Computer Science Colloquium

Theory and Applications of Multimedia Information Hiding

Qiang Cheng
University of Illinois


Monday, December 3, 2001
2:30pm
CSB 232


Abstract

The ease of misappropriations of digital data has jeopardized the intellectual property right and the security of digital data. Multimedia information hiding has emerged as a powerful technique to protect the copyright as well as data security. It has also been found very useful as a tool to convey side information to achieve additional functionalities. In this talk, theoretical investigations for multimedia information hiding is presented, and applications and designs are constructed. First, an information theoretic approach is explored to achieve high channel capacity, where an ordinary communication model is used to transmit messages. Then a verification model is considered, where the existence of a specific message is to be verified. Both additive and multiplicative methods are studied. It has been found out that the commonly used correlation detector is far from optimum. New optimal detection schemes are devised and their performance analyses are examined using large deviation theory. The theoretical results are validated by experiments. An interesting visible watermarking scheme exploiting human perceptual models is introduced. A combined audio and video watermarking technique is also proposed. The multimedia information hiding techniques can be applied to copyright protection, content authentication, as well as side information transmission in commerce, journalism, defense, and law.


About the Speaker

Qiang Cheng received his B.S. and M.S. degrees from Beijing University in 1994 and 1996, respectively. Now he is a Ph.D. candidate in the Dept. of Electrical and Computer Engineering, UIUC. He was awarded Invention Achievement Award from T. J. Watson Research Center, IBM in Jan. 2001. He was the holder of Graduate School Fellowship, UIUC, 1997-1998, and Guanghua University Fellowship, Beijing University, Beijing, China, 1995-1996. His research interests include multimedia watermarking, multimedia analysis, statistical learning, computer vision, and communication.