【作者】 刘继新；

【导师】 陆哲明；

【作者基本信息】 哈尔滨工业大学 ， 信息与通信工程， 2010， 博士

【摘要】 信息的数字化已经给我们的社会和日常生活带来了巨大的和革命性的影响。数字多媒体技术的快速发展和国际互联网的广泛普及给我们的生活带来了方便,给我们的事业带来了机遇,但是与此同时也给我们带来了许多具有挑战性的问题。比如如何实现对数字多媒体作品的有效的版权保护以及如何实现安全的秘密通信等等。由于传统加密方法的种种缺点和应用局限,信息隐藏技术作为一种信息安全领域的新技术、新方法得到了越来越多的研究和应用群体的关注。信息隐藏技术主要有数字水印和数字密写两大分支,分别应用于数字多媒体作品的版权保护和隐蔽通信。矢量量化技术最早是作为一种更有效的数据压缩方法被提出来,随后广泛的应用于静态和动态图像编码、语音和音频编码以及语音识别等领域。而且近年来随着矢量量化研究的深入和各种研究方向的交叉,使得矢量量化的应用领域越来越广,如:信息隐藏、图像检索、数据库压缩和图像恢复等等各个领域都可以或多或少的找到矢量量化应用的例子。现有的研究多集中于以静态图像或视频为载体的基于矢量量化技术的信息隐藏,目前罕有将矢量量化技术应用到音频信息隐藏的报道。本文在这一方面做了有益的尝试,围绕着将矢量量化技术和信息隐藏技术相结合这一主题进行研究,做了如下四个方面的工作:(1)提出了一种基于矢量量化技术和修改的离散余弦变换的鲁棒音频数字水印算法,并采用小波变换来对音质进行改进,另外采用嵌入自同步码的方式实现了嵌入水印的同步。(2)提出了一种采用离散余弦变换的可以同时进行数字版权保护和防篡改认证的基于矢量量化技术的多功能音频数字水印算法,并通过实验验证了其有效性。(3)提出了一种基于矢量量化技术与低比特率语音编码器相结合的语音信息隐藏算法,并通过实际开发一个经由公共电话网络的隐蔽通信系统验证了其可行性。(4)提出了一种基于小波变换的音频指纹算法,根据每个小波分量系数的方差之间的关系设计了音频指纹算法,该算法在保持较高的最佳识别率的同时对常见的保留信号内容的攻击操作和加性高斯白噪声,尤其对线性速度变化攻击具有很好的鲁棒性。论文最后在总结本文研究成果的基础上,对基于矢量量化的音频和语音信息隐藏技术的研究前景做了展望。更多还原

【Abstract】 The digitization of information has produced enormous and revolutionary impaction both on our society and on our daily life. The rapid development of the digital multimedia technology and the widespread popularity of the Internet have brought about convenience to our life, opportunity to our career, but at the same time, many challenging questions to us. Such questions include how to achieve effective intellectual protection to the digital multimedia arts and how to accomplish secure secret communication, etc. Due to the various shortcomings and application constraints of conventional cryptography method, information hiding, as a new technology and method in information security domain, has drawn more and more attention from both research communities and application groups. Information hiding technology includes digital watermarking and steganography, applied to the copyright protection of the digital multimedia arts and the covert communication for the secret message.At first, the Vector Quantization (VQ) technology was coming up as a more valid data compression method. VQ has been applied to static and dynamic image coding, speech and audio coding, speech recognition and so on. Moreover, with the thorough study of VQ and the crossover of multiple disciplines, the application area of VQ has been becoming wider. Examples can be found in information hiding, image retrieval, database compression and image restore, etc. The existing research on VQ-based information hiding technique is fasten on using static image or video as the carrier signals, and there are rare reports on the combination of VQ and audio (or speech) information hiding. This dissertation makes a beneficial attempt in this direction, and the main works focusing on the combination of VQ and audio (or speech) information hiding technology are listed as follows:(1) A robust audio digital watermarking algorithm based on VQ and Modified Discrete Cosine Transform (DCT) is presented, and then the improvement, focusing on the audio quality, is made by using the Discrete Wavelet Transform (DWT). Furthermore, the synchronization of embedded watermark is fulfilled by embedding the self-synchronization code. (2) A multipurpose audio digital watermarking algorithm based on VQ and DCT for digital copyright protection and tampering authentication is presented. Experiments are carried to verify its effectiveness.(3) A speech information hiding algorithm based on VQ technology and low bit-rate speech codec is proposed. And a covert communication system based on this algorithm via Public Switching Telephone Network (PSTN) is implemented to justify the algorithm’s validation.(4) An audio fingerprinting algorithm based on Discret Wavelet Transform (DWT) is proposed. This algorithm designed an audio fingerprinting method according to the relationship of the variance of every wavelet components. The algorithm is robust to the common content preserving attacks and the addition of Gaussian White Noise, especially to the Linear Speed Change attack, while remains high identification rate.Finally, the research achievements on audio and speech information hiding technique based on vector quantization are summarized, and some new highlights in this research direction which need to be further explored are pointed out.更多还原