基于Markov过程和伪极坐标快速傅里叶变换的重采样篡改检测  

Resampling tampering detection based on Markov process and pseudo-polar fast Fourier transform

作　　者：周治平[1] 胡成燕[1] 朱丹[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《计算机应用》2014年第11期3323-3326,3331,共5页journal of Computer Applications

摘　　要：针对图像中的重采样篡改操作导致的离散余弦变换(DCT)系数之间相关性的变化,提出了一种新的图像重采样篡改检测方法。首先,提取Markov特征,利用高阶统计量分析重采样图像中离散余弦变换系数之间的关系;然后,将图像所在的笛卡儿坐标映射到伪极坐标(Pseudo-Polar)轴上提取图像的光滑度作为纹理特征,利用纹理特征检测图像的重采样操作;最后,将提取的两类特征输入到支持向量机(SVM)中训练和分类,从而检测出图像中的重采样篡改操作。实验结果表明:所提方法可以检测出图像中的重采样篡改操作,并具有较好的检测率,且对于一定范围内的加噪处理也具有鲁棒性。Resampling tampering in digital images would bring about change of the correlation between the Discrete Cosine Transform（ DCT） coefficients. A new resampling tampering method was proposed to solve this problem. At first,Markov features were extracted, then the relationship between DCT coefficients in images was analyzed by the high-order statistics. Secondly, the Cartesian coordinate was mapped to the pseudo-polar axis, and the smoothness of the image was extracted as the texture feature, then the texture feature was used to detect the resampling operation. Finally, these two types of features were sent into Support Vector Machine（ SVM） for training and classification to detect the resampling tampering operation. The experimental results show that this method can detect the resampling tampering operation in image and has a high detection rate, and it is robust for noise in certain range.

关 键 词：图像取证 重采样检测 MARKOV过程 伪极快速傅里叶变换 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]