Protecting multimedia authenticity with ICA vaccination of digital bacteria watermarks

We propose the application of independent component analysis (ICA), via unsupervised neural networks, to authenticity protection for multimedia products. We give an overview of the current state of multimedia authenticity protection, including the requirements of various multimedia applications, current approaches to the problem, and the robustness of the approaches. For watermark security, a covert independent-component watermarking signal can serve as a vaccination against a dormant digital bacteria protecting the multimedia data. Unauthorized removal of the watermark triggers the bacterium, which then responds appropriately against the pirated data. We insure that our digital bacteria meet established requirements for beneficial virus-like programs. Overall, we show how these new approaches contribute to a flexible, robust, and secure system for protecting the authenticity of multimedia products.