Resampling forgery detection in JPEG-compressed images

When digital forgery is introduced in a digital images, in order to create a convincing match, it always includes operation of scaling, rotation or skewing, which procedures typically require resampling and interpolation step. However, the resampling detection is always severely affected by the blocking artifacts introduced by JPEG compression. In order to solve this problem, in this paper an effective method has been proposed to distinguish the periodicity introduced by resampling and JPEG compression. At first, the probability map of an image was obtained by using the expectation-maximization algorithm. Then, it was Fourier-transformed and matched with affine-transform templates and a scaled JPEG template to detect whether the image was resampled, JPEG-compressed or both of them. Experimental results are provided to show the given method is dependable and effective in different kinds of digital tampering when resampling and JPEG compression occurred. Moreover, compared to the existed methods, better results are obtained by us.