Exposing Digital Forgeries From JPEG Ghosts

Abstract

Abstract-When creating a digital forgery, it is often necessary to combine several images, for example, when compositing one person's head onto another person's body. If these images were originally of different JPEG compression quality, then the digital composite may contain a trace of the original compression qualities. To this end, we describe a technique to detect if part of an image was initially compressed at a lower quality than the rest of the image. This approach is applicable to images of high and low quality and resolution.

Automated Image Forgery Detection through Classification of JPEG Ghosts

“…Thus, as a preprocessing step for guiding a human expert towards a JPEG ghost location, we consider these three classifiers highly suitable. Additionally, the good discrimination for small values of δ improves over the results reported in [5], which reports δ ≥ 20 as a good quality distance for detection. In [7], detection rates for δ ≤ 10 vary between 50% and 70%.…”

“…We varied the length w of a window between 8 and 64 pixels. Following [5], we excluded very smooth image regions, i. e. windows with an intensity variance below 5 points. For the training of the classifiers, we used 10% of the images from all three classes.…”

“…We briefly restate Farid's ghost observation [5]. Let I q1 be an input image that has been compressed with JPEG quality q 1 .…”

“…The quality difference δ can be decreased to an empirical minimum of 5 points. In such cases, other methods often exhibit difficulties; e. g. [5] and [7] reported increased error rates for δ < 20 and δ < 10, respectively.…”

“…While the majority of the presented approaches rely on statistics, Farid [5] recently presented a perception-oriented SD-JPEG method, called "JPEG ghost" detection. It assumes that the first compression step was conducted on a lower quality level than the second step.…”