论文标题:信息安全中的数字水印技术研究
Studies of Digital Watermarking of Information Security
论文作者 张志明
论文导师 王磊,论文学位 博士,论文专业 检测技术与自动化装置
论文单位 西北工业大学,点击次数 85,论文页数 102页File Size6226k
2003-06-01论文网 http://www.lw23.com/lunwen_711760172/ 数字水印;信息隐藏;混沌;人类视觉系统;奇异值分解变换;版权保护;人工神经网络;差错控制编码;稳健性增强
Digital Watermarking;Information Hiding;Chaos;Human Visual System;Copyright Protection;Artificial Neural Networks;Error Control Coding;Robustness Enhancement;Singular Value Decomposition
基于计算机和网络的信息交换为多媒体数字作品的使用、传播提供了便利的途径，然而由于数字作品极易被非法复制和篡改的特性，使得数字作品的信息安全保护和版权保护成为迫切需要解决的实际问题。数字水印(Digital Watermarking)技术利用数字作品中普遍存在的冗余数据与随机性，向数字作品中加入不易察觉但可以判定区分的秘密信息-水印(Watermark)，水印可以是一段文字、标识、ID序列号等，从而保护数字作品版权或完整性。数字水印技术已经成为数字信息安全问题，尤其是数字作品版权保护的一种有效手段，得到广泛的研究和应用。 本文主要探讨网络信息安全中的数字图像水印技术，介绍了数字水印的研究背景、基本原理、应用领域以及当前国内外的发展状况，文中结合混沌技术、SVD变换、神经网络和差错控制编码等技术提出了多种不同类型水印算法方案，它们分别是：混沌技术在数字水印技术中的应用；SVD变换域中的数字水印方案；基于神经网络和纠错编码应用的稳健性增强方案。这几种数字水印(原理)方案涵盖了图像数字水印的主要方面，能够满足各种应用领域的实际要求。 本文利用HVS人类视觉系统模型，提出了图像内容自适应的数字水印算法，分别实现了时空域中的统计原理水印算法和DCT变换域中的有意义水印算法。混沌序列在这两类算法中分别作为随机水印信号发生器和二值图像水印的混沌加密手段，具有安全性高、形式简单的优点；利用HVS特性，分别在时空域和DCT频率变换域中计算图像内容的自适应嵌入强度，增强了算法的稳健性；水印的嵌入和检测计算复杂度小，算法简单。 本文基于图像矩阵的SVD奇异值分解特性分析，提出了在新的SVD变换域中进行的数字水印算法，水印信息嵌入到分块变换的最大奇异值分量系数中，应用混沌变换加密和空域置换，改善了空域性能，安全性高；利用图像分块矩阵的奇异值分解稳定性好的特点，采用图像内容自适应方法计算水印的嵌入强度，增强了算法的稳健性；采用二值图像作为有意义水印进行嵌入和检测，水印在感知上是可视的。根据实际应用目的的不同，提出了两类水印方案，一种为稳健性较强的非盲水印方案，另外一种为采用系数量化技术实现的盲水印嵌入／提取方案。最后，通过实验验证了本文万尹了尹头掌博士学位论文所设计的水印方案的隐形性和稳健性。 本文针对现有的数字水印系统中最重要的性能一稳健性和隐形性(载体对象的降质)这两个互相矛盾的性能指标的分析，提出了两种不同原理的解决方案，嵌入大容量的水印(二值标志图像)信息数据。本文在保证水印不可见性的基础上和载体图像的允许降质范围内，利用RBF神经网络技术实现数字水印信息的最大强度嵌入以增强数字水印抵御各种恶意攻击的稳健性;在通信论和信息论模型基础上，引入ECC(差错控制编码)技术降低水印的误码率，使得提取水印信息更加可靠，使可能的水印攻击变得更加困难，从而获得更好的水印稳健性和可靠性。实验结果说明，本文所提出的原理性方案可以推广应用到不同类型和变换域的数字水印系统中。 最后，本文对全文内容和研究成果作了一个概要的回顾，介绍了论文的创新点，并对未来值得关注的研究和发展方向作了简单讨论和预测。
With the rapid development of computer, network and communication, especially the popularization of the Internet, information security and copyright protection have been a key and attractive problem. A potential solution to alleviate this problem is to use invisible digital watermarking scheme, which refers to the embedding of secret messages, unobtrusive marks or labels that can be represented as bits into the digital multimedia.In this paper, we mainly describe the digital watermarking technologies of information security. We firstly introduced the research backgrounds, basic principles, application areas and recent development of the digital watermarking technology. Then we adopted some new technologies such as chaos, SVD (Singular Value Decomposition), ANNs (Artificial Neural Networks), ECC (Error Correction Code), into the digital watermarking and proposed several different kinds of watermarking schemes: watermarking with chaos control; watermarking in SVD domain; robustness enhancement with ANNs and ECC. The proposed schemes in this paper could cover the primary aspects of the image digital watermarking and fulfill the practical application"s demands.This paper employed the characteristics of the HVS (Human Visual System) responded to the image contents and realized the two kinds of image-adaptive watermark algorithm, one is in the spatial domain based on the statistics theory and another in the DCT domain with meaningful digital watermark. The chaos sequence was applied in these two algorithms to increase its security and was served separately as the random watermark generator and the watermark image encryption techniques. Due to the analysis of the image characteristics, we determined the watermark strength by the classification of the image content and obtained better robustness and imperceptibility.New algorithms of meaningful digital watermarking in SVD domain with chaotic transform encryption are proposed. Based on the SVs characteristic analysis of image matrix, a visually recognizable binary image watermark is embedded into maximal singular value coefficient in block-based SVD transform domain of the cover image. Here we brought forward two primary schemes: one need original signal and the other is blind (without the originalcover). Experimental results show that our schemes can extract reliable copy of the hidden watermark from images that have been significantly degraded or altered through several common geometric distortions and signal processing operations.The most two important requirements of digital watermarking are the imperceptibility and robustness but it is hard to obtain a good balance between them. In this paper, we proposed two algorithms, which embed a visually recognizable binary image watermark into the host image, to enhance the robustness within the imperceptibility limit and degraded quality limit. The first one used an RBF neural network to mimic the HVS to achieve the adaptively different embedding maximum-strength according to the frequency component feature of the cover image. Experimental results show that the proposed techniques have good imperceptibility and can survive of common image processing operations and JPEG lossy compression with high robustness. The second one considered the watermarking system as a special digital communication system and introduced the ECC as channel coding technique. By using the Turbo coding technique, we are able to reduce the watermark bit error rate and improve the robustness of the watermark so as to make counterfeiting attacks more difficult.Finally, we make the summary of our works and introduce the achieved innovations in briefly. We also discussed the future works should be done of the digital watermarking.
|
