coding technique to the watermark can increase the correctness of watermark extraction. Enbedding the watermark on the low-frequency and middle-frequency of DCT using mean_quantization theory improves the watermark’s imperceptility and robust. The watermark can be extracted without the original audio.(2) Applying Spectrm-spread communication theory, an improved DCT audio watermarking algorithm has been put forward. First, the audio signal has been divided and DCT, the watermark image will be embedded into the low-frequency of DCT based on mean_quantization theory. Experimental results show that the performance of this scheme is good infragility. The watermark can be extracted without the original audio.(3) A new blind digital audio watermarking algorithm is proposed for audio data security. First, the audio signal has been divided and SVD, the watermark image will be embedded into the low-frequency of SVD based on mean_quantization theory. The watermark image has been processed by Arnold disorder transform. Experimental results demonstrate that the method is robust and imperceptible. The watermark can be extracted without the original audio.更多还原