基于重采样检测的JPEG图像拼接篡改取证

数字图像在进行拼接篡改时,为了不留下视觉上的明显篡改痕迹,往往会对篡改的区域进行缩放、旋转等重采样操作。针对这一现象,本文提出一种新的基于重采样检测的JPEG图像拼接篡改取证算法,该算法通过对JPEG图像局部区域二阶导数进行Radon变换,并求其自协方差后进行快速傅里叶变换,在频域中消除JPEG压缩的影响,最后判断该局部区域是否经过重采样操作,以作为判断被检测的JPEG图像是否经过拼接篡改的证据。实验结果表明,本文算法对于经过包括缩放和旋转等重采样操作后拼接成的JPEG图像有较好的篡改取证效果。

Splicing Tampering Forensics of JPEG Images Based on the Resampling Detection

In order not to leave any visual evidence of tampering when the digital images were tampered by splicing, the tampered region might be dealt with some resample operations, such as scaling, rotation, and so on. Targeting this phenomenon, this paper proposes a new JPEG image splicing forgery algorithm based on the re-sampling detection. Firstly, calculating the second derivative of the local JPEG image region and doing the Radon transform to it, then finding the auto-covariance and doing fast Fourier transform to it. After those processing, the impact of JPEG compression in the frequency domain is eliminated. Secondly, judging that whether the local region is operated by re-sampling is to be the evidence of whether the detected JPED images are tampered with splicing. Experimental results demonstrate that the proposed algorithm performs well in tamper forensics of JPEG images spliced after the re-sampling operations that include scaling and rotation.