基于条件共生概率矩阵的移位JPEG双压缩检测  

Detecting shifted JPEG double-compression based on conditional co-occurrence probability matrix

作　　者：张玉金[1] 袁野[2] 王士林[3] 李生红[1] 赵旭东[1] 

机构地区：[1]上海交通大学电子工程系上海200240 [2]上海对外贸易学院商务信息学院,上海201620 [3]上海交通大学信息安全工程学院,上海200240

出　　处：《光电子．激光》2012年第10期1932-1939,共8页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61071152,60702043,61271316);国家重点研究发展“973”计划(2010CB731403,2010CB731406);“十二五”国家科技支撑计划重点(2012BAH38B04)资助项目

摘　　要：针对合成JPEG图像的小区域移位JPEG双压缩(SD-JPEG压缩)篡改问题,提出一种基于条件共生概率矩阵(CCPM)的SD-JPEG压缩篡改检测算法。为了减小图像内容的影响,增强SD-JPEG压缩效应,首先对JPEG量化的离散余弦变换(DCT)系数的幅度矩阵进行水平、垂直、主对角和副对角4个方向差分和阈值化处理,然后使用CCPM对这4个阈值化的差分矩阵进行建模,选取CCPM的元素作为特征数据,并用主分量分析(PCA)对其降维处理,最后通过支持向量机(SVM)技术判决图像块是否经过SD-JPEG压缩。实验结果验证了本文算法的有效性。For the small tampered region coming from two times of JPEG compression with inconsistent block segmentation in composite JPEG images,in this paper,an effective shifted double JPEG（SD-JPEG） compression tampering detection method based on conditional co-occurrence probability matrix（CCPM） is proposed.In order to reduce the effects caused by the diversity of image content and enhance the SD-JPEG compression,difference 2-D arrays are generated along four directions,i.e.,horizontal,vertical,main diagonal and minor diagonal directions,for magnitudes of JPEG quantized discrete cosine transform（DCT） coefficients.The thresholding technique is used to deal with these difference arrays for reducing computational cost.Conditional co-occurrence probability matrix is used to model these thresholded difference 2-D arrays.All elements of these conditional co-occurrence probability matrices are served as features for SD-JPEG compression tampering detection.Principal component analysis（PCA） is introduced to reduce the dimensionality of the proposed features.Support vector machine（SVM） is employed as the classifier to identify whether an image block has been SD-JPEG compressed.Experimental results demonstrate the effectiveness of the proposed method.

关 键 词：被动图像取证 复制-粘贴篡改 移位JPEG双压缩(SD-JPEG压缩) 条件共生概率矩阵(CCPM) 主分量分析(PCA) 支持向量机(SVM) 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]