【作者】 刘培培；

【导师】 朱中梁；

【作者基本信息】 西南交通大学 ， 通信与信息系统， 2008， 博士

【摘要】 信息时代的到来使我们的社会和日常生活发生了深刻的革命性变化。计算机网络技术、通信技术和多媒体技术的快速发展和广泛应用,使网络信息的安全与保护问题日益成为亟待解决的关键问题。在这样的时代背景下,信息隐藏技术作为版权保护和隐蔽通信等的主要手段,近年来成为国际信息安全领域一个新的研究热点。针对保障网络信息传递中图像内容的真实性、完整性和保密性的需求,结合科研项目的研究需要,本论文致力于研究以图像为载体的脆弱数字水印技术和数字隐写技术。首先分析了现有的用于图像精确认证的脆弱水印算法,从提高算法的安全性角度出发,利用混沌映射高度的随机性和对初值的极端敏感性,提出了一种安全的混沌脆弱水印算法。该算法实现了基于图像内容生成的水印信息与图像内容的随机交错分布,加强了算法的安全性。仿真结果表明算法能够精确检测和定位对图像内容的恶意篡改,同时水印图像具有极好的不可见性,并且提取水印不需要原始图像。从保证秘密信息的不可见性角度出发,研究应用于隐蔽通信的图像隐写算法,指出秘密信息的不可见性是保证算法安全性的前提条件。结合人眼视觉模型,构造了一种基于混沌映射的DCT域自适应图像隐写算法,利用人眼视觉模型导出的恰可见差异阈值调节嵌入秘密信息的强度和位置,实现了不可见性和鲁棒性之间的最佳折衷,仿真结果验证了算法的有效性。从提高图像隐写算法的安全性角度出发,针对SPA隐写分析算法对传统LSB隐写算法攻击的有效性,理论分析了F₁替换和F_-1替换对图像统计量的影响作用,推导了可以抵抗SPA分析的F₁替换和F_-1替换的理想嵌入比例关系。在此基础上提出了一种抵抗SPA分析的图像隐写策略,并实现了基于误差校正的图像隐写算法。仿真实验结果表明,该算法能够有效抵抗SPA隐写分析,安全性高。同时,具有对载体图像修改少,嵌入秘密信息的载体图像视觉质量好,提取秘密信息简单等特点。在分析现有的迭代混合图像隐藏算法性能的基础上,构造了一种新的基于小波对比度的迭代混合图像隐藏算法。该算法将各图像块的小波对比度绝对值之和作为视觉敏感性的测度,选取适合嵌入秘密信息的图像块,最后以迭代混合的方式将混沌置乱后的秘密图像嵌入这些图像块。仿真结果表明该算法在实现大容量信息隐藏的同时,载密图像仍具有很好的视觉质量。最后,从提高图像隐藏算法的隐藏容量角度出发,将迭代混合的思想扩展应用到整数小波变换域,实现了一种新的自适应迭代混合图像隐藏算法。该算法采用整数小波变换,克服了传统小波变换域图像隐藏算法存在的边界误差和舍入误差问题,同时充分考虑人眼视觉特性,有效协调了不可见性和鲁棒性之间的矛盾。实验结果表明算法隐藏容量大,同时具有好的不可见性和鲁棒性。更多还原

【Abstract】 The arrival of information age caused a revolution in our society and our daily life. Rapid development and extensive use of computer network technique, communication technique and multimedia technique, especially the rocketing development of Internet, have made the security and protection of network information be a problem, which needs to be urgently resolved. In this situation, information hiding technique used as the main method for copyright protection and covert communication has been focused on the international information security field.Clear focus on protecting the fidelity, integrity and security of digital content, this thesis devotes to researching fragile watermarking technique and digital steganographic technique with image as the carrier.Firstly, the current fragile watermarking algorithm used for precise authentication of image is analysed. In order to enhance the security of this algorithm, the properties of super randomicity and high sensitivity on initial value of chaotic map are utilized, and a secure chaotic fragile watermarking algorithm is proposed. Randomly interlaced distribution of original image content and watermark content generated from original image enhances the security of the proposed algorithm. Experimental results show that the proposed algorithm can precisely detect and localize the malicious tamper of image content; simultaneously, the watermarked image has some merits such as good imperceptibility; the extraction of watermark without the original image.Secondly, from the point of view of protecting the imperceptibility of secret information, image steganographic algorithm used for covert communication is researched, and the conclusion is obtained that the imperceptibility of secret information is the premise condition for protecting the security of the algorithm. Based on HVS （Human Vision System） model, a chaos-based self-adaptive image steganographic algorithm in DCT （Discrete Cosine Transform） domain is proposed. The JND （Just Noticeable Difference） obtained from HVS model is used to adjust the embedding strength and position, which realizes the perfect compromise of imperceptibility and robustness. Experimental results validate the validity of the proposed algorithm.Thirdly, the influence of image statistic caused by F₁ substitution and F_-1 substitution is theoretically analysed, and the ideal embedding ratio relation of F₁ substitution and F_-1 substitution used to against SPA steganalysis is obtained. Hence, the idea of error correction is utilized and a novel image steganographic algorithm against SPA steganalysis is proposed. Expermental results show that the proposed secure algorithm can resist SPA steganalysis effectively. Simultaneously, the less modification and imperceptible stego-image can be obtained.Fourthly, on the basis of the analysis of the current iterative blending image steganographic algorithm, a novel image steganographic algorithm based on wavelet contrast is proposed. The sum of absolute values of wavelet contrasts of each image block is adopted as the measurement of visual sensitivity. Under the direction of this measurement, suitable image blocks are selected and the scrambled gray scale image is embedded into these suitable image blocks by iterative blending method. Simulation results show that the proposed algorithm has good invisibility and large capacity for information hiding.Finally, in order to increase the capacity of image steganographic algorithm, the idea of iterative blending is applied to integer wavelet domain and a novel self-adaptive image steganographic algorithm is presented. The employment of integer wavelet overcomes the boundary difference and rounding difference that exist in traditional image steganographic algorithms. Simultaneously, the HVS model is sufficiently used to adjust the imperceptibility and robustness. Experimental results show that the proposed algorithm has a large capacity, good imperceptibility and strong robustness.更多还原