Scalable fingerprinting scheme using statistically secure anti-collusion code for large scale contents distribution

Fingerprinting schemes use digital watermarks to determine originators of unauthorized/pirated copies. Multiple users may collude and collectively escape identification by creating an average or median of their individually watermarked copies. Previous fingerprint code design including ACC (anti-collusion code) cannot support large number of users, which is a common situation in ubiquitous contents distribution environment. We propose a practical scalability solution, which extends previous ACC codebook generation scheme. We design a scalable ACC scheme using a Gaussian distributed random variable to increase the robustness over average and median attack. We implemented our scheme using human visual system based watermarking scheme, and the fingerprinted copy of standard test images show good perceptual quality. The result shows good collusion detection performance over average and median collusion attacks for large scale user population.