G.Leela Sankar, III B.Tech, ECE, P.Jnananand, III B.Tech, ECE,

In the wake of increasing copyright misuses, several new technologies are developed to preserve the authentication of information. Digital water marking is one of them. It is a technique that allows us to add hidden copyright notices, verification messages or any other useful information to the original signal. The hidden message can be a group of bits that describe something about the signal or about the genuine author of the signal. The description can be about the name, place or something related to this. Much of the research wo rk has been done in the field of digital watermarking in the recent period of time. In this paper, a brief description about various digital water marking techniques for different media types is given. Watermark insertion and watermark decoding is described as well. The watermarking techniques described for various media types include audio water marking, video water marking, image water marking and text water marking. Some of the properties and applications are also discussed.

JNTUCE, Kakinada.
watermarking is classified basing on the ability to withstand attacks, into robust and fragile watermarking. Fragile watermark is capable of detecting any minute transformation made to the watermarked content of the host signal. It is easy to embed this into the signal. This can be accomplished by accommodating it in the unimportant portions of the original data. But it cannot withstand wide range of attacks. For the applications wherein wide range of attacks needs to be tolerated, a robust watermark is used.

Original watermark F

Watermarked signal Watermark embedding


Distortion medium

To begin with, the meaning of “watermark”(on a paper) is a mark that is not perceptible to the naked eye in ordinary circumstances but is visible when focused through a special light. A digital water marked signal is also difficult to differentiate from the original signal normal circumstances. In fact, there should be no detectable difference between the original signal and watermarked signal. The desire to communicate secretly is a human trait that dates back to the earliest times. This led to the invention of steganography at first and encryption later on. Before the origin of watermarking technique, there were two techniques called “steganography” and “encryption”. Steganography takes its origin from the old Greek language that means ‘cover writing’. It is nothing but, a way of transmitting hidden messages between two persons. Encryption is used to provide communication privacy and deny access to unauthorized persons from damaging the content. The usage of the watermark does not limit the usage of the original file in any way. Watermarking can basically be classified into visible and invisible watermarking. Visible watermarking obviously includes usage of logos, trademarks and other related things for unique identification. Invisible
Corrupted signal Actual


Host H
Watermark decoding


Decoded signal

key C


Fig 1: block diagram of a watermarking system Robust watermarking techniques are especially useful for military applications, authentication, copyright protection and so forth. Herein, the watermark is placed in the significant portion of the data to be watermarked. All in all, the digital watermarking techniques are expected to withstand assails including compression,

A/D and D/A conversion, linear and non-linear filtering and additive noise as well.

3.1 Digital audio watermarking Here in, the watermark i.e. the digital audio watermark consists of the information pertaining not only to the creator of the audio information that could be any one of several forms of song, music etc., but also to the authenticated user of the audio information. This allows that particular information to be used by only the authorized user and no one else. The watermark can also be used for secondary applications other than authentication. e.g. providing information about the song, album, style of tunes, etc. This watermark is useful in voice conferencing halls to indicate the person who is speaking at the moment. This application can be extended to videoconferences as well. The method of inserting watermark consists of concealing the watermark in the low frequency components of the message signal (audio signal), so that the insertion of watermark is not noticeable by the human ear.

the power contained in each frame, which is useful in the later parts of the method. • From the above calculations, the low frequency components are removed from the signal. • Now, the watermark is placed in the low frequency region, where the low frequency components are removed earlier on. This finishes the process of inserting the watermark in the audio file. Now that the process of insertion is accomplished, the user has to extract the watermark and get it verified with his license to access the information in the audio file. The process of extraction of watermark is similar to the process of inserting the watermark as well.

water marked Partitioning into frames signal

Ananlysis using FFT

Carrier processing

Fig 3. Block diagram for removal of watermark from the audio watermarked file. Ø The watermarked signal to be processed is subjected to framing of some finite duration. This amount of time must be the same as that of duration during insertion process. Once again, spectral analysis is carried out using techniques such as FFT, to calculate the low frequency components and the power constituted in each frame as well. From the above analysis, the power contained in the low frequency components is observed which determines the watermark.

Message Signal

Partitioning Into frames

Analysis for spectral comp.

Elimination of low frequency comp. componenco


Insertion of watermark

Watermarked signal Aw Fig 2.block diagram for audio watermark insertion • • At first, the audio signal to be processed is partitioned into some frames, allotting a finite time interval to each frame. Following this, analysis is done to calculate the spectral components, which is accomplished using FFT, more often than not. This process also allows us to find out

Ideally, this watermark should be exact reproduction of the original watermark that was inserted during insertion process. This is subje cted to certain constraints like power in each frame.

