Game theory based false negative probability of embedded watermark under unintentional and steganalysis attacks

Steganalysis attack is to statistically estimate the embedded watermark in the watermarked multimedia, and the estimated watermark may be destroyed by the attacker. The existing methods of false negative probability, however, do not consider the influence of steganalysis attack. This paper proposed the game theory based false negative probability to estimate the impacts of steganalysis attack, as well as unintentional attack. Specifically, game theory was used to model the collision between the embedment and steganalysis attack, and derive the optimal building/embedding/attacking strategy. Such optimal playing strategies devote to calculating the attacker destructed watermark, used for calculation of the game theory based false negative probability. The experimental results show that watermark detection reliability measured using our proposed method, in comparison, can better reflect the real scenario in which the embedded watermark undergoes unintentional attack and the attacker using steganalysis attack. This paper provides a foundation for investigating countermeasures of digital watermarking community against steganalysis attack.