Image fingerprinting system based on collusion secure code and watermarking method

Abstract

According to watermark security based on the Kerckhoffs' principle, we should mind that illegal users will be able to access to the host signal which is targeted for embedding in a fingerprinting system. In our system, each user's ID is encoded by a fingerprinting code, and then it is embedded into the host signal employing the obfuscation technique. Considering the operation in the obfuscation technique, we discover that the operation is equivalent to a simple spread spectrum (SS) watermarking method using longer sequences. Its effects are intensively evaluated by simulation in this paper.

Obfuscation in digital fingerprinting

“…To our knowledge, there is only one technique which claims to thwart both the MMX and uniform attacks, namely the obfuscation described in [Kur16,KS15]. Previous works have applied obfuscation as one of several advanced features in a complex system, and evaluated the complete system only.…”

“…Obfuscation, as introduced in [KS15], is a new layer between the host and watermarking layers. This is shown in Figure 2.…”

“…Applying the permutation R before the DCT transform T −1 means that we can use a relatively short codeword w, and have the energy spread across a longer host signal. Another approach to prevent attacks on the individual co-ordinate positions is modulation [KS15], which is essentially obfuscation as in Equation (26) with a slightly dierent interpretation. If S •T is an orthogonal matrix, which is the case in many implementation of DCT, we get…”

“…Our previous construction [KS15] uses two layers of obfuscation, or, as we phrased it then, obfuscation in addition to modulation. Eectively this gives the following embedding of a codeword w:…”

“…We analyse the MMX and Uniform attacks which has previously proved eective in simulations, and we demonstrate their ecacy mathematically. Finally, we provide a new mathematical framework to study the obfuscation technique which has thwarted the MMX attack in simulations [KS15]. A security proof for obfuscation remains an open problem, but we do identify some interesting properties which may be relevant for further research.…”