Multimedia authenticity protection with ICA watermarking and digital bacteria vaccination

Abstract

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. An unauthorized removal of the watermark triggers the bacteria payload, which then degrades the quality of the unauthorized data. We argue that such digital bacteria meet all the established requirements for beneficial virus-like programs, and their payload would merely affect pirated media. We show how these new approaches contribute to a flexible, robust, and secure system for protecting the authenticity of multimedia products.