Digital watermarking can be used to embed various types of data, depending on
the particular application and intended use. For example, a watermark in a
digital movie file might simply identify the name or version of the movie.
Alternatively, it might convey copyright or licensing information from the
movie’s creator. Or it might embed a customer or transaction number that could
be used to identify individual payment or transaction data relating to that
particular copy of the movie. But the number of bits that can be contained in a
watermark itself today is typically modest – enough to provide some basic codes
or identifiers, but not enough to include the equivalent of a full sentence of text.
The general elements of a digital watermarking system are as follows.
1. Embedding of watermark in content – Every watermarking application
starts by placing a watermark into digital content. This involves modifying
the content using a special algorithm. The algorithm translates the data to
be conveyed by the watermark into specific, subtle modifications to the
2. Subsequent reading of watermark by device/software – Every
watermarking application includes some capability for the embedded
watermarks to be subsequently recognized. Recognizing the watermark
requires knowledge of the algorithm used to embed it, because the reader
device or software needs to know what modifications to look for. Therefore,
readers are system‑ or vendor‑specific; there are no readers capable of
recognizing and deciphering all watermarks from all watermarking vendors.
3. Back‑end database for determining meaning of watermark – Most
watermarking applications involve maintaining a database for storing and
looking up data associated with specific watermarks. For example, the
information contained in a watermark itself might be simply a serial
number, while the database would enable that serial number to be correlated
with rights information or a specific consumer. Similarly, the information in
a watermark might consist of some type of coded message, requiring access
to the database to decode its meaning.
4. Actions triggered upon reading of watermark – In many watermarking
applications, the recognition or reading of a watermark triggers or enables
some type of action. Some actions may occur automatically, via
appropriately programmed hardware or software that looks for watermarks
and responds in predetermined ways. Other actions may depend on the
individualized decisions and responses of people to whom the information
in the watermark has been communicated.