【作者】 陈海涛；

【导师】 王阿川；

【作者基本信息】 东北林业大学 ， 计算机应用技术， 2010， 硕士

【摘要】 本文简要介绍了感知图像哈希的研究背景和国内外的发展现状,深入的研究了感知图像哈希的生成方法,并对特征提取过程进行了系统的阐述。对小波变换和离散余弦变换两种特征提取方法进行了比较和研究,同时结合多媒体信息的特点选取离散余弦变换作为本文特征提取方法。利用非负矩阵分解将图像矩阵进行分解,保留其系数矩阵作为后期处理的数据,这样不但使得计算的效率提高,还能有效的降低存储空间。利用离散余弦变换提取特征,用人类视觉模型进行系数处理,同时利用混沌序列进行加密,实验证明该方法具有很好的鲁棒性和检索性。采用离散余弦变换进行特征提取的哈希算法不能有效的抵抗几何攻击,为了提高该算法的鲁棒性能,本文将Zernike矩的概念引入,利用Zernike矩的旋转不变性对图像矩阵进行矩的提取,生成哈希序列将其加入到感知哈希序列中,作为抗几何攻击的特征检验序列。本文提取图像的前十阶Zernike矩生成哈希序列,并通过Matlab进行实验验证。实验结果表明该方法能够有效的抵抗图像旋转攻击,从而提高了整个算法的鲁棒性。本文对混沌理论的图像加密技术进行了研究。由于图像保护技术不仅要保护图像的鲁棒性还要有很强的安全性,所以对图像进行加密就尤为重要了。本文结合多媒体信息加密的特点又充分利用了混沌理论的特性,提出采用混沌理论对图像哈希进行各个阶段的加密方法。这种加密的方法不仅能满足哈希序列的安全性要求还具有很好的灵活性和抗攻击性能。感知哈希技术作为多媒体处理的新兴技术,必将在多媒体管理中发挥越来越重要的作用。本文从感知哈希技术的核心出发,将感知哈希技术同混沌理论、非负矩阵分解和人类视觉模型有效的结合,对已有的哈希提取算法进行了改进和提高,对图像保护技术也有很大的促进作用。更多还原

【Abstract】 This thesis briefly introduces the study backgrounds and research status of Perception Image Hash at home and abroad and makes further studies on the methods with illustrating on the extraction process of characteristics in system. Then, we compare and study the characteristics extraction methods of Wavelet Transform and Discrete Cosine Transform, and select Discrete Cosine Transform as the characteristics extraction method combined with the characteristics of multimedia information.We use the Non-negative Matrix Factorization to decompose the image matrixes and keep the Coefficient Matrix as data that needs post treatment, which will not only improve the computational efficiency but also reduce the storage space effectively. The experiment shows that the method has good robustness and retrievement through using Discrete Cosine Transform to extract characteristics and using human visual model to process coefficient and encrypting with Chaotic Sequence.Generally, the Hash Algorithm using Discrete Cosine Transform to extract characteristics could not resist geometric attack. In order to improve the robustness of this calculation method, we introduce the concepts of Zernike moments and use its rotation invariance to extract the moment from Image Matrix to generate hash sequence which is added into Perceptual Hash sequence as Characteristics of test sequences to against geometric attacks. This article extracts the first ten steps of image as we called Zernike moments to generate hash sequence and use Matlab to make experimental verification. The results show that the method has improved the robustness effectively and could resist the image rotation attack.In this paper, the image encryption techniques of chaos theory are researched. The image protection technology not only has to protect Robustness of image but also has strong safety, so it is especially important to encrypt image. Combined with the characteristics of multimedia information encryption and made full use of the characteristics of chaos theory, we put forward a method that uses Chaos Theory to encrypt each stage of hash image which can not only satisfy the safety requirements of Hash Algorithm but also has good flexibility and anti-attack capabilityPerceptual hashing technology as an emerging technology will play an increasingly important role in multimedia management. The core of Perceptual Hashing technology has been used as a basis and it combines efficiently Perceptual Hashing technology with the Chaos Theory and Non-negative Matrix Factorization and human vision model. It improves existed hash Extraction Algorithm and has a great role in promoting image protection technology.更多还原