计算机工程 ›› 2013, Vol. 39 ›› Issue (4): 165-168,172.doi: 10.3969/j.issn.1000-3428.2013.04.038

• 安全技术 • 上一篇    下一篇

图像隐写分析技术综述

张 军1,熊 枫1,张 丹2   

  1. (1. 广东商学院信息学院,广州 510320;2. 四川大学计算机学院,成都 610207)
  • 收稿日期:2012-05-04 出版日期:2013-04-15 发布日期:2013-04-12
  • 作者简介:张 军(1966-),男,教授、博士,主研方向:多媒体信息安全;熊 枫,副教授、硕士;张 丹,学士
  • 基金项目:
    国家自然科学基金资助项目(60873198);广东省自然科学基金资助项目(10151032001000003);教育部人文社会科学研究规划基金资助项目(12YJA630157);广东省哲学社会科学“十二五”规划基金资助项目(GD11CGL16);广东省高等学校科技创新基金资助项目(2012KJCX0056);广州市科技计划应用基础研究专项基金资助项目(2012J4100068)

Overview on Image Steganalysis Technology

ZHANG Jun 1, XIONG Feng 1, ZHANG Dan 2   

  1. (1. School of Information, Guangdong University of Business Studies, Guangzhou 510320, China; 2. College of Computer Science, Sichuan University, Chengdu 610207, China)
  • Received:2012-05-04 Online:2013-04-15 Published:2013-04-12

摘要: 通过归纳典型专用隐写分析方法和通用隐写分析方法的机制,指出在该领域中,低嵌入率的检测问题、图像源不匹配问题和隐写分析方法的适用性问题是3个亟待解决的问题,进而提出基于富模型和数字取证的隐写分析是两大研究趋势,前者合并不同域的差异特征后,利用集成分类器区分载体和含密图像,后者先用数字取方法证识别图像的类型,再采用该类的隐写分析器检测图像,由此克服图像源不匹配问题,提高检测性能。

关键词: 隐写术, 专用隐写分析, 通用隐写分析, 数字取证, 信息隐藏

Abstract: This paper summarizes the schemes of typical targeted and universal steganalysis methods, and points out three challenges in this area: the detection of low embedding rate, the mismatch between the training image source and the test image source, and the ability to adapt unknown steganography algorithms. And it shows that steganalysis based on rich model and digital forensics are two research trends in the field. The former merges different features from different domains and then distinguishes between cover and stego-images by using ensemble classifier. The latter identifies the image type by using digital forensics in advance, and then detects the image by utilizeing stegoanalysizer of the corresponding type. So that the problem is solved and performance of detection is improved.

Key words: steganography, targeted steganalysis, universal steganalysis, digital forensic, information hiding

中图分类号: