WEST LAFAYETTE, Ind. -- Researchers have created an invisible "digital watermark" to protect copyrights for images placed on the World Wide Web.
The watermark, developed at Purdue University, can be added to multimedia images, everything from news photos and NASA pictures of the Martian landscape to original art and video.
"The idea is to ensure the intellectual property rights of people who create digital media," says Edward Delp, professor of electrical and computer engineering at Purdue. "Adding a digital watermark to an image identifies the owner and protects copyrights."
Some television networks currently use a visible video watermark, a small network logo that appears in the lower corner of the television screen.
"We're interested in adding a watermark similar to that to digital or video images, but in such a way that you can't see it," Delp says. "It's invisible, but I can extract it to verify the image is mine."
Delp will present a paper on his research group's digital watermarking research Monday (10/5) in Chicago at the Institute for Electrical and Electronics Engineers' International Conference on Image Processing. Delp's research team includes colleagues Christine Podilchuk, at Bell Laboratories, and Raymond Wolfgang, one of Delp's graduate students.
Several companies are using their own digital watermarking techniques, and a few commercial programs also are available. Delp's method is different from others because it is tailored for Web-based imaging.
The technique Delp developed embeds the watermark in the very pixels of an original digital image. A pixel is the basic unit or picture element that makes up an electronic image.
For example, one of Delp's watermarks is a pattern of black speckles on a white background. The pixels of the watermark are "added" to, or superimposed upon, the pixels in an original image, but integrated in such a sophisticated way that a typical viewer cannot detect any change. Only a very close, detailed examination of the watermarked image on the pixel level would reveal any alteration.
The procedure can be used on any type of image file, and the watermark can be added before or after an image is compressed. If the image is copied and/or altered, the watermark is copied and/or altered as well.
"We can search the Web for our watermark in images in order to find unauthorized copies," Delp says. "We can find the watermark even if it was altered when the image was altered."
Delp's technique also can determine the area of the image that has been changed, which can help detect forgeries. On his office computer, Delp calls up a photo of the Mars landscape, taken last summer on the planet's surface by cameras on the Mars rover. In the sky of this image is an unidentified flying object.
"Someone went in and put an object into the picture so that it looks like the explorer's camera had captured a craft flying through the sky on Mars," Delp says. "If that original image had been watermarked, then you could prove the image had been altered and in what way it had been altered."
Delp says a watermark can be made unique for a particular image, or it could be a unique "signature" of the creator.
Delp, who also teaches courses in cryptography and secure communications, has been working on the watermarking project for about four years. He and his Purdue team currently are working with an electronics company to incorporate watermarking technology into digital cameras.
"Watermarking technology can help verify whether an image taken with a digital camera is authentic," Delp says. "If the image is watermarked when it is taken, you would be able to verify later whether the image has been altered." This is a particularly important issue in the case of verifying the authenticity of news images, historical images, and images that might be used as evidence in court.
Delp says his Purdue research group is on the forefront of digital watermarking technology, but other research groups around the country are pursuing research in this area as well. Delp's group is concentrating on images, video, audio and security for scanned documents.
Digital watermarking technology is beginning to be incorporated in commercial computer security packages, and Delp says it will become more widespread in the near future.
"Audio companies, the people making compact disks, want to put watermarking technology into the audio file so that you can't copy it and redistribute it," Delp says. "The same thing is happening with video. The DVD technology that stores video on a disk will be watermarked and copy protected so that you physically won't be able to copy the videos."
Purdue has applied for patents for the technology, and Delp's watermarking technique has been licensed by Louisiana-based Intellectual Protocols 2, or IP2, a computer technology company. IP2 has incorporated the Purdue technology into a product called Copysight.
"Digital watermarking is not foolproof," Delp says. "You really need to embed it in a larger security product that includes a whole set of security tools, and that's essentially what Copysight is."
Delp's research currently is supported by IP2 and Texas Instruments, and also has been funded by the AT&T Foundation.
Source: Edward Delp, (765) 494-1740; e-mail, ace@ecn.purdue.edu;
Web,
http://dynamo.ecn.purdue.edu/~ace/
Writer: Amanda Siegfried, (765) 494-4709; e-mail, amanda_siegfried@uns.purdue.edu
Web links
Edward Delp: http://dynamo.ecn.purdue.edu/~ace/
Intellectual Protocols 2: http://www.ip2.com/
Copysight: http://www.ip2.com/copysight.cgi
Abstract
The Effect of Matching Watermark and Compression Transforms in Compressed
Color Images
Raymond B. Wolfgang, Christine I. Podilchuk, Edward J. Delp
The growth of networked multimedia systems has complicated copyright enforcement relative to digital images. One way to protect the copyright of digital images is to add an invisible structure to the image (known as a digital watermark) to identify the owner. In particular, it is important for Internet and image database applications that as much of the watermark as possible remain in the image after compression. Image adaptive watermarks are particularly resistant to removal by signal processing attacks such as filtering or compression. Common image adaptive watermarks operate in the transform domain (DCT or wavelet); the same domains are also used for popular image compression techniques (JPEG, EZW). This paper investigates whether matching the watermarking domain to the compression transform domain will make the watermark more robust to compression.
Note To Journalists: Ed Delp will present a paper on digital watermarking technology at 10 a.m. Oct. 5 at the IEEE International Conference on Image Processing, Westin Hotel, Chicago. He can be reached through Oct. 2 and after Oct. 9 at Purdue at (765) 494-1740. He can be reached in Chicago Oct. 4-7 at (312) 943-7200.