Study Of Multimedia Watermarking In Different Domains Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Abstract: Many years are the witness of fast development of video technology. To develop a system with copyright protection and authentication techniques are most interesting field in video technology. In this paper we achieved a survey on different present watermarking schemes. Along with this we studied the watermarking techniques gathering the application specific criteria for MPEG1, MPEG2, and MPEG 4. Later we accomplished the analysis based on robustness of different watermarking algorithms in multimedia systems.

Key Terms: Video Watermarking, MPEG 4.

Introduction: Internet and digital video broadcast networks made the development of multimedia, communication network and distribution of the digital video or images very easy. The convenience of distribution makes the channels ideal for pirating and illegal use of digital content. Piracies can be replicated easily and illegally by Digital Video Recorders (DVR), which are able to make exact digital recordings of digital TV broadcasts. The Digital watermark can cope with this kind of piracies. In the watermarking solution, an invisible and robust watermark are embedded into the image or video. Watermark is a kind of digital data inserted in multimedia substances so that water mark can be identified and it can be claimed also. In this way an authorized copies can be traced. Usually watermark includes information about the possession, foundation, intention, copy control, transaction etc. Possible applications of digital watermarking are transaction tracing, copyright protection, authentication, hidden annotations, improvement in legacy system and data connectivity [60].

Video based applications like video conferencing, wireless videos, video recorders, internet multimedia, videophones require much bigger compression to achieve good data transfer and most excellent video quality. Various video encoders and decoders are grown in meeting the present need of video based applications merchandise. MPEGS use different conversion and block sizes. This leaded the development of novel algorithms robust watermarking techniques.

In the section II we estimated the fundamentals of digital watermarking, terms and techniques in video watermarking. In section III we shortly talked about the common video watermarking techniques. Comparison of these techniques against different attacks is done in section IV. Finally we are concluding this paperwith section V.

Fundamental Concepts about for Video Watermarking

Watermark insertion or watermark embedding symbolizes the information insertion method into multimedia. This is known as original media e.g. text, image, audio or video. The inserted information or watermark could be an identification serial number, unique expression of ownership, copyright messages, signals, dates of transactions or information about inventors The embedded information or watermark can be a serial number or random number sequence, ownership identifiers, copyright messages, control signals, transaction dates, information about the inventors of the work etc. Various watermarking algorithms can be found in the text [9]-[10], image [11]-[12], audio [13] and video [14]-[15]. Original media is transformed to generate the watermarked media using these algorithms.. Fig.1 shows an overview of different types of watermarking methodologies depending on their working domains, cover media, perceptibility and application areas. After the watermark insertion, the media to be watermarked launched over Internet or some other transmission channels. The embedded information about the inventor of media is decoded to identify copyright owner each time there is rise of question of the copyright of the digital media. The procedure of decoding can pull out the watermark from the watermarked media (watermark extraction) or can distinguish the existed watermark inside it (watermark detection).

Fig.1 Different Types Of Methods For watermarking

Video watermarking terms:

Video watermarking expresses the method of embedding information in video data [2]. Some of the important terms referencing to digital video watermarking are as follows:

Digital Video: The collection of successive and evenly time spaced motionless images is known as digital video.

Payload: The amount of information accumulated in a watermark is a payload. Watermark Granularity is the vital concept related to the video watermarking payload. It is defined as the amount of data required for embedding one unit of watermark information.

Perceptibility: When human cannot identify the original video form watermarked video then it is known as imperceptible methodology for video watermarking.

Robustness: Easy breakable watermark must not be strong against intended modification techniques because there is a failure in detecting the watermark; it leads to non-authentication of received data. In case of copyright protection, it is enviable that watermark remains in the video data all time. This is happened even if the video data is focused to deliberate and unintended signal processing attacks. So watermark is made robust or semi fragile manner depending on requirement

Security: The encryption methodology and security of the watermarking algorithm are guaranteed in the same way as in the Kerckhoff's assumption.

Techniques for video Watermarking:

The different watermark techniques in multimedia presently found are give below.

The video watermarking techniques faces more challenges than the image watermarking systems. These might be redundancy in frames, disturbance between motion and still parts, realtime broadcast of video. Video series always comes under the attacks such as lossy compressions, digital to analog conversions, frame exchange, overlapping are the pirate attacks. Video watermarking methods faces the issues like localized recognition, real-time hanging point representation, power dissipation etc [77].

There are some other watermarking schemes in the working domains such as watermark in pixel domain and in transform domain. The vital benefit of the pixel domain is that these are easy to understand and hence real-time implementable.

In case of transform domain watermarking schemes, the signal from host is converted into different domains. The methods in transform domain are DCT (Discrete Cosine Transformation) and DWT (Discrete Wavelet Transformation). As an example in the frequency domain, there is guaranty of robustness and imperceptibility criteria for applying more properties for the systems.

Video Watermarking Survey :

Watermark is embedded different manners such as directly in the unprocessed video or combined during the encoding process or executed after the video compression. Here we will discuss about some general video watermarking techniques.

Efficient Video Watermarking using Motion Estimation Approach in spatial domain [17] is a method for inserting additive digital watermarks into uncompressed and compressed video sequences in a spatial domain. The scheme is implemented to develop the motion regions between frames using motion estimation method. It is tested on different types of video (compressed DVD quality movie and uncompressed digital camera movie). The scheme is supposed to be robust against the attack of frame dropping, averaging, statistical analysis and lossy compression. The point of view of the author is making use of a digital watermarking technique to protect the digital multimedia intellectual copyright, and proposing a new algorithm of color video watermarking applied in the spatial domain. This algorithm uses the predictive motion estimation [1] to find the best matched blocks in order to embed the proposed watermark (random Gaussian distribution) in between frames.

