Scalable Processing History Detector for JPEG Images

Abstract

Knowing the history of global processing applied to an image can be very important for the forensic analyst to correctly establish the image pedigree, trustworthiness, and integrity. Global edits have been proposed in the past for "laundering" manipulated content because they can negatively affect the reliability of many forensic techniques. In this paper, we focus on the more difficult and less addressed case when the processed image is JPEG compressed. First, a bank of binary linear classifiers with rich media models are built to distinguish between unprocessed images and images subjected to a specific processing class. For better scalability, the detector is not built in the rich feature space but in the space of projections of features on the weight vectors of the linear classifiers. This decreases the computational complexity of the detector and, most importantly, allows estimation of the distribution of the projections by fitting a mutlivariate Gaussian model to each processing class to construct the final classifier as a maximum-likelihood detector. Well-fitting analytic models permit a more rigorous construction of the detector unachievable in the original high-dimensional rich feature space. Experiments on grayscale as well as color images with a range of JPEG quality factors and four processing classes are used to show the merit of the proposed methodology.

Deep Learning for Detecting Processing History of Images

“…The purpose of the initial experiment in this section is to assess the capability of the proposed CNN in comparison with our previous work [7] that employed a maximum likelihood detector operating in projections of rich feature representations. To this end, we use a rather simple (sandboxed) experimental setup, which is exactly the same as in [7], to limit the distortion applied to the image after processing and to limit the input source diversity. The processing is applied to never-compressed images and then the images are directly saved as JPEGs.…”

“…This work started with our prior research on this topic [7], which was a maximum-likelihood (ML) detector based on modeling the distribution of projections of a rich feature on eigen-processes determined by binary linear classifiers trained to distinguish between unprocessed images and images processed with a fixed class of processing. In this paper, we propose an alternative detector constructed using the tools of deep learning as a CNN with softmax over five output neurons corresponding to four processing classes and the unprocessed class.…”

Facing the Cover-Source Mismatch on JPHide using Training-Set Design

“…We have therefore investigated whether it is possible to use the same feature set for detecting the IPP. Several papers have been published showing the usefulness of steganalysis features for digital image forensics and in particular for detecting image manipulations [3,17]. The ensemble classifier (EC) yielding excellent results for steganalysis using large feature sets, in the current paper the EC is also used for constructing a supervised classifier of the IPP.…”

Recasting Residual-based Local Descriptors as Convolutional Neural Networks

“…In fact, the noise residual, extracted through some high-pass filtering of the image, contains a wealth of information on the in-camera and out-camera processes involved in the image formation. Such subtle traces, hardly visible without enhancement, may reveal anomalies due to object insertion [14], [15] or can detect different types of image editing operations [7], [9], [16].…”