Using a theoretical approach based on random processes, signal processing, and information theory, we study the performance of digital watermarks subjected to an attack consisting of linear shift-invariant filtering and additive colored Gaussian noise. Watermarking is viewed as communication over a hostile channel, where the attack takes place. The attacker attempts to minimize the channel capacity under a constraint on the attack distortion (distortion of the attacked signal), and the owner attempts to maximize the capacity under a constraint on the embedding distortion (distortion of the watermarked signal). The distortion measure is frequency-weighted mean-squared error (MSE). In a conventional additive-noise channel, communication is most difficult when the noise is white and Gaussian, so we first investigate an effective white-noise attack based on this principle. We then consider the problem of resisting this attack and show that capacity is maximized when a power-spectrum condition (PSC) is fulfilled. The PSC states that the power spectrum of the watermark should be directly proportional to that of the original signal. However, unlike a conventional channel, the hostile attack channel adapts to the watermark, not vice versa. Hence, the effective white-noise attack is suboptimal. We derive the optimum attack, which minimizes the channel capacity for a given attack distortion. The attack can be roughly characterized by a rule-of-thumb: At low attack distortions, it adds noise, and at high attack distortions, it discards frequency components. Against the optimum attack, the PSC does not maximize capacity at all attack distortions. Also, there is no unique watermark power spectrum that maximizes capacity over the entire range of attack distortions. T...

One envisioned application of digital watermarking is fingerprinting, in which different information is embedded into several copies of the same original signal. Several attackers may collude by combining their copies to produce an attacked signal. In the case of independent watermarks, a collusion-attack model is presented and shown to be analogous to the Gaussian multiple-access channel. The attack parameters are optimized to minimize the information rate under a constraint on the distortion of the attacked signal. Another fingerprinting method, collusion-secure codes, is then related to the attack. Finally, independent and collusionsecure watermarking are compared for the same attackedsignal distortion and probability of false identification.

This paper presents a review of some influential work in the area of digital watermarking using communications and information-theoretic analysis. After a brief introduction, some popular approaches are classified into different groups and an overview of various algorithms and analysis is provided. Insights and potential future trends in the area of watermarking theory are discussed.

Digital watermarking is the enabling technology to prove ownership on copyrighted material, detect originators of illegally made copies, monitor the usage of the copyrighted multimedia data, and analyze the spread spectrum of the data over networks and servers. Most watermarking methods for images and video can be viewed as a communications problem in which the watermark must be transmitted and received through a watermark channel. This channel includes distortions resulting from attacks and interference from the original digital data [1]. It is well accepted that an effective watermarking scheme may be described as the secure, imperceptible, robust communication of information by direct embedding in and retrieval from digital data. For verifying the security and robustness of watermarking algorithms, specific attacks have to be applied to test them. In this paper, using a theoretical approach based on random processes, signal processes, and communication theory, we propose a stochastic formulation of a new watermarking attack using blind source separation-based concept. The proposed attack consider the watermarking channel as a “black-box”. A host image was passed through the “black-box”, which include the watermarking embedding process, and then the watermarked image was produced. The watermarked image is viewed as linear mixtures of unknown source signals, and then we attempt to recover sources from their linear mixtures without resorting to any prior knowledge by using blind source separation theory.

This paper describes a game-theoretic methodology to design and embed watermarks in images. The optimality criterion is probability of error of the decoder. Analytical solutions are presented for problems involving Gaussian host images and illustrated with examples. Significant improvements over previous designs are obtained. 1.

Abstract—In most watermarking systems, masking models, inherited from data compression algorithms, are used to preserve fidelity by controlling the perceived distortion resulting from adding the watermark to the original signal. So far, little attention has been paid to the consequences of using such models on a key design parameter: the robustness of the watermark to intentional attacks. The goal of this paper is to demonstrate that by considering fidelity alone, key information on the location and strength of the watermark may become available to an attacker; the latter can exploit such knowledge to build an effective mask attack. First, defining a theoretical framework in which analytical expressions for masking and watermarking are laid, a relation between the decrease of the detection statistic and the introduced perceptual distortion is found for the mask attack. The latter is compared to the Wiener filter attack. Then, considering masking models widely used in watermarking, experiments on both simulated and real data (audio and images) demonstrate how knowledge on the mask enables to greatly reduce the detection statistic, even for small perceptual distortion costs. The critical tradeoff between robustness and distortion is further discussed, and conclusions on the use of masking models in watermarking drawn. Index Terms—Attacks, mask attack, masking models, robustness, watermarking, Wiener attack. I.

One of the main problems, which darkens the future of digital watermarking technologies, is the lack of detailed evaluation of existing marking schemes. This lack of benchmarking of current algorithms is blatant and confuses rights holders as well as software and hardware manufacturers and prevents them from using the solution appropriate to their needs. Indeed basing longlived protection schemes on badly tested watermarking technology does not make sense. In this paper we will present the architecture of a public automated evaluation service we have developed for still images, sound and video. We will detail and justify our choice of evaluation profiles, that is the series of tests applied to different types of watermarking schemes. These evaluation profiles allow us to measure the reliability of a marking scheme to different levels from low to very high. Beside the known StirMark transformations, we will also detail new tests that will be included in this platform. One of them is intended to measure the real size of the key space. Indeed, if one is not careful, two different watermarking keys may produce interfering watermarks and as a consequence the actual space of keys is much smaller than it appears. Another set of tests is related to audio data and addresses the usual equalisation and normalisation but also time stretching, pitch shifting. Finally we propose a set of tests for fingerprinting applications. This includes: averaging of copies with different fingerprint, random exchange of part between different copies and comparison between copies with selection of most/less frequently used position