Digital Watermarking Research

Comments about this page can be sent to bchen15 at comcast.net.

This page has been accessed several times since October 25, 1998.
Page last updated March 25, 2004.

About digital watermarking

Digital watermarking, information embedding, data hiding, and steganography have received considerable attention recently. These terms, which can for the most part be used interchangeably, refer to the process of embedding one signal, called the ``embedded signal'' or ``digital watermark'', within another signal, called the ``host signal''. The host signal is typically a speech, audio, image, or video signal, and the digital watermarking must be done in such a way that the host signal is not perceptibly degraded unacceptably. At the same time, the digital watermark must be difficult to remove without causing significant damage to the host signal and must reliably survive common signal processing manipulations such as lossy compression, additive noise, and resampling.

Many emerging digital watermarking applications are a direct result of the explosive increase in the use of digital media for communications. For example, the ease with which digital signals can be copied has lead to a need for digital watermarking techniques to protect copyrighted music, video, and images that are distributed in digital formats. To satisfy this need it has been proposed to embed a digital watermark within these source signals that is designed to thwart unauthorized copying. This embedded signal could be a digital ``fingerprint'' that uniquely identifies the original purchaser of the copyrighted work. If illicit copies of the work were made, all copies would carry this fingerprint, thus identifying the owner of the copy from which all illicit copies were made. Alternatively, the digital watermark could either enable or disable copying by some duplication device which checks the digital watermark before proceeding with duplication. Such a system has been proposed for allowing a copy-once feature in digital video disc (DVD) recorders. Not all digital watermarking applications relate to copyright protection, however. For example, automated monitoring of airplay of advertisements on commercial radio broadcasts can be facilitated by information embedding techniques as well. Advertisers can embed a digital watermark within their ads and count the number of times the digital watermark occurs during a given broadcast period, thus ensuring that their ads are played as often as promised. In other digital watermarking applications, the embedded signal may be used for authentication of, or detection of tampering with, the host signal. For example, a digital signature could be embedded in a military map. Numerous other digital watermarking applications also exist such as covert communication, transmission of auxiliary information, and backwards-compatible upgrading of legacy communication networks.

Digital watermarking research

As mentioned above digital watermarking systems must be designed to cause as little degradation to the host signal as possible, that is they must minimize the embedding-induced distortion. At the same time, the system must be robust to common signal processing manipulations and deliberate attempts to remove the watermark so that the watermark can be reliably extracted. Finally, one would like to maximize the information-embedding rate, which is defined as the amount of information that can be embedded in a host signal of a fixed size. For example, the information-embedding rate for digital watermarking of an image is typically measured in bits of embedded information per image pixel. These three goals --- minimizing embedding-induced distortion, maximizing robustness and reliability, and maximizing embedding rate --- are typically contradicting and, thus, a major challenge of digital watermarking research is to develop systems that can obtain the best possible trade-off among these three performance goals.

Digital watermarking research efforts in the Digital Signal Processing Group at the Massachusetts Institute of Technology have lead to a new class of digital watermarking techniques called quantization index modulation (QIM) that efficiently perform the trade-offs among embedding rate, embedding-induced distortion, and robustness. An example of this class of techniques that has been developed by the group is called dither modulation, in which the embedded signal modulates the dither signal of a dithered quantizer. Quantization index modulation techniques in general, and dither modulation in particular, have considerable performance advantages over previously proposed spread-spectrum and low-bit(s) modulation techniques and are described in the following publications:

B. Chen and G. W. Wornell, " Quantization index modulation: A class of provably good methods for digital watermarking and information embedding," IEEE Trans. on Information Theory, vol. 47, no. 4, pp. 1423-1443, May 2001. [pdf]

B. Chen and G. W. Wornell, " Quantization index modulation methods for digital watermarking and information embedding of multimedia," Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology, Special Issue on Multimedia Signal Processing, vol. 27, no. 1-2, pp. 7-33, Feb. 2001. [Invited paper.] [pdf]

B. Chen, Design and Analysis of Digital Watermarking, Information Embedding, and Data Hiding Systems, Ph.D. Dissertation, MIT, Cambridge, MA, June 2000. [pdf]
Companies interested in developing commercial digital watermarking products based on QIM or dither modulation can contact the MIT Technology Licensing Office for licensing information about several related pending patents:
B. Chen and G. W. Wornell, Multiple patents pending in digital watermarking, data hiding, and information embedding. Licensing info: MIT Technology Licensing Office.

More information about the two inventors of this technology, Brian Chen and Prof. Greg Wornell, is available at their respective web pages. An on-line list of Brian Chen's publications is also available. Finally, a list of recent publications of the entire Digital Signal Processing Group is available at the group's website.