针对Seam-carving图像篡改的内容自适应检测方法

吴学谦; 曹纭; 赵险峰; 刘长军

引用本文：

吴学谦,曹纭,赵险峰,刘长军.针对Seam-carving图像篡改的内容自适应检测方法[J].信息安全学报,2018,3(6):92-102 [点击复制]
WU Xueqian,CAO Yun,ZHAO Xianfeng,LIU Changjun.A Content Adaptive Method to Detect Seam-Carving-based Image Forgery[J].Journal of Cyber Security,2018,3(6):92-102 [点击复制]

本文已被：浏览 5960次下载 5190次	码上扫一扫！
针对Seam-carving图像篡改的内容自适应检测方法
吴学谦^1,2, 曹纭^1,2, 赵险峰^1,2, 刘长军^1,2
0 字体:加大+\|默认\|缩小-
(1.中国科学院信息工程研究所信息安全国家重点实验室北京中国 100093;2.中国科学院大学网络空间安全学院北京中国 100093))

摘要:

针对借助于Seam-carving缩放技术实现的数字图像篡改，本文提出了一种内容自适应的检测方法。该方法考虑了Seam-carving的操作特点，首先对待测图像中高度疑似篡改区域进行定位，进而仅基于定位的区域进行特征提取及分类。该方法在特征计算时剔除了篡改发生可能性较小的区域带来的影响，从而能够有效提高所提特征的代表性。在特征选择方面，本文选用了扩展Markov特征对Seam-carving操作引起的像素间相关性破坏程度进行度量，实验表明，与先前的非自适应方法相比，新方法在针对性检测有效性方面有着明显的优势。

关键词: Seam-carving 自适应检测 Markov特征数字图像取证

DOI：10.19363/J.cnki.cn10-1380/tn.2018.11.08

投稿时间：2018-09-06修订日期：2018-09-25

基金项目:本课题得到国家自然科学基金（No.U1736214），北京市科委项目（No.Z181100002718001），国家重点研发计划（No.2017YFC0822704，No.2016YFB0801003，No.2016QY15Z2500），中国科学院信息工程研究所基础前沿项目（No.Y7Z0371102）资助。

A Content Adaptive Method to Detect Seam-Carving-based Image Forgery

WU Xueqian^1,2, CAO Yun^1,2, ZHAO Xianfeng^1,2, LIU Changjun^1,2

(1.State Key Laboratory of Information Security, Institute of Information Engineering, Chinese Academy of Sciences, Beijing 100093, China;2.School of Cyber Security, University of Chinese Academy of Sciences, Beijing 100093, China)

Abstract:

This paper proposes a content adaptive method to detect the digital image tampered by Seam-carving. Taking into account the operational characteristics of Seam-carving, in this method, the highly suspectable tampered areas in the image are located first, and then feature extraction and classification are carried out based on those located areas. The extracted features in this method are more representative for the influence of the area with less possibility is eliminated in feature calculation. Based on extended Markov feature, this paper measures the damage degree of pixels correlation caused by Seam-carving. Compared with previous non-adaptive methods, the experiment shows that the proposed method is more effective in targeted detection.

Key words: Seam-carving adaptive method Markov feature digital image forensics