Digital watermarks are catching on quickly as digital artists and publishers begin to post and distribute their copyright works online. Watermarks provide a cheap means of copyrighting information and a way to notify would-be users how to contact the copyright owner to gain permission for use and pay copyright fees. It's the way to go if a copyright holder who needs to distribute works without fear of losing major revenue to unauthorized use.

Digital watermark technology still has a long way to go, however.I say this because while watermarks digitally copyright online materials, they are by no means foolproof, nor do they prevent an unauthorized user intent on reproducing the work.

A digital watermark is similar in function to a traditional watermark. It identifies a person's stationery so a digital watermark underlying a graphic, photo, or document identifies the creator or owner of a certain work. Some forms of watermarking are referred to as signatures, or fingerprints. I was taught there is a functional distinction between watermarks and fingerprints. But be aware that digital watermarking and fingerprinting are often used interchangeably.

A watermark identifies the copyright owner and even relays permission information, a fingerprint is essentially a recorded digital imprint left by the user. An intelligent fingerprint is a telltale script of the violator's location and use(s) of the copyrighted work. To enforce copyright protection, you just follow the fingerprint trail. A publisher "reads" the fingerprints to determine who used the work, how it was used, and how many times it was used. Many watermarking software makers are now offering integrated intelligent fingerprinting features.

The more rudimentary watermarking techniques require the original work with which to compare the digital watermark or the coded value (checksum) used in algorithm-based encoding. The more advanced techniques highlighted below have been developed to eliminate the use of an original to extract the watermarked message.

Spread-spectrum watermarking and steganography are the most common forms of digital watermarking today. Steganography encodes, or hides, coded sequences within the binary file of the image, video, or audio. A classic example of steganography is writing a secret message with invisible ink between the lines of a letter. Likewise, digital steganography hides a message in a computer image or digital audio clip. Digital watermarks can be uniquely created so that each watermarked image persists with each digital copy made, and persistent watermark reproduction is key to deterring digital reproduction. The encoded watermark is often referred to a work's digital signature.

The most available form of watermarking is based on spread spectrum communications techniques. The watermark information is hidden using any modulation scheme (any method of communicating over a continuous channel that works at low signal-to-noise ratios). MIT Media Lab and Dice Company watermarks are prime examples of how spread spectrum-based–watermarks are created via the direct-sequencing technique. Think of the original image as the noise and the hidden message as the signal. In the direct-sequence method, the hidden message chooses a key and uses it to generate a pseudorandom carrier function that is then modulated by the information to be encoded, and added to the original image. To extract the information, an image is demodulated with the carrier generated by the original key.

MIT Media Lab's PixelTag media watermark encodes bits of copyright information in the pixel brightness values, rather than the image itself. The benefit of the PixelTag technique is that the watermark is retained despite changes in format or digital-to-analog conversion such as what occurs in printing. Another key feature is the original image is not required to extract the watermark information, enabling the use of a Web crawler to systematically search for illegal copies of images. The decoding process reveals an actual message (e.g. author's email address) instead of a checksum.

On the other hand, Dice Company's Argent refers to its watermarks as digital signatures, but the process of embedding one into a digital sample stream is not accomplished via a digital signature calculation. These signatures are single, continuously integrated numbers, or messages, over a large area of the carrier signal. Dice's definition of watermark is "a continuous integration of...many repetitions of an informational message over arbitrary carrier signal areas." The information encoded by Argent can be digitally signed to certify its validity when it is extracted. Argent watermarks are removed with an authorized key. Without a key, removal will damage the content.

Two well-known developers in the field of digital rights management, Dr. Jian Zhoa and R. Eckhard Koch, employ a pseudorandom position sequence embedded watermark using their new SysCop system. SysCop has a two-step process to embed and retrieve copyright labels. The first step is to generate a pseudorandom position sequence for selecting blocks where the code is embedded, using extracted image data together with a user-supplied key. This step produces the actual copyright code and a random sequence of locations for code embedding. The second step embeds the code at specified locations, using a set of embedding methods dependent on the type or even the content of the multimedia materials. It also retrieves the code from the blocks specified in the position sequence using different watermarking methods.

The embedded copyright label is reportedly tamperproof and does not reduce the quality of the multimedia data. The method is said to resist damage from data compression, low pass filtering and file format conversion. A Web interface to the SysCop has been developed to provide Web users with services to mark their materials.

(Digimarc is one of the better known watermarks as it is available as an Adobe Photoshop plug-in.) A Digimarc watermark placed throughout the image is created by imitating naturally occurring image variations. To further hide the watermark, the Digimarc automatically varies the intensity of the watermark in order to remain invisible in both flat and detailed areas of an image. The Digimarc watermark contains both ownership and usage permission information. To read the information, the user must have a Digimarc reader, also bundled with tools like Adobe Photoshop.

Although the watermark is embedded digitally within the image, it remains part of the image even when printed and can be read later by scanning the image into a computer. A Digimarc watermark is reportedly retained through copying, editing, and most file format conversions.

Highwater's FBI fingerprints which uses FBI Write and FBI Detector is another Adobe Photoshop watermark plug-in. The tools apply and detect FBI fingerprints. The fingerprint does not increase the image file size, nor degrade the reproducible quality. FBI offers two key advantages. It can detect when a part of a protected image has been used, such as in a photo montage. Most importantly, FBI fingerprints can be detected in printed output, giving owners an extra level of copyright control.

Watermarking alone does not deter reproduction. The watermarks need to be combined with other technical mechanisms (e.g. cryptographic containers) to physically prohibit the reproduction of works. Optical character recognition (OCR) processing software used in scanners raises some technical challenges to preserving watermarks. Intelligent page analysis software is needed to correctly identify and reproduce the watermark's unique signature or line spacing when OCR processing. Otherwise, the watermark becomes scrambled or destroyed.

Background watermarks also can be destroyed by the same tools used to clean up water stains and other discolorations on scanned documents. Watermarking does not restrict users from copying or printing. If the rights holder can restrict printing a document or image, then he or she is assured that the secured materials cannot be reproduced beyond the desktop.