降低图像统计特性差异的隐写分析

The basic concept of the current image steganalysis is to analyze the embedding mechanism and the statistical changes of the image data caused by secret message embedding. However, the embedding changes are not only correlated with the steganography methods, but also with the image content and the local statistical characteristics. Compared with the process of embedding, the image contents make a more significant impact on the differences of image statistical characteristics. This makes the image steganalysis to be a classification problem with bigger within-class scatter distances and smaller between-class scatter distances. As a result, the steganalysis features will be inseparate caused by the differences of image statistical characteristics..This project focuses on steganalysis based on reducing the differences of image statistical characteristics caused by various content and processing methods. In this project, an image will be assumed to be a local stationary Markov source, and the theoretical and technical methods of image engineering and image steganalysis will be studied and used. The main contents of the project include: analyze the correlation between the complexity of the content and the statistical features, and build the statistical characteristics based steganalysis model; study the model of image quality degradation caused by embedding secret messages, and propose the calibration methods based on content and image restoring; according to the proposed steganalysis model, study the steganalysis algorithms based on reducing the differences of image statistical characteristics. .This project can effectively decrease the influence of the image content and make the steganalysis features of the cover or stego images distribute more centralized within the same class of images. The project tends to improve the steganalysis performances especially when the train and test images are with different statistical features and for adaptive embedding. As a result, the project will have obvious theoretical meaning and practical value for studying and applying image steganalysis.

隐写分析研究表明，隐写对图像的影响不仅与秘密信息的嵌入机制有关，并且与目标图像的内容特征、统计分布关系密切相关。通常，图像内容对统计特性差异性的影响强于秘密信息的嵌入过程，这使得图像隐写分析成为了一个“类间聚合、类内分散”的分类问题。.本项目从自然图像的统计模型入手，以降低由内容、处理手段等造成的图像统计特性差异为着力点，提出可靠的隐写检测方法。主要研究内容包括：依据图像视觉内容属性呈现的数据分布关系，分析图像统计特性与隐藏信息存在性特征间的关系，建立降低图像统计特性差异的隐写分析技术模型；分析数字隐写所构成的图像质量退化模型，构建基于图像内容和载体图像恢复的图像校准方法；结合基于统计特性的隐写分析模型和图像校准方法，提出降低图像统计特性差异的隐写分析算法。.本项目能够有效降低隐藏信息存在性特征类内分散现象，提高隐写分析算法的性能，对于图像隐写分析的发展和应用有明显的理论意义和实用价值。

图像内容所造成的统计特性差异会对隐写分析算法的性能产生影响，本项目针对此问题进行研究，主要成果如下：研究降低图像统计特性差异的隐写分析模型，提出了基于图像分类、基于图像分割以及分类和分割相结合的隐写分析模型，通过对具有代表性的隐写分析特征绘制特征分布曲线、计算可分性判据等，验证了模型的有效性；提出了四种降低图像统计特性差异的隐写分析算法，与传统隐写分析方法相比，检测性能有明显提升；对复杂网络环境下载体、载密图像统计特性“失配”的问题进行了研究，提出了两种相应的隐写分析方法，与实验室环境下基于分类器设计的隐写分析算法相比，性能有所提升，提高了算法的实用性；研究了基于深度学习的图像隐写分析算法和负载定位算法，得出了图像内容复杂度与负载定位的准确度相关的结论，为本项目的后续研究打下了基础。