Pedro Comesaña-Alfaro

Detection in quantization-based watermarking: performance and security issues

In this paper, a novel method for detection in quantization-based watermarking is introduced. This method basically works by quantizing a projection of the host signal onto a subspace of smaller dimensionality. A theoretical performance analysis under AWGN and fixed gain attacks is carried out, showing great improvements over traditional spread-spectrum-based methods operating under the same conditions of embedding distortion and attacking noise. A security analysis for oracle-like attacks is also accomplished, proposing a sensitivity attack suited to quantization-based methods for the first time in the literature, and showing a trade-off between security level and performance; anyway, this new method offers significant improvements in security, once again, over spread-spectrum-based methods facing the same kind of attacks.

Optimal counterforensics for histogram-based forensics

There has been a recent interest in counterforensics as an adversarial approach to forensic detectors. Most of the existing counterforensics strategies, although successful, are based on heuristic criteria, and their optimality is not proven. In this paper the optimal modification strategy of a content in order to fool a histogram-based forensics detector is derived. The proposed attack relies on the assumption of a convex cost function; special attention is paid to the Euclidean norm, obtaining the optimal attack in the MSE sense. In order to prove the usefulness of the proposed strategy, we employ it to successfully attack a well-known algorithm for detecting double JPEG compression.

Modeling Gabor Coefficients via Generalized Gaussian Distributions for Face Recognition

Gabor filters are biologically motivated convolution kernels that have been widely used in the field of computer vision and, specially, in face recognition during the last decade. This paper proposes a statistical model of Gabor coefficients extracted from face images using generalized Gaussian distributions (GGDs). By measuring the Kullback-Leibler distance (KLD) between the pdf of the GGD and the relative frequency of the coefficients, we conclude that GGDs provide an accurate modeling. The underlying statistics allow us to reduce the required amount of data to be stored (i.e. data compression) via Lloyd-Max quantization. Verification experiments on the XM2VTS database show that performance does not drop when, instead of the original data, we use quantized coefficients.

Transportation-theoretic image counterforensics to First Significant Digit histogram forensics

First-order statistics of First Significant Digits (FSD) have been recently exploited in multimedia forensics as a powerful tool to reveal traces of previous coding operations. As an answer, adversarial approaches aimed at modifying the FSD histogram and fooling such forensic methods have been proposed. However, the existing techniques have limitations in terms of distortion introduced in the multimedia object. In this paper, a transportation-theoretic formulation of the problem is presented which provides a close-to-optimal solution. Such strategy is tested in a well-known image forensic scenario, where FSDs of 8 × 8-DCT coefficients after single or double quantization are modified in order to restore a certain target histogram and the distortion with respect to the provided compressed image is measured in terms of MSE.

A new method for perspective correction of document images

In this paper we propose a method for perspective distortion correction of rectangular documents. This scheme exploits the orthogonality of the document edges, allowing to recover the aspect ratio of the original document. The results obtained after correcting the perspective of several document images captured with a mobile phone are compared with those achieved by digitizing the same documents with several scanner models.

Are you threatening me?: Towards smart detectors in watermarking

We revisit the well-known watermarking detection problem, also known as one-bit watermarking, in the presence of an oracle attack. In the absence of an adversary, the design of the detector generally relies on probabilistic formulations (e.g., Neyman-Pearsons lemma) or on ad-hoc solutions. When there is an adversary trying to minimize the probability of correct detection, game-theoretic approaches are possible. However, they usually assume that the attacker cannot learn the secret parameters used in detection. This is no longer the case when the adversary launches an oracle-based attack, which turns out to be extremely effective. In this paper, we discuss how the detector can learn whether it is being subject to such an attack, and take proper measures. We present two approaches based on different attacker models. The first model is very general and makes minimum assumptions on attackers beaver. The second model is more specific since it assumes that the oracle attack follows a weel-defined path. In all cases, a few observations are sufficient to the watermark detector to understand whether an oracle attack is on going.

A Random Matrix Approach to the Forensic Analysis of Upscaled Images

The forensic analysis of resampling traces in upscaled images is addressed via subspace decomposition and random matrix theory principles. In this context, we derive the asymptotic eigenvalue distribution of sample autocorrelation matrices corresponding to genuine and upscaled images. To achieve this, we model genuine images as an autoregressive random field and we characterize upscaled images as a noisy version of a lower dimensional signal. Following the intuition behind Marčenko-Pastur law, we show that for upscaled images, the gap between the eigenvalues corresponding to the low-dimensional signal and the ones from the background noise can be enhanced by extracting a small number of consecutive columns/rows from the matrix of observations. In addition, using bounds provided by the same law for the eigenvalues of the noise space, we propose a detector for exposing traces of resampling. Finally, since an interval of plausible resampling factors can be inferred from the position of the gap, we empirically demonstrate that by using the resulting range as the search space of existing estimators (based on different principles), a better estimation accuracy can be attained with respect to the standalone versions of the latter.

On the effectiveness of meta-detection for countering oracle attacks in watermarking

We evaluate the performance of smart metadetection as a way to combat oracle attacks in watermarking. In a recent work, we have shown that few queries are sufficient for a simple metadetector (namely, a metadetector based on the closeness of queries to the watermark detection boundary) to detect an oracle attack. A limitation of our prior analysis is the assumption that all the queries correspond to either honest users or malicious ones. In this paper, we address a more realistic scenario in which honest queries are interspersed with queries derived from an oracle attack. By focusing on this more general situation, we evaluate the performance of the metadetection and derive conditions under which powerful testing is possible.

Flat fading channel estimation based on Dirty Paper Coding

A novel complex flat fading channel estimation scheme is proposed. Contrarily to previous schemes in the literature, this new approach is not based on introducing pilot sequences, but on reducing the interference caused by the information-bearing signal on the estimation-aiding signal by using Dirty Paper Coding. We show through simulations that our method outperforms the Partially-Data Dependent scheme, which is a state-of-the-art technique based on superimposed pilots.

Random Matrix Theory for Modeling the Eigenvalue Distribution of Images Under Upscaling

The stochastic representation of digital images through a two-dimensional autoregressive (2D-AR) model offers a proper way to approximate the empirical distribution of the eigenvalues coming from genuine images. By considering this model, we apply random matrix theory to analytically derive the asymptotic eigenvalue distribution of causal 2D-AR random fields that have undergone an upscaling operation with a particular interpolation kernel. This eigenvalue characterization is useful in developing new forensic techniques for image resampling detection since we can use theoretical bounds to drive the decision of detectors based on subspace decomposition. Moreover, experimental results with real images show that the obtained asymptotic limits turn out to be excellent approximations, even when working with images of small size.