With the prevalence of adopting data-driven convolution neural network (CNN)-based algorithms into the community of digital image forensics, some novel supervised classifiers have indeed increasingly sprung up with nearly perfect detection rate, compared with the conventional supervised mechanism. The goal of this paper is to investigate a robust multi-classifier for dealing with one of the image forensic problems, referred to as source camera identification. The main contributions of this paper are threefold: (1) by mainly analyzing the image features characterizing different source camera models, we design an improved architecture of CNN for adaptively and automatically extracting characteristics, instead of hand-crafted extraction; (2) the proposed efficient CNN-based multi-classifier is capable of simultaneously classifying the tested images acquired by a large scale of different camera models, instead of utilizing a binary classifier; and (3) numerical experiments show that our proposed multi-classifier can effectively classify different camera models while achieving an average accuracy of nearly 100% relying on majority voting, which indeed outperforms some prior arts; meanwhile, its robustness has been verified by considering that the images are attacked by post-processing such as JPEG compression and noise adding.

With the continuous development of computer hardware equipment and deep learning technology, it is easier for people to swap faces in videos by currently-emerging multimedia tampering tools, such as the most popular deepfake. It would bring a series of new threats of security. Although many forensic researches have focused on this new type of manipulation and achieved high detection accuracy, most of which are based on supervised learning mechanism with requiring a large number of labeled samples for training. In this paper, we first develop a novel unsupervised detection manner for identifying deepfake videos. The main fundamental behind our proposed method is that the face region in the real video is taken by the camera while its counterpart in the deepfake video is usually generated by the computer; the provenance of two videos is totally different. Specifically, our method includes two clustering stages based on Photo-Response Non-Uniformity (PRNU) and noiseprint feature. Firstly, the PRNU fingerprint of each video frame is extracted, which is used to cluster the full-size identical source video (regardless of its real or fake). Secondly, we extract the noiseprint from the face region of the video, which is used to identify (re-cluster for the task of binary classification) the deepfake sample in each cluster. Numerical experiments verify our proposed unsupervised method performs very well on our own dataset and the benchmark FF++ dataset. More importantly, its performance rivals that of the supervised-based state-of-the-art detectors.

Nowadays, Deepfake videos are widely spread over the Internet, which severely impairs the public trustworthiness and social security. Although more and more reliable detectors have recently sprung up for resisting against that new-emerging tampering technique, some challengeable issues still need to be addressed, such that most of Deepfake video detectors under the framework of the supervised mechanism require a large scale of samples with accurate labels for training. When the amount of the training samples with the true labels are not enough or the training data are maliciously poisoned by adversaries, the supervised classifier is probably not reliable for detection. To tackle that tough issue, it is proposed to design a fully unsupervised Deepfake detector. In particular, in the whole procedure of training or testing, we have no idea of any information about the true labels of samples. First, we novelly design a pseudo-label generator for labeling the training samples, where the traditional hand-crafted features are used to characterize both types of samples. Second, the training samples with the pseudo-labels are fed into the proposed enhanced contrastive learner, in which the discriminative features are further extracted and continually refined by iteration on the guidance of the contrastive loss. Last, relying on the inter-frame correlation, we complete the final binary classification between real and fake videos. A large scale of experimental results empirically verify the effectiveness of our proposed unsupervised Deepfake detector on the benchmark datasets including FF++, Celeb-DF, DFD, DFDC, and UADFV. Furthermore, our proposed well-performed detector is superior to the current unsupervised method, and comparable to the baseline supervised methods. More importantly, when facing the problem of the labeled data poisoned by malicious adversaries or insufficient data for training, our proposed unsupervised Deepfake detector performs its powerful superiority.

With the development of computer hardware and deep learning, face manipulation videos represented by Deepfake have been widely spread on social media. From the perspective of symmetry, many forensics methods have been raised, while most detection performance might drop under compression attacks. To solve this robustness issue, this paper proposes a Deepfake video detection method based on MesoNet with preprocessing module. First, the preprocessing module is established to preprocess the cropped face images, which increases the discrimination among multi-color channels. Next, the preprocessed images are fed into the classic MesoNet. The detection performance of proposed method is verified on two datasets; the AUC on FaceForensics++ can reach 0.974, and it can reach 0.943 on Celeb-DF which is better than the current methods. More importantly, even in the case of heavy compression, the detection rate can still be more than 88%.

The problem of authenticating a re-sampled image has been investigated over many years. However, few current research proposes a statistical model-based test, resulting in that statistical performance of the re-sampling detector could not be completely analyzed. To fill the gap, we utilize a parametric model to expose traces of re-sampling forgery, which is described with the distribution of residual noise. Afterwards, we propose a statistical model describing the residual noise from a re-sampled image. Then the detection problem is cast into the framework of hypothesis testing theory. By considering the image content with designing texture weight map, two types of statistical detectors are established. In an ideal context in which all distribution parameters are perfectly known, the Likelihood Ratio Test (LRT) is presented and its performance is theoretically established. An upper bound of the detection power can be successfully obtained from statistical performance of LRT. For a practical use, when the distribution parameters are not known, a generalized LRT with three different maps based on estimation of parameters is established. Numerical results on simulated data and real natural images highlight the relevance of our proposed approach.

In current steganalysis, relying on a large scale of samples, widely-adopted supervised schemes require the training stage while few studies focus on the design of a training-free unsupervised adaptive detector with high efficiency. To fill the gap, we investigate an adaptive statistical model-based detector designed for detecting JPEG steganography. First, in virtue of hypothesis testing theory, together with the distribution of quantized DCT coefficients, we establish the general framework of the statistical model-based detector. Second, based on the framework, we mainly analyze the performance of the detector relying on the selection of the statistical model, parameters estimation, and less significant payload prediction. Third, to improve the reliability of detection, based on the strategy of assigning weights for DCT channels, the novel adaptive statistical model-based detectors are proposed to aim at detecting JPEG steganography, involving the channel-selected or non-channel-selected algorithm. Extensive experiments highlight the effectiveness of the proposed methodology. Moreover, when detecting JPEG images adopted by two steganographic schemes with the small payload, the experimental results show the Area Under Curve (AUC) of our proposed optimal adaptive detector can achieve as high as 0.9567 and 0.9895 respectively, which are both better than that of non-adaptive detector.