【作者】 尹显东；

【导师】 李在铭； 姚军；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2007， 博士

【摘要】 多维随机序列敏感视觉信息隐藏技术是近年才出现的一个多学科渗透的前沿技术。该技术利用人眼的视觉缺陷来掩盖信息的视觉“存在性”,其理论和应用都在快速发展之中。因此系统地分析和研究该领域内的一些问题,对创立完整的多维视觉信息隐藏系统理论体系有着非常重要的意义。本论文主要探讨与多维视觉信息隐藏有关的一些理论和技术问题,独立完成了如下一些研究内容:(1)详细研究了多维随机视觉信息序列中敏感图像隐藏通信系统的一般理论、三元模型及其系统方程。针对四种典型应用,建立相应的敏感图像信息隐藏通信系统框架和功能方程。(2)对多维视觉敏感图像隐藏系统所涉及的各相关序列进行了状态空间分析,定义了图像信息自隐藏空间、图像视觉形象完全不同的隐藏空间和图像视觉形象不发生明显变化的隐藏空间。在此基础上设计出两种在多维涉密视觉信息序列中,自动分割出敏感图像区域,并将敏感图像重新隐藏到去密图像序列中的隐藏方案。(3)分析图像视觉形象与图像位平面之间的关系,设计出在宿主图像位平面中隐藏敏感图像的方法。该隐藏方法在嵌入相当大的数据量时,仍然较好地保持了宿主图像的质量和统计特性。(4)分析图像视觉形象与图像信号DWT变换域系数之间的关系,设计出基于小波变换的图像信息隐藏算法,该隐藏方法具有较强的鲁棒性、抗信道干扰和抗压缩的性能。(5)研究敏感图像信息隐藏前的混沌加密和数据置乱加密技术,提出一种敏感图像混沌加密和DWT域系数置乱隐藏的安全大容量的敏感图像信息隐藏算法,并通过实验证明了该方法的优越性。(6)对涉密图像序列中敏感视觉信息的识别与分割技术进行研究。深入系统地分析了基于视觉特征的图像场景理解过程,提出场景理解的五层次模型,进一步应用该模型建立相应场景图像分析理解系统,并在此基础上设计出基于色彩、纹理、轮廓、几何形状等视觉特征的敏感子图像检测系统模型,并从理论和实验两方面对检测定位算法的效率和效果进行了分析。(7)针对特殊应用环境,利用上述的相关理论和技术,设计并实现基于DM642-FPGA的涉密场景中银灰球区域图像实时隐藏通信系统,来验证多维随机序列中敏感视觉信息自动隐藏技术的可实现性。最后分析了当前在信息隐藏和隐密通信研究中存在的问题,探讨了该领域需要进一步研究的几个方面。更多还原

【Abstract】 The hiding technology of sensitive visual information in multidimensional stochastic series is a new research field involved in multi-subject, which is developed rapidly these years. It conceals visual "existence" of information by making use of visual characteristics of human eyes. Now relative technology and applications are changed quickly, so it’s very important to analyze and study some crucial problems in this field systematically for establishing integrated multidimensional visual information hiding system theoryIn this dissertation, some issues involved in sensitive visual information hiding theory and techniques are studied, and the work finished independently is introduced in detail as follows:(1) General theory and mathematical model of sensitive image hiding communication system in multidimensional visual information series are investigated particularly. For 4 typical applications the scheme and function equation of corresponding hiding systems are established.(2) The concerned series state space for multidimensional visual sensitive image hiding is analyzed, and three types of state space are defined including image information auto-hiding space, image visual dissimilarity hiding space and image visual unapparent changing space. Thus two sensitive image-hiding methods are designed, which finish extracting sensitive sub-image region and hiding this sensitive information to the remains image series.(3) Relation between image visual character and image bit plane is studied, and a hiding method of sensitive information based on host image bit plane is designed. Its advantage is remaining the good quality and statistics of raw image in hiding course while large computations are embedded.(4) Relation between image visual character and coefficients of DWT transformation is investigated, and image information hiding algorithm is proposed based on discrete wavelet transform. It has good performance on robustness, anti-jamming and anti-compressing.(5) The technology of chaos encryption and series scrambling encryption before sensitive image information hiding is studied, and a hiding algorithm for chaos encryption and DWT coefficients hiding of sensitive image is suggested. And the algorithm doesn’t need the raw image to extract sensitive information. The sightless property and secrecy of the algorithm are analyzed, and the results of simulation experiments show its good performance.(6) The technology of automatic recognition and segmentation of sensitive visual information in secret image series is studied. The general course of scene recognition based on visual characteristic is analyzed systematically, and a five-layer model of scene recognition is proposed. According to the model corresponding scene recognizing analysis system is built; and thus a detection system of sensitive sub-image is designed using junior visual characteristics as color, texture, sketch and shape etc. The efficiency and effect of detection position algorithm are analyzed in detail according to theoretic analysis and experiment.(7) For special application, a real-time scene communication system of recognizing and segmenting automatic silver gray ball image based on DM642-FPGA is designed. It can identify feasibility of automatic hiding sensitive information in multidimensional visual information series.Finally, the thesis is concluded by pointing out some open issues on the research of information hiding and steganogaphy as well as highlighting some further research directions.更多还原