【作者】 郭志川；

【导师】 程义民；

【作者基本信息】 中国科学技术大学 ， 电路与系统， 2006， 博士

【摘要】 基于语音的信息隐秘传输一直是保密通信、军事通信等领域的一个研究热点，在战时利用公用网络进行机密数据的传输一直是信息战研究的前沿课题之一。隐秘传输采用信息隐藏技术，将机密信息嵌入到普通多媒体信息中不易被人眼、耳所感知且不易被侦测的部位，通过公共信道进行秘密传输，比传统的加密技术多加了一层防护。语音信息是传输中常用的多媒体信息，研究基于语音的信息隐秘传输及实现方法在信息安全、保密通信等领域具有重要的意义。本论文以方法研究为主，重点研究了基于语音的信息隐秘传输及其相关的若干关键技术： 提出了一种基于无线网中RPE-LTP语音的高嵌入效率的机密汉字隐秘传输方法，可以将机密汉字信息嵌入到RPE-LTP语音中不易被感知的部位，通过公共信道进行隐秘传输，该方法可以在RPE-LTP语音中每秒嵌入300 bits的机密汉字数据。 针对流媒体信息嵌入的大容量和较高码率的问题，提出了一种将一路2.4kb／sMELP低码率语音嵌入到另一路13kb／s RPE-LTP语音中，通过公共信道进行隐秘传输的方法，嵌入MELP语音后的RPE-LTP语音相对于原宿主语音，人耳较难感知。 目前大多嵌入／提取算法时间复杂度较高和不能在线实时处理，为满足实时、高速、较大容量信息的隐秘传输，嵌入／提取算法往往需要采用硬件实现。对嵌入／提取模块的硬件实现方法进行了研究，给出了基于数位信息嵌入模块的硬件设计和仿真。 此外，在基于图像的信息隐藏抗突发性误码和无损信息隐藏方面也进行了研究。将SP交织、纠错编码、基于数位信息的大容量嵌入方法相结合，给出了一种基于图像的抗突发误码和抗JPEG压缩的嵌入方法。在无损信息隐藏方面，给出了一种基于改进整数变换的无损信息隐藏方法，具有较低的时间复杂度。并应用于语音、遥感图像、医学图像和MPEG-2彩色视频中。更多还原

【Abstract】 Covert communication of secret information based on speech has been a hotspot in secure communication and military communication. Secret data transmitted in open networks in war has been one of the advanced subjects studied in information warfare. Using data hiding techniques, secret data was embedded in imperceptible multimedia data which did not raise any suspicion by our ears and eyes and was not easily to be detected. Secret data was secretly transmitted in open channels, and so covert communication adds a safeguard compared to traditional encryption techniques. Speech signal was widely used in communication, so research on data covert communication based on speech was very important in information security, secure communication etc. This dissertation focused on the development of methods of covert communication especially covert communication based on speech and its relative key techniques.A high data embedding efficient secret Chinese characters covert communication method based on RPE-LTP speech of wireless network was proposed. Secret Chinese characters were embedded in some imperceptible parts of the host data and secretly transmitted in open channels. Using the method, 300 bits/s of data were embedded in the host speech.In order to reach the demands of large capacity and high rates of streaming media data embedding, a new scheme capable of hiding a 2.4kb/s low-bit-rate mix-excitation linear prediction (MELP) speech into another RPE-LTP speech was proposed. The MELP speech can be secretly transmitted in open channels. It was difficult to detect the difference between the host RPE-LTP speech embedded a MELP speech and the original host RPE-LTP speech.Now, most data embedding and extracting algorithms have higher time complexity and could not process signal on line. In order to reach the demands of real-time, high speed and large capacity data covert communication, data embedding and extracting algorithms must be implemented by hardware generally. The hardware更多还原