【作者】 高庆；

【导师】 刘泉；

【作者基本信息】 武汉理工大学 ， 通信与信息系统， 2008， 硕士

【摘要】 随着互联网以及多媒体通讯的发展,对多媒体产品版权保护和鉴别要求越来越强烈,而数字水印技术是解决数字版权保护问题的有效手段。目前,对静态图像数字水印技术的研究己经日趋成熟,音频数字水印也正成为新的研究热点。本文在详细分析数字音频水印技术特点和需求的基础上,以直观的二值图像作为水印信息,以音频数据为嵌入对象,从提高数字水印的不可察觉性和鲁棒性出发,探讨了两种新的用于版权保护的数字音频盲水印方法。一是基于离散小波变换的盲音频水印算法。在对作为水印的二值图像进行Arnold置乱处理,将其转化为一维序列后,再对原始音频信号实施分段离散小波变换并根据离散小波变换近似分量的能量自适应确定嵌入强度,选择小波变换近似分量的偶数点系数嵌入水印,水印嵌入的次数等于原始音频分成的段数。水印检测不需要原始音频信号,提高了水印的工程应用价值。二是引入了基于倒谱变换和离散余弦变换相结合的音频水印算法。算法中对二值水印图像进行了置乱处理,增加了水印信息的安全性。将原始音频信号分段并对每段音频数据实施倒谱变换,再对倒谱系数进行离散余弦变换,采用量化原理将水印信息嵌入到离散余弦变换的低频系数中,保证了水印的不可感知性以及鲁棒性。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。与目前大部分文献采用的倒谱变换方法相比,本算法有更好的不可感知性及鲁棒性。仿真实验表明,本课题提出的两种算法具有较好的不可感知性,对一些基本的信号处理操作具有很强的鲁棒性,并且均实现了水印的盲检测,具有很好的实用价值。更多还原

【Abstract】 With the development of the internet and multimedia, the protection for the intellectual copyright becomes more and more urgent and important, meanwhile, the digital watermarking is the effective means to solve the intellectual copyright problems. Recently, research on the still image watermarking technology becomes more and more matured; while the research on audio digital watermarking is becoming a hot topic.With the two-dimension binary image as watermarking information, the audio data as embedded target, on the basis of analyzing the digital technological characteristic and demand of audio watermarking in detail, in order to enhance the imperceptible and robust of the digital watermarking, this paper discusses two new digital blind watermarking methods of audio for copyright protection.At one hand, there is the digital blind watermarking algorithm of audio based on discrete wavelet transform. First, the watermarking image has been processed by Arnold disorder transform, then converted it into a one-dimension sequence; Second, the audio data has been divided and discrete wavelet transform; Third, determine the embedding strength of watermarking adaptively according to the energy of the DWT approximate coefficient, and choose the even number of wavelet coefficient to embed watermarking. The times of the watermarking has been embedded repeatedly equal to the segments number of the audio has been divided. The watermark can be extracted without the original audio signal. This helps it to use in the actual project.On the other hand, it discusses the digital watermarking algorithm of audio based on cepstrum transform and discrete cosine transform. The algorithm increases the watermarking security by disordering the watermarking. The audio signal has been divided and cepstrum transform, and use the coefficient of cepstrum transform to discrete cosine transform. Embedding the watermarking on the low-frequency of the discrete cosine transform using quantization theory improves the watermarking’s imperceptible and robust. .The watermarking can be extracted without the original audio. Compared with the methods based on cepstrum transform proposed in the most papers at present, the algorithm of this paper has better imperceptible and robust.The experimental result demonstrates that the watermarking is imperceptible and the algorithms are robust to many basic signal operations. The watermark can be extracted without the original audio. The blind watermarking detection has very good practical value.更多还原