基于物理与视觉感知模型的信息隐藏与保护

发布时间：2021-08-10 18:21

　　信息保护在当今已经成为一个重要的研究方向。有史以来人们对于私密信息保护的探索从未停止。到了现代,我们的社会已经被各种各样的信息充斥、填满,其中就包括很多不能公开的数据。这些私密的数据在传播的过程中,就面临着被盗用或者刻意改写的风险。传统的信息攻击大都基于互联网,这是因为在网络中,所有的数据都是以数字信号的形式存储的。传统保护信息的方法主要有密码学和隐写术。在本文中,我们分别从两个方面,结合了当前技术的进展,研究了基于物理与视觉感知模型的信息保护问题。首先,我们对物理密码破解的过程进行了建模,以此来帮助用户更好地保护信息。在研究中,我们假设攻击者在用户输入完密码并且离开之后才进行密码的窃取。他们使用红外相机对用户输入密码的痕迹进行偷拍和分析,进而分析出密码。我们探索了基于红外相机的密码攻击对于信息安全的威胁性。我们首先对红外序列进行了建模,然后利用最大似然函数法进行模型的求解。在这个过程中,最大的挑战在于模型的建立,其次是如何优雅并且有效率地求解该模型。我们在研究中通过对红外序列的深入分析解决了上述两个问题。最后,我们提出了一系列可以帮助用户保护密码的建议。其次,我们研究了针对单人用户的... 

【文章来源】：上海交通大学上海市 211工程院校 985工程院校 教育部直属院校

【文章页数】：150 页

【学位级别】：博士

【文章目录】：
Acknowledgements
ABSTRACT
摘要
Nomenclature
Chapter 1 Introduction
    1.1 Background
        1.1.1 Password Protection
        1.1.2 Visual Information Protection
    1.2 Related Work
        1.2.1 Cryptography and Image Steganography
        1.2.2 Physical Password Breaking
        1.2.3 Visual Masking
        1.2.4 Information Hiding by TPVM
    1.3 Challenges
        1.3.1 Modeling and Solving Thermal Attack
        1.3.2 Information Hiding for One Single User
        1.3.3 Information Hiding for Multiple Users
        1.3.4 Misleading Image for Information Hiding
    1.4 Contributions
    1.5 Organization
Chapter 2 Physical Password Breaking via Thermal Sequence Analysis
    2.1 Introduction
    2.2 Problem Description
    2.3 Methodology
        2.3.1 Workflow of the System
        2.3.2 Physical Model for Temperature Decreasing on the Keys
        2.3.3 Estimate Key Touching Time with the Physical Model
        2.3.4 Password Estimation
    2.4 Experiment and Discussion
        2.4.1 Benchmark
        2.4.2 Evaluation Metrics
        2.4.3 Experiment
        2.4.4 Discussion
    2.5 Conclusion
Chapter 3 Information Hiding for One Single User
    3.1 Introduction
    3.2 Multi-channel visual masking
        3.2.1 Visual Masking by Linear Combination
        3.2.2 Visual Masking with Masking Effect Map (MEM)
    3.3 Experiments and Discussion
        3.3.1 Linear Combination v.s. MEM based Approach
        3.3.2 Comparison with the Benchmark
        3.3.3 Effectiveness of Misleading Images
        3.3.4 Discussion of Visual Masks
        3.3.5 Influence of Attack Types
    3.4 Conclusion
Chapter 4 Information Hiding for Multiple Users
    4.1 Introduction
    4.2 Method
        4.2.1 Network with the Disturbing Module
        4.2.2 Loss Function
        4.2.3 Patch GAN
    4.3 Experiments
        4.3.1 Benchmark
        4.3.2 Restoration Accuracy
        4.3.3 Converging Speed
        4.3.4 Generalization
        4.3.5 Ablation Study
    4.4 Conclusion
Chapter 5 Misleading Information Selection for Information Protection
    5.1 Introduction
    5.2 Related Work
        5.2.1 Place Recognition
        5.2.2 Domain Adaptation
    5.3 Method
        5.3.1 Retrieving Images
        5.3.2 Information Hiding with Misleading Image Selection
    5.4 Experiments
        5.4.1 Datasets and Evaluation Metric
        5.4.2 Results and Discussion
    5.5 Conclusion
Chapter 6 Summary and Future Work
    6.1 Summary
    6.2 Future Work
Bibliography
Publications
Patents



本文编号：3334553