3.2 Digital video watermarking A watermark inserted in a video signal can be used for DVD copy protection. First of all, the watermark

should satisfy the requirements of strength and imperceptibility. Besides this, it can be bestowed on to convey the information to restrict copying in many ways such as copy never i.e. the DVD is copy protected, copy no more i.e. it is possible to copy a finite number of times, copy liberally i.e. i can be t copied any number of times, copy only once. The design depends upon the content in the DVD.A general process of digital video watermarking consists of inserting a group of bits along with the video frame to be transmitted. The process of insertion of watermark is through a ‘hidden key’ that is known to the transmitting person and the receiving person as well.

space required for a digital form of the video signal. This is essential to meet the bandwidth requirements. So it is necessary that the watermark signal be inserted directly into this signal, which in compressed form without having to decode the video signal to reduce intricacy.

Received signal

Extraction process


Hidden key Fig 5. Extraction of watermark from watermarked signal
Video watermarking process

actual video signal watermark hidden key

watermarked signal to be transmitted

This is accomplished by changing the span of the codes and subsequently inserting the watermark. This eases the insertion process. The drawback t this o process is that the watermark is vulnerable to remove entirely at the time of detection. To make the watermark stable even after detection, the watermark is inserted by imposing energy differences between different parts of a video signal.

Fig watermark insertion 3.3 Image watermarking: Actual video signal refers to a single image of a video sequence. Now that, the watermark is inserted into an image of a video sequence, it is susceptible to wide range of attacks in the channel through which it is transmitted. These attack could include deliberate attacks and attacks by chance, i.e. not deliberately done. Some of the deliberate attacks are reducing frames, conspiracy and geometric attacks. Herein, the signal is not completely damaged or distorted beyond use, for the pirate who attacks thus would like to use it anyway. Other range of attacks that the signal could suffer is those that arise from the errors due to transmission and also due to errors generated during processing of the signal. Now, the process of extraction or removal of watermark is an intricate process. This process consists of extraction of watermark from the received watermarked signal, provided the ‘hidden key’ is known. By the time the signal arrives at the watermarking process, it might in a compressed form i.e. it might have undergone compression techniques. Compression is a technique that reduces the storage For watermarking a still image, the watermark is directly inserted on to the picture information itself such as colour, brightness, and luminescence etc. the watermark may be inserted directly on to the pixels, which are either in the spatial domain or in the transform domain. Earliest form of spatial domain techniques consists of inserting a finite number of sequences of bits into the least significant data of the host signal. This type of insertion provides for efficient visible insertio n technique. Another advantage of using this technique is that the implementation requires very low cost. Later on this technique was extended to two dimensions as well. This also proved to be an effective fragile watermarking technique. While decoding, a correlation detector is used for the retrieval of watermark. Another method where a correlation detector is used does the watermarking by introducing a texture-based watermark into the region, which is quite identical to the original image. This enhances the transparency between the watermarked and the actual image.

Watermarking the image in transform domain comes up with many advantages like capability of withstanding the compression techniques. 3.4 Text watermarking: Text watermarking is the process of inserting watermark into the text for preserving the creator’s identity. This is especially applicable to such valuable texts such as poetry. An earlier form of watermarking the texts was accomplished by inserting spaces in specific parts of the text. The disadvantage associated with this form of watermarking was that it was quite easy to remove the watermark. Later on, watermarks were placed in the text that depends on linguistic properties like synonyms, sequence of words etc. minor modifications are made to characters, and such characters are used for watermarking. Text watermarking schemes are applicable to any type of texts from usual word documents to PDF files. At detection stage, the watermark is removed by processing which involves the correction for noise, and distortion introduced, if any (during the insertion process) as well. This scheme protects texts like poetry well as, the changes brought about in the text will damage the worth of it. This is also resistant to attacks like photocopying because these do not alter the watermark.

Mid frequency components are chosen for embedding the watermark in a simple way, owing to the shortcomings with the low frequency components and high frequency components. The difficulty with the low frequency components is that they are highly susceptible to distortion. The high frequency components do not bear much of the information reliably, since their removal from the actual signal does not affect the signal’s quality too much. The watermarking process is a process known to all, thus the safety of the signal completely depends up on the hidden key (confidential key ‘C’ shown in the Fig 1). This key ‘C’ determines the sequence of bits, which are to be embedded into the host signal. The key acts in such a way that, it would be difficult (almost impossible) to recover the original signal without knowing the confidential key ‘C’. There are simple additive watermarking methods, which can be represented using simple mathematical equation: F’= H + P Where F’ is the watermarked signal, ‘H’ is the host signal (from fig 1). The new signal ‘P’ is nothing but the signal originating from the hidden key ‘C’ and the original watermark ‘F’. This method of representation is not applicable to all embedding processes. In some of the processes, the properties of the host signal are used for the embedding process.



