计算机工程

• 安全技术 • 上一篇    下一篇

基于Boosting算法融合的图像隐写分析方法

万宝吉,张 涛,侯晓丹,朱振浩   

  1. (解放军信息工程大学信息系统工程学院,郑州 450002)
  • 收稿日期:2012-11-19 出版日期:2013-12-15 发布日期:2013-12-13
  • 作者简介:万宝吉(1986-),男,硕士研究生,主研方向:信息隐藏技术,信息融合技术;张 涛,副教授、博士;侯晓丹、朱振浩,硕士研究生
  • 基金项目:
    国家自然科学基金资助项目(60903221, 61272490)

Image Steganalysis Method Based on Boosting Algorithm Fusion

WAN Bao-ji, ZHANG Tao, HOU Xiao-dan, ZHU Zhen-hao   

  1. (Institute of Information System Engineering, PLA Information Engineering University, Zhengzhou 450002, China)
  • Received:2012-11-19 Online:2013-12-15 Published:2013-12-13

摘要: 现有盲检测技术在实际检测中,由于嵌入算法未知导致检测困难。为此,提出一种基于Boosting算法融合的图像隐写分析方法。通过训练分类器建立不同隐写算法下的分类器模型,利用Boosting算法计算各分类器的分类性能,对各分类器的概率输出进行融合,得到最终检测结果。基于典型空间域隐写算法和JPEG隐写算法的实验结果表明,该方法实现了对多种隐写算法的有效检测,应用Boosting算法融合后整体检测性能提升了约2%。

关键词: 信息隐藏, 数字隐写, 隐写分析, Boosting算法, 分类器融合, 支持向量机

Abstract: The existing detection algorithms are difficult to obtain high detection accuracy when applied to the condition, in which the embedding algorithm of the stego-images is unknown. Therefore, this paper proposes a steganography-unknown image steganalysis method based on Boosting fusion. It obtains various classifying results by establishing steganography algorithm classifier models in the training phase, and acquires the performance of these classifies according to the Boosting algorithm. The final detection result is obtained by combinational rule based on probability output. The detection work is presented to attack the current different spatial domain and JPEG steganographic algorithms. Extensive experimental results show that this proposed method is effective for multi-steganographic algorithms, and Boosting takes advantage of the individual strengths from each detection system and whole detection performance is probably increased by 2%.

Key words: information hiding, steganography, steganalysis, Boosting algorithm, classifier fusion, Support Vector Machine(SVM)

中图分类号: