Digital watermark
A digital watermark is a technical mark in carrier media (works) such as image, video or audio data or text that is work-related and mostly imperceptible. In contrast to metadata , digital watermarks are intertwined directly with the content to be marked using steganographic methods . Digital watermarks are therefore seen as a commercial application of steganography. They are used, for example, to embed rights holder information in a work for the purposes of copyright infringement tracking. In various research and development projects, which were initially limited to the image area, methods were developed mainly for commercially relevant content such as audio and video data and for 3D models. There are also methods such. B. for marking databases , geodata , music notes , still images and texts.
Depending on the design, digital watermarks can also withstand digital-analog-digital conversions, i. H. the carrier medium does not necessarily have to be digital. A prominent example of such a watermark can be found on euro banknotes. Another example are digitally projected cinema films, the locations of which can be taken from the location of the performance, filmed as a digital watermark.
For example, digital watermarks can prove the authenticity of a file and ensure its traceability. A carrier can contain several different watermarks at the same time. In contrast to conventional watermarks , digital watermarks cannot be perceived directly by humans, but are intended to be detected and read out only using a prescribed, likewise digital process.
Differentiation from steganography
In contrast to steganography, with digital watermarks it is usually not the imperceptibility and secrecy of the use of steganography that comes first, but the robustness against attacks. In extreme cases, this can lead to noticeable changes in the carrier medium. Robust digital watermarks are usually designed in such a way that the elimination of the digital watermark degrades the wearer into an unusable state.
Watermark Criteria
Watermarks and their algorithms are classified according to various criteria, which are explained in more detail in this section. Understanding them is necessary in terms of applicability in a given case.
The criteria are:
- Blindness to the wearer
- Perceptibility
- Publicity
- Fragility / robustness
- Embedding space
Blindness to the wearer
In the case of digital watermarks, one speaks of blindness to the wearer when the carrier is not needed to read the watermark. Accordingly, one method is non-blind when the original signal is needed to be read out.
Normally, non-blind schemes are more robust than blind ones, since the differences between the watermarked signal and the original can be trivially determined. In practice, on the other hand, one mainly encounters blind watermarking algorithms, since the lack of need to have access to the original data makes reading out more flexible.
Perceptibility
There are noticeable and imperceptible watermarks. Perceptible watermarks are, for example, logos or copyright notices that are subsequently incorporated into images in order to identify the author. In contrast, imperceptible watermarks aim not to change the perceptible impression of a work. Imperceptibility has been cited by many scientists researching in the field as a necessary criterion for a watermark to be a digital watermark .
Publicity
Watermark algorithms can be distinguished in terms of their public availability or their area of use. Private algorithms can only be detected by appropriately authorized users. With these algorithms, priority is given to ensuring that unauthorized users cannot read out the private watermark. In contrast to these private algorithms, public watermarks are readable for everyone.
In general, private watermarks are more robust than public ones, since knowledge of the watermark makes it easier for an attacker to disrupt or remove public watermarks.
Fragility / robustness
One of the most important criteria by which watermarks are distinguished is their fragility (fragility) or - inversely - their robustness. Robust digital watermarks withstand processing steps such as cropping, enlargement / reduction, digital-analog-digital conversion, etc., depending on the process. This type of digital watermark is widespread due to its use in copyright infringement proceedings . On the other hand, fragile methods can be used to prove integrity and authenticity. The procedures that fall under this can be broken down into global approaches, which only allow binary statements, and local approaches, through which manipulated areas of the carrier can be shown, since the digital watermark is destroyed there.
There are hybrid forms of fragility, so-called semifragile watermarks. There are processes that are robust to certain processing steps, but fragile to others. For example, lossy storage can be tolerated, but processing such as trimming cannot.
Embedding space
For embedding a watermark, signals can be modified as such as well as in a transformed state. The former is referred to as a modification in local space or, in the case of audio signals, as a period . Transformed signals are changed in the frequency domain . Examples of transformations are Fourier transformation , discrete cosine transformation or wavelet transformation .
The embedding in the frequency domain generally leads to increased robustness, since, for example, amplitude normalization in the spatial domain has no significant effects on the transformed signal.
Applications
The type of embedded message is determined by the application. This can e.g. B. in the case of proof of authorship information about the copyright holder. In the general case, metadata is embedded. (English meta data labeling ).
However, due to practical limitations of the message length ( capacity ), not all metadata can be embedded directly. Accordingly, considerably shorter link information to detailed data is embedded via the medium (see primary key ).
Examples of special applications are:
| application | possibly embedded information |
|---|---|
| Recognizing a medium | Unique identification number of the content comparable with the ISBN |
| Proof of authorship | Author's identification number |
| Proof of the rightful owner (by personalized copy) | Customer number, credit card number |
| Flag to track data flows | Transaction number possibly in connection with a user identification number, e.g. B. by marking laser color printouts (transaction watermarks) |
| Identification of media for advertising purposes | Number to identify the respective advertising measure |
However, the names for different applications are not always clear. Instead of the term transaction watermark, the term fingerprinting is also used (based on the idea that the customer leaves their fingerprint on their content). The term fingerprinting is also used for the recognition of a content based on its intrinsic features, as well as for perceptual hashing processes that calculate a characteristic digital fingerprint for digital content.
Both process classes - digital watermarks and digital fingerprint processes - belong to the class of passive protection mechanisms. In contrast to active protection mechanisms, such as For example, encryption , these mechanisms do not prevent unauthorized access to content.
They are therefore seen by various groups as a better alternative to DRM . Due to their robustness against format conversions and various operations, they can cover a wide range of applications.
The reversible watermarking methods are a special class. Here the embedded watermark can be removed and the original message restored. For this, the recovery information is placed in the watermark in addition to the newly introduced (watermark) data. Applications of this type of watermark can be found e.g. B. in medical image processing.
Characteristics
Watermarking processes have various properties:
- The perception is the influence on the quality of the selected content. The differential perceptibility threshold is often used to evaluate the perceptibility compared to the original ,
- the robustness , as described above, and
- the capacity , which is determined by the amount of information that can be put into a medium. It ranges from binary watermarks (marked or unmarked) to watermarks with a capacity of several hundred bytes (depending on the carrier size and method).
These characteristics are interdependent. This can be illustrated with a triangle with these three properties at the corners. If two parameters are selected, the third parameter results.
Further characteristics for differentiating between different procedures are
- the (non) detectability , which indicates how easily a watermark can be recognized by analyzing the processed data stream (e.g. by a spectrogram or bar analysis methods ), and which is strongly related to the
- Security attack (engl. Security ) corresponding to the defined to be operated at the expense deliberate removal of the marking and depends on the reversibility of the embedding algorithm, and
- the blindness , which expresses whether the original data set is required to read out the information of the watermark, or whether the embedded data can be obtained directly from the marked data stream.
The significance of the individual features depends on the respective application. Usually, perceptibility is the most important criterion, followed by robustness and capacity.
Specifically, the dependency described above means that perceptibility cannot be selected independently of robustness: the less an embedded message is to be perceived, the lower the possible changes to the content without exceeding the perceptibility threshold. At the same time, these changes can also be removed more easily.
In addition, there are other features depending on the respective application scenario, such as B. watermarking minimum segment (WMS, smallest subdivision of a content that contains a complete watermark message ) or the real-time capability of the algorithm.
See also
Web links
Process-specific information:
Individual evidence
- ^ A b c Neil F. Johnson, Zoran Duric, Sushil Jajodia: Information Hiding . Steganography and Watermarking - Attacks and Countermeasures. Kluwer Academic Publishers, 2001, Chapter 2.2.2 Watermarking Techniques.
- ^ Fabien Petitcolas, Stefan Katzenbeisser: Information Hiding Techniques for Steganography and Digital Watermarking . 1st edition. Artech House, Boston MA 2000, ISBN 1-58053-035-4 .
- ↑ Kerstin Kohlenberg: Pirate copiers - The film industry is fighting against Internet pirates . In: The time . No. 7 , 2013, p. 15 .
- ↑ Jana Dittmann : Digital watermarks: Basics, procedures, areas of application . 1st edition. Springer Verlag, Berlin / Heidelberg 2000, ISBN 3-540-66661-3 .
- ^ A b Frank Y. Shih: Digital watermarking and steganography: fundamentals and techniques . 1st edition. Taylor & Francis, Boca Raton FL 2008, ISBN 978-1-4200-4757-8 .
- Jump up ↑ Ingemar J. Cox, Matthew L. Miller, Jeffrey A. Bloom, Jessica Fridrich, Ton Kalker: Digital watermarking and steganography . 2nd Edition. Morgan Kaufmann, Burlington MA 2008, ISBN 978-0-12-372585-1 .