Adaptive Detection for Group-Based Multimedia Fingerprinting

Grouping strategy has been proposed recently to leverage the prior knowledge on collusion pattern to improve collusion resistance in multimedia fingerprinting. However, the improvement is not consistent, as reduced performance of the existing group-based fingerprinting schemes than the non-grouped ones is observed when the grouping does not match the true collusion pattern well. In this letter, we propose a new adaptive detection method, where the threshold for the group detection can be adjusted automatically according to the detection statistics that reflect the underlying collusion pattern. Experimental results show that the proposed adaptive detection outperforms nonadaptive detection and provides consistent performance improvement over non-grouped orthogonal fingerprinting schemes under various collusion scenarios.