图像隐写取证中的软件识别与载体检测关键问题研究

The research of steganography forensics is of great theoretical value and practical significance for discovering, determining and tracking behaviors of covert communication based on information steganography. Existing researches on steganography forensics are mainly dependent on the existence detection of secret messages in carriers, which is means of a single. This project will explore and discuss the image steganography forensics from another point of view, and focus on studying the key issues of software recognition and image steganography in image steganography forensics. Based on the techniques and methods such as reverse engineering, software feature extraction and steganography behaviour analysis, the caculation feature of image steganography will be extracted, and the steganography behavior model of steganography softwares will be built. Then, on the basis of that, the recognition method for unknown softwares of known steganographic method, as well as the recognition method for the "variations" of known steganographic software will be presented. By theoretical analysis and the analysis of changing rule of steganography to images, the extraction method of image identification features will be studied, and then, the steganography algorithm recognition method for stego images will be proposed. In addition,based on the aspects of the reduction of high dimensional feature, the optimization design of ensemble classifier, and the recognition of steganography encoding parameters, this project will try to present detection and steganalysis methods for modern steganography methods such as HUGO (Highly Undetectable steGO) series steganography. The research of this preoject is expected to make some breakthroughs in the aspects of steganography software recognition, stego-images' steganography algorithm recognition, and steganography encoding parameters recognition, which may provide new ways and means for steganography forensics.

开展隐写取证研究，对发现和判定基于信息隐写的隐蔽通信行为具有重要理论价值和现实意义。现有隐写取证主要依赖于载体中隐秘信息的存在性检测，手段单一。本课题拟从另一角度来探讨图像隐写取证，重点对图像隐写取证中的软件识别和图像检测关键问题展开研究。拟基于逆向分析、软件特征抽取、隐写行为分析等技术和方法，提取图像隐写计算特征、构建隐写软件的隐写行为模型，在此基础上给出基于已知隐写方法的未知软件识别方法，以及已知隐写软件的"变种"软件识别方法；通过理论分析和隐写对图像的更改规律分析，研究图像辨识性特征的获取方法，给出隐密图像隐写算法识别方法；从高维特征降维和集成分类器优化、隐写编码参数识别等角度，探索HUGO（Highly Undetectable steGO）系列新隐写的检测与分析方法。课题期望在隐写软件识别、隐写算法识别、隐写编码参数识别等方面有所突破，为隐写取证提供新的方法和手段。

开展隐写取证研究，对发现和判定基于信息隐写的隐蔽通信行为具有重要理论价值和现实意义。传统的隐写取证主要依赖于载体中隐秘信息的存在性检测，手段单一。针对该问题，本课题组围绕图像隐写取证所涉及到的软件识别与载体检测等方面展开研究，并取得了一系列研究结果。学术论文发表和人才培养超过了预期目标，在“IEEE Transactions on Circuits and Systems for Video Technology”、“IEEE Signal Processing Letters”、“IEEE Access”、 “Journal of Internet Technology”、“Digital Investigation”、《计算机研究与发展》等国内外重要学术刊物上发表相关论文41篇，其中SCI期刊论文26篇，EI论文15篇，1篇被ICCCS2017（International Conference on Cloud Computing and Security）国际会议评为优秀论文。培养毕业博士生4人，硕士生8名，其中2篇博士学位论文分别获2015、2016年度ACM中国郑州分会优秀博士学位论文奖，2人分别获2015、2016年度ACM中国郑州分会新星奖。此外，申请发明专利5项，获得省部级奖励3项，其中2015年度教育部技术发明一等奖、中国电子学会技术发明一等奖各1项，2016年度河南省科技进步二等奖1项。