Small representation of IA-W marked fruit image + corresponding IA-W watermark Multimedia Security Research at Purdue University


Digital Watermarking and Data Hiding

With the recent growth of networked multimedia systems, techniques are needed to prevent (or at least deter) the illegal copying, forgery and distribution of digital audio, images and video. It is also desireable to determine where and by how much the multimedia file has been changed from the original. One way to improve one's claim of ownership over an image, for instance, is to place a low-level signal directly into the image data. This signal, known as a digital watermark, uniquely identifies the owner and can be easily extracted from the image. Our goal has been to develop both fragile and robust digital watermarks for images. This research is led by Prof. Edward J. Delp at the Video and Image Processing Laboratory (VIPER).

Watermarking Overview

This sections discusses the general concept of digital watermarking.

Many approaches are available for protecting digital data; these include encryption, authentication and time stamping. One approach to protect images is to add an invisible structure to the image data itself that can be used to authenticate it. These structures are known as digital watermarks. If the image is copied and distributed, the watermark is distributed along with the image. This is in contrast to the (easily removed) ownership-information fields allowed by the MPEG-2 syntax. Figure 1. shows the general watermarking embedding procedure.

Figure 1. Top-level image watermarking procedure.

Several papers are available which provide readable, top-level overviews of digital watemrarking:

R. B. Wolfgang, C. I. Podilchuk, and E. J. Delp, "Perceptual Watermarks for Digital Images and Video", submitted to the Proceedings of the IEEE, 1998. The readme file , PDF file, and the ftp site.

R. B. Wolfgang and E. J. Delp, "Overview of Image Security Techniques with Applications in Multimedia Systems," Proceedings of the SPIE Conference on Multimedia Networks: Security, Displays, Terminals, and Gatweys, Vol. 3228, November 2-5, 1997, Dallas, Texas, pp. 297-308. The readme file, compressed postscript file, PDF file, and the ftp site.


Different Types of Watermarks

This section describes some of the current types of watermarks, and defines some commonly used watermarking terms. Some of the different types of watermarks that have been developed in the past few years are listed below:

watermark excerptVariable-Watermark Two-Dimensional Algorithm (VW2D)

A semi-fragile, spatial watermark we have developed is called the Variable-Watermark Two-Dimensional technique. This technique generates a pseudo-random sequence from a small key file which is uniquely associated with an owner. The sequence, consisting of either {0,1} or {-1,1}, fills multiple 8 x 8 pixel blocks to eventually cover the entire original image. The collection of blocks forms the full watermark, which is the size of the original image. The marked image is formed by arithmetically adding the watermark to the original pixel data.

As part of the marking procedure, the inner (dot) product between each watermark block and corresponding marked image block is computed. Testing consists of computing the inner product between a test image block and the mark, and subtracting this from the previously stored value. This block-based procedure allows us to localize changes made to an image down to an 8 x 8 pixel resolution. A user-defined threshold on the magnitude of the change determines whether a block is unacceptably altered from the original. This allows minor changes to a particular block to be overlooked in examining the authenticity of the entire image. Figure 2 shows the testing procedure for VW2D.

Figure 2. VW2D testing procedure.

The VW2D algorithm has been shown to detect linear and non-linear filtering, and JPEG compression. Sample images, and the corresponding watermark and user-key file are available.

R. B. Wolfgang and E. J. Delp, "A Watermark for Digital Images," Proceedings of the IEEE International Conference on Image Processing, September 16-19, 1996, Lausanne, Switzerland, pp. 219-222. The readme file, compressed postscript file, PDF file, and the ftp site.

R. B. Wolfgang and E. J. Delp, "A Watermarking Technique for Digital Imagery: Further Studies," Proceedings of the International Conference on Imaging Science, Systems, and Technology, June 30 - July 3, 1997, Las Vegas, pp. 279-287. The readme file, compressed postscript file, PDF file, and the ftp site.


Image-Adaptive Watermarks in the DCT and Wavelet Domain

Research on robust, image-adaptive (IA) watermarks is being conducted with Dr. Christine Podilchuk of Bell Labs, Lucent Technologies.

The image-adaptive watermarks are based on the robustness of the basic spread-spectrum technique. The spread-spectrum technique embeds a normally distributed zero mean pseudorandom sequence, W, into the transform coefficients (DCT or wavelet) of an original luminance image X. To verify a (possibly altered) marked image Z, one first obtains the normalized correlation coefficient (the test statistic) between the watermark W and the version of the watermark extracted from Z. A value near 1 indicates the presence of the mark; a small value (< 0.1) indicates that the watermark under test is not in the image Z, or that the image has been severely (and visibly) degraded to the point of removing the watermark.

The work in image-adaptive watermarking shows that using visual models in the embedding scheme can produce an even more robust watermark than the standard spread-spectrum technique. Specifically, visual models allow the user to raise or lower the amplitude of the watermark according to the image content (hence the name, image-adaptive). These visual models provide thresholds for how much a given transform coefficient can change, before such changes are noticeable under standard viewing conditions. Figure 3 shows the procedure for image-adaptive watermarking.

Figure 3. Image-adaptive watermarking procedure.

Several papers on spread-spectrum and image-adaptive watermarking are listed below. Sample images, and their corresponding watermarks are also available.

I. J. Cox, J. Kilian, F.T. Leighton and T. Shamoon, “Secure spread spectrum watermarking for multimedia,” IEEE Transactions on Image Processing, vol. 6, no. 12, pp. 1673-1687, December, 1997.

C. Podilchuk and W. Zeng, “Image-adaptive watermarking using visual models,” IEEE Journal on Selected Areas in Communications. vol. 10, no. 4, pp. 525-540.

R. B. Wolfgang, C. I. Podilchuk, and E. J. Delp, "Perceptual Watermarks for Digital Images and Video", submitted to the Proceedings of the IEEE, 1998. The readme file , PDF file, and the ftp site.

R. B. Wolfgang, C. I. Podilchuk, and E. J. Delp, "The Effect of Matching Watermark and Compression Transforms in Compressed Color Images," Proceedings of the IEEE International Conference on Image Processing, October 4-7,1998, Chicago, Illinois. The readme file, compressed postscript file, PDF file, and the ftp site.


Research Sponsorship

We would like to knowledge support from Digimarc for supporting our work in video watermarking. We also acknowlegde the Center for Education and Research in Information Assurance and Security (CERIAS) at Purdue University.


Publications

Our recent publications in Multimedia Security.

A complete set of recent publications at VIPER.


Useful Resources


Multimedia Security Research at Purdue University - Professor Edward J. Delp