【作者】 汪爱莲；

【导师】 李淑芝；

【作者基本信息】 江西理工大学 ， 计算机应用技术， 2007， 硕士

【摘要】 本文以直观二值图像作为水印信息,在分析数字音频水印技术特点和需求的基础上,从提高数字水印的不可察觉性和鲁棒性出发,结合倒谱变换和小波变换的特性,对已有算法进行改进,提出两种用于版权保护的频域数字音频盲水印算法。其一,针对基于倒谱域统计均值的数字音频水印算法在透明性和健壮性方面的不足进行了改进。在水印嵌入前的预处理中,使用混沌序列加密和重复编码(一种简单的纠错编码)的方法来提高水印信息的安全性和数字水印的鲁棒性。本文中采用5倍重复码对水印信息进行纠错。其二,为提高数字音频盲水印算法的鲁棒性,在引入重复码的基于倒谱域统计均值的数字音频盲水印算法的基础上,讨论采用离散小波变换和复倒谱变换相结合的二次变换来取代只使用复倒谱变换方法的可行性。小波变换具有多分辨率特性,复倒谱变换具有解相关的能力和倒谱系数均值大多趋于零等优点,本文把两者的优点结合起来,用离散小波变换得到的低频分量做复倒谱变换得到其倒谱系数用于嵌入数字水印信息的方法来提高数字水印算法的性能。本文对上述两种算法均进行大量不可感知性和鲁棒性实验。实验结果表明,两种音频盲水印算法都具有良好的不可察觉性和鲁棒性,重复编码的引入在总体上增强了算法对噪声污染、低通滤波等常见信号攻击的抵抗能力。更多还原

【Abstract】 This paper takes visual binary image as watermark information, it is on the basis of the analysis of both the digital technical characteristics and the demand of audio frequency watermark, combines the characteristic of cepstrum transform and the wavelet transform to enhance the imperceptibility and robustness of the digital watermarking, and sets an improvement of the exist algorithm. It researches two kinds of frequency domain digital audio blind watermarking robust algorithm for copyright protection.Firstly, there is an improvement in the embedding process of the existing algorithm—the sigital audio blind watermark algorithm which is based on the statistical mean of cepstrum domain. In the pre-process of embedding, both chaos encryption sequence and repletion code (a kind of simple error-correction code) are used to improve the security of watermark information and robustness of digital watermarking. This paper uses 5 times repetition codes for the correction of the watermark.Secondly, in order to improve the robustness of the digital audio bland watermark algorithm, this paper discusses the feasibility of substitute the complex cepstrum transform of twain transform, which are combined by discreet wavelet transform and complex cepstrum transform, on the basis of introducing the repetition code of digital audio bland watermark algorithm based on cepstrum transform. The discreet wavelet transform has the characteristic of multi-resolution and the cepstrum transform is excellent with the ability of de-correlation, and most of their mean of coefficient are around zero. The paper combines the advantages of them, and uses the low frequency weight of discreet wavelet transform to make complex cepstrum transform, and then in return the coefficient of the cepstrum transform is used to embed information in the digital watermarking, thus raise the performance of digital watermarking algorithm. There are a lot of imperceptible and robust experiments for both the two kinds of better algorithm. The experimental results show that both of the tow audio watermarking algorithms have well not awareness and robustness. In general, the introduction of repetition coding strengthened algorithm’s resistance against common signal attack, such as noise pollution、low-pass filtering and so on.更多还原