Method for decoding a probabilistic anti-collusion code

Abstract

The invention relates to a method for decoding a probabilistic anti-collusion code intended to identify at least one sequence of code present in a multimedia content having been used to create an illegal copy of this multimedia content, this method comprising a step of estimating the collusion strategy used to constitute the illegal copy associated with a step of identifying the sequences having been used in creating the illegal copy. The estimation of the collusion strategy used and the identification of sequences having been used to create the illegal copy are carried out sequentially and iteratively in order to refine the precision of each of these according to the following steps: Creating two parameters, a vector of probabilities of the presence of a sequence in the collusion and a vector of probabilities θ representing the collusion strategy; Initializing the iterative method by initializing one of the two parameters; Estimating the parameter vector of probabilities of the presence of a sequence in the collusion; Estimating the parameter vector of probabilities θ of the collusion strategy which maximizes the probability of the collusion strategy; Iterating the steps c) and d) until a determined number of iterations or until convergence of the parameter vector of probabilities θ of the collusion strategy; Identifying at least one sequence of the code present in a multimedia content having been used to create the illegal copy.