【作者】 于琛；

【导师】 李庆忠；

【作者基本信息】 中国海洋大学 ， 信号与信息处理， 2006， 硕士

【摘要】 随着Internet技术和多媒体信息技术的飞速发展,数字多媒体信息在网上传播与传输越来越便捷,但这同时也带来了信息安全的隐患问题。首先,信息安全技术直接关系到一个国家的安全问题,如果一些机要部门的信息被窃取,就会给整个社会带来严重的危害。此外,信息安全技术还关系到一个企业的商业机密和企业的生存问题以及个人通信的隐私问题。图像信息隐藏技术是多媒体通信和多媒体处理领域近年来出现的一个新的研究方向,尚处于其发展的初级阶段,从理论方法到实际应用仍然有许多问题亟待解决。图像信息隐藏技术的关键是在保证隐蔽图像视觉不可见的质量要求下,研究具有较强稳健性的信息嵌入策略与算法,以抵抗信息在网上传输时有意和无意的不良攻击,并提高载体图像的信息嵌入量。本文根据以上研究目标,针对目前图像信息隐藏方法中存在的问题,对频域内稳健的图像隐藏技术进行了较为深入的研究,主要完成的研究工作如下:(1) DCT域内稳健图像信息隐藏方法的研究。首先分析研究了国内外在DCT域内的各种信息隐藏方法的特点,为进一步提高目前隐藏方法的稳健性、安全性,和提高信息的隐藏量,主要作了以下三方面的改进工作:为了克服秘密图像提取时存在严重失真问题,本文提出了一种基于像素值合理分配与标志性嵌入的新的嵌入策略与方法;为了提高DCT域内载体图像信息的隐藏量,本文根据人类视觉系统(HVS)的特点,引入模糊理论,提出了DCT子块的模糊分类方法,实现了秘密信息嵌入强度的自适应确定;为提高秘密隐藏图像的网上传输的安全性,本文研究了基于混沌理论的图像加密方法,实现了基于混沌技术的秘密图像位置和像素值的双重加密算法。最后通过实验验证了以上方法的有效性。(2)彩色图像由于数据量大,在网上传输时隐藏比较困难,为此本文将DCT域内灰度图像的隐藏方法延伸至彩色图像的隐藏,实现了基于YCbCr色彩空间下的彩色秘密图像的隐藏算法。(3)整数小波变换是一种无失真可逆变换,为进一步提高频域内秘密信息的隐藏量,并与目前发展流行的JPEG-2000压缩标准相兼容,本文研究了基于提升小波变换的图像信息隐藏方法,提出了基于HVS的小波系数分类方法,利用Daub 5/3小波实现了小波域内秘密信息的隐藏。本研究将促进图像信息隐藏技术实用化、商品化的进程,解决多媒体信息传输与传播中目前亟待解决的信息安全问题,具有重要的学术价值和社会、经济意义。更多还原

【Abstract】 With the drastic expand of Internet and multimedia technology; it is more and more convenient to transmit and spread multimedia information over Internet. Unfortunately, it also brings upon the information security problem. Information security technology is directly concerned with information security of a nation. If the secret information of some confidential departments is wiretapped, grievous harms would be done to the entire society. Moreover, it is also of utmost importance to preserve the commercial secrets of enterprises and individual privacy of correspondence. In recent years, image hiding is becoming a new research direction in multimedia communication and processing, so there are numerous unsolved problems, which hamper the actual implement of the technology. The key of image hiding technology is to investigate the robust data embedding method, which should maintain good perceptual quality for stego-image and have large embedding capacity, and be robust against intentional and unintentional attacks. Focusing on the above goal we studied the robust image hiding approaches in frequency domain. The following research has been accomplished.(1) The characteristics of various image-hiding methods in DCT domain were first analyzed. In order to improve the robustness, security, and embedding capacity of the existing approaches, we have carried out the following research. To overcome the grievous distortion problem due to distortions introduced in the secret image by rounding off errors, we proposed a new embedding scheme consisting of a reasonable partition of pixel value and a sign embedding operation. By incorporating fuzzy theory with human visual system (HVS), a fuzzy classifier for DCT sub-blocks was developed to realize the adaptive selection of the embedding strength. Moreover, the secret image for embedding was encrypted by exploiting the shuffling operations for pixel values and pixel positions based on chaotic cryptography. Finally, extensive experiments have been conducted to evaluate the performance of the proposed methods.(2) Color image is characterized of large amount of data, which brings difficulties in transmission and hiding by Internet. By extending the grey image hiding method in DCT domain, a robust color image hiding method based on YCbCr color system was developed and implemented.(3) Integral wavelet transform is a completely reversible transform with high fidelity. To further improve the embedding capacity of secret information in frequency domain and make compatible with the popular international image compression standard of JPEG-2000, we presented an image hiding method based on the Daub 5/3 lifting form of wavelet transform. This method allows inserting data adaptively in DWT domain.This research will shorten the distance between the image hiding theory and its practical application. It is of significance for secure information communication over Internet.更多还原