计算机工程 ›› 2011, Vol. 37 ›› Issue (8): 225-227.doi: 10.3969/j.issn.1000-3428.2011.08.078

• 图形图像处理 • 上一篇    下一篇

基于张量分解的数字图像取证

刘 铭,俞能海,李卫海,周 浩   

  1. (中国科学技术大学多媒体计算与通信教育部-微软重点实验室,合肥 230027)
  • 出版日期:2011-04-20 发布日期:2012-10-31
  • 作者简介:刘 铭(1980-),男,硕士研究生,主研方向:数字图像盲取证;俞能海,教授、博士生导师;李卫海,讲师;周 浩,本科生
  • 基金项目:
    国家“863”计划基金资助项目(2008AA01Z117);国家自然科学基金资助项目(61003136)

Digital Image Forensics Based on Tensor Decomposition

LIU Ming, YU Neng-hai, LI Wei-hai, ZHOU Hao   

  1. (MOE-Microsoft Key Laboratory of Multimedia Computing and Communication, University of Science and Technology of China, Hefei 230027, China)
  • Online:2011-04-20 Published:2012-10-31

摘要: 提出一种基于张量分解的数字图像盲检测方法,从全局处理角度对JPEG压缩数字图像进行真伪盲检测。对于来自某一相机拍摄的一批参考图像组成的张量,利用张量分解的方法,从分解残差中分析提取图像特征,通过支持向量机分类器鉴别待检测图像是否直接来自该数码相机。实验结果表明,该方法对数字图像的来源鉴定具有较高准确性和较强的鲁棒性。

关键词: 图像盲取证, 张量分解, 支持向量机, 数字图像鉴别

Abstract: A blind digital image forensics method is proposed based on tensor decomposition to detect the authenticity of JPEG compressed image in global processing view. Image features are extracted from decomposition residual, which are obtained via tensor decomposition of a group of reference images shot by the claimed camera. Then the Support Vector Machine(SVM) classifier is applied to classify whether the image is from the claimed camera or not. Experimental results show that this method has high accuracy and strong robust.

Key words: blind image forensics, tensor decomposition, Support Vector Machine(SVM), digital image identification

中图分类号: