K. Mıhçak; "An Information-Theoretic Approach to Information Hiding.."

An Information-Theoretic Approach to Information Hiding for Gaussian Sources and Application to Images

Kıvanç Mıhçak

 

Abstract: We consider a communication theoretic approach to the problem of information hiding. In particular, we show that it can be modeled as a communications problem, with side information at the encoder. We begin the talk with a classification of information hiding problems, and provide some background with a few illustrative examples. Then, we present an information-theoretic framework for the information hiding problem and show that we can employ a game-theoretic approach, where attacker and the information hider are two players with opposing costs with different constraints. We present capacity results (i.e., the maximum rate of information that can be reliably transmitted asymptotically as signal length tends to infinity) for discrete memoryless Gaussian sources and parallel Gaussian sources under the squared error distortion metric. We also find asymptotically optimal covert and attack channel distributions. Then, we apply these results to image sources, where we use locally approximately independent identically distributed Gaussian models in the discrete wavelet transform domain, and derive capacity results for images. Next, we show some practical watermarking code constructions based on information-theoretic optimality results. We conclude with discussions and some open problems.

May 3, 2005, 14:00, L063