Fig 6. Diagram showing changed ‘N’(character alteration). . 1. Here the word “fundamental” is printed in the ordinary way. 2. It shows the word “fundamental”, in which the letter ‘N’ is modified slightly.

‘Detection’ and ‘identification’ are not the same. The method of deciding about whether a signal received contains a watermark or not is ‘detection’. But, identification is the actual process of decoding the unique watermark from a finite number of possible choices. Two kinds of errors could occur at the detection. The first one is detection of a watermark in the received signal when it really does not exist. The second one is quite opposite to this. Herein, the watermark, which inherently exists in the received data, is not detected. Synchronization failure between watermark embedding and detection process could disable the entire communication process. This could result in attacks on the signal.

Some of the additional applications of digital watermarking are broadcast monitoring, usage of contractual watermarks for special purposes, secret communication and copy control.

Broadcast monitoring: Watermarks are used in broadcast monitoring by inserting watermarks in the message that is to be transmitted. This is done before signal (audio or video) is broadcasted. Contractual watermarks: These types of watermarks are also called fingerprints. One of the applications of these is in the distribution of daily film copies. Many people and organizations might be involved in the distribution of a film. In this process, there could be possibilities of leakage of certain copies. Placing a watermark in every copy, which contains the information of the recipient, can prevent this. This doesn’t affect the actual version of the copy, and it is possible to identify the person or source responsible for leakage, if any. Secret communication: This was one of the earliest applications of watermarking technique. It is a process of transmitted secret messages in concealed form. Specially designed programs are there to accomplish these tasks. Copy control: Many watermarks do not actually prevent the copying of content into which they are embedded, but serve as identification marks of the genuine owner of it. But using certain mechanisms that detect the watermarks and inhibit the copying of the information, it is possible to preserve the integrity of the information.

another. Consequently, the signal is made to withstand attacks, which may differ from one application to another. Where there is only a necessity to know whether any distortion of signal has occurred and nothing else, the strength is not at all a constraint. Instead, the watermark may be a fragile one. The situation may be quite different in some other cases. One such case is where the attacks are quite capricious. Herein, the signal is watermarked in such a fashion it can withstand every possible distortion. Tamper resistance: If there is a possibility of aggressive or antagonistic attacks, then the watermarked signal must be tamper resistant. Tamper resistance is immaterial if there are slim chances of aggressive attacks. The watermarked signal must be tamper resistant to attacks like passive, active, and falsification attacks. In passive form of attacks, the attacker just likes to determine whether a watermark is present in the signal or not. In contrast to this, an active attack is one in which the attacker tries to remove the watermark from the watermarked signal. There is yet another form of active attack, called collusion attack wherein, the attacker tries to reconstruct the replica with no watermark from different sets of copies with different watermarks. In falsification attacks, the attacker inserts another watermark into the genuine one, thereby, changing the identification of the copy. Reliability: A watermarked signal may be considered to be highly reliable, if the distortion caused by the inclusion of watermark is quite unnoticeable by the human eye. Cost involved: Depending upon the application, the number of watermark embedders and watermark detectors vary. Their working speeds vary as well according to the type of application. The cost involved thus varies greatly depending up on all these constraints.

Some of the properties of the watermarking techniques are strength, fidelity, cost involved, and the tamper resistance. Strength: Strength is an important parameter because it decides the capacity of the watermarked signal to withstand the wide range of attacks. Especially, the image and video signals need to tolerate the attacks including the geometric ones. Besides, it should be able to endure ordinary attacks like A to D conversion and lossy conformations. More often than not, a watermarked signal is expected to endure common processing transformations only during the time interval between the insertion and decoding of the signal. The toughness of the signal is not limited to any one constraint, but it can be extended to many others. It means, a signal, which is capable of withstanding an attack, may not be able to tolerate

Digital watermarking is an outstanding technology that can serve a wide range of applications in various media types. But, the employment of schemes depends upon the requirements that may vary widely. A sound knowledge of communication techniques and signal processing methods is essential for efficient development of watermarking techniques. Besides this, the attacks that the signal is expected to endure the assessment of the amount of distortion

that the signal suffered which in turn determine the efficacy of the system and the effective cost involved need to be taken into consideration while designing watermarking technique for a particular application. Thus the standards for all the applications don’t remain alike.

1. Digital communications, By Bernard Sklar. 2. Digital image processing By Ganzaly and Woods.

3. JPEG based image compression for low bit rate coding By P.P.Gandhi. 4.

Sign up to vote on this title
UsefulNot useful