MOTS is the method used for envisaging blocks motion vectors (motion estimation) is transformed version of the OTS standard method. This OTS method is very easy but effective algorithm. It makes separate horizontal and vertical scan. The point of minimum distortion over the horizontal axis is search throughout the horizontal scan. As it is found the scan is stopped. The process of modification is found using the existed motion vectors of the adjacent blocks inter-block correlation. The preset of motion vectors of the previous blocks is used in (i.e., top, left, and top block to calculate the initial motion vector of the tested block). Advantage of MOTS over OTS is that MOTS determines the motion vectors with minimum prediction error compared to OTS [1]. The flowchart of the MOTS method is shown fig. 2.

Fig 2. MOTS method Flowchart

The proposed scheme achieves the watermarking algorithm in the spatial domain. The intensities or color values of some selected pixels are modified by spatial domain. The illustrated quality of the watermarked data is needed to be as higher as possible because the measure of distortions introduced by watermarking process. The degradation of the data due to the watermarking operation is imperceptible which is known as visual quality. PSNR is a visual quality measurement known as peak signal to noise ratio. The watermark is extracted by using original frame, the watermark W* is extracted as shown below.


I: the original frame data

Iw: the watermarked frame dat

W: the watermarked data

Α: the scaling factor

x, y: o,……….m-1 where m is block size

Practical Real-Time Video Watermarking Scheme [19] was proposed for inserting the watermark into an arbitrary size of a host video. The extraction of the watermark is under a preset false error rate. The watermark embedding as well as extraction is made in the uncompressed domain. The expressed watermark scheme is relevant for DCT-based coding videos (e.g., MPEG-2) as well as MPEG-1 and MPEG-4. As per the format of the input video the watermarking system uses a video coder as well as decoders. To use in practical, these embedder and detector are developed as a directshow filters to achieve easy connectivity in the directshow applications. Code book generation and the decision of block size pattern are the two basic steps in watermark insertion. The size of the pattern is depend on the no. of messages inserted into one decoded frame and no. of encoded pattern repeated within it.

Visibility is reduced by using Dithering. (see Fig. 3(c)). There are different sizes for the pre-defined matrices. So the dithered value gives a smooth region of 2-D scaled watermark while value of a detail region is taken using small size. The scaling factor α is calculated by pixel-by-pixel estimation. Spatially perceptual masking is adapted to the original frame and then local is calculated. Finally, the watermarked frames Yi are obtained by adding the dithered version of the watermarks to host video frames Xi. For boosting detector performance the same watermark is inserted into a fixed number of t consecutive frames. Since it should need high computational costs to estimate α and add the watermark to the host frames. MMX technology is utilized to accomplish real-time insertion process [6]. Figure 3 illustrates this watermark embedding scheme.

Fig.3. Block diagram of practical real-time video watermarking embedding scheme: (a) overview of watermark embedding into video streams, (b) the details of the watermark embedder, and (c) used 2-2, 3-3, and 4-4 matrices as dithering.

A novel video watermarking scheme based on motion estimation for color video sequence in a frequency domain is a frequency domain watermarking scheme [27]. The filter used to apply wavelet transform is a non reversible biorthogonal transform. This transform can only be used for lossy coding. The FWT is applied on a frame by transforming the rows and then the columns of the frame. It yields two dimensional decomposition (four-channel decomposition). The wavelet transform compacts most of the image energy into the LL sub band which consists of a few coefficients in comparison with the whole number of wavelet coefficients (which is equal to the total number of pixels). Unlike conventional transforms, wavelet decomposition produces a family of hierarchically organized decompositions. The selection of a suitable level of the hierarchy depends on the signal nature and experience. In this work, the wavelet decomposition is two level. For the second level of wavelet transform, since the watermark size is (1024 bytes), there is a need of 1024/ (m/2+m/2) blocks to embed the watermark data in each frame, where m*m is the block size. For example if m = 8 (block size is 8*8) then 128 blocks are needed to embed the watermark. The insertion and extracting processes of the proposed watermark in the frequency domain (added only to HL and LH sub bands) is given in fig. 4. The main advantage of using the frequency domain in this video watermarking scheme is that it is resistant against some attacks. This evaluation is done through the similarity measure of a watermark before and after attack.

Fig4. Watermark Embedding and Extraction process in frequency domain

IV Comparison Of the watermarking schemes:

In this section, we made the comparative analysis of the above mentioned watermarking schemes. We compared these watermarking schemes against robustness, reliability, practicality and impreciability among each other. This comparison is indicated by the quantifiers like Good, Acceptable and Poor. This is shown in the following table 1.

Watermarking Techniques










Practical Real Time





Robust video watermarking





Table 1. Comparison of watermarking schemes

The following tables 2, 3,4 shows the representation of these schemes depending on the robustness against different kinds of attacks. The table 2 shows the test results conducted for MOTS watermarking scheme. These results are obtained after conduction of some tests while frame dropping, compression.

Table2 Robustness of MOTS watermarking schemes against attacks

The practical real time watermarking scheme is robust against various attacks such as cropping, format conversion, frame rate change, color-to-gray conversion attack as shown in Table 3. We found that this schemes not only good at robustness against downscaling attack and other malicious attacks but also it allows to be confident for the specific error rate .

Input Video



Frame Rate change

Color conversion


























Table 3. Summary of robustness against attacks

The Table 4 presents the comparison between the similarity of robustness against three different types of attacks in frequency domain as well as spatial domain tested on uncompressed video sequence for two successive frames.

V. Conclusion:

The Efficient video watermarking technique is based on motion estimation for color video sequence. This technique is tested on compressed (taken from DVD high quality film) and uncompressed (taken by digital camera) video movies. A new principle is used during the motion estimation process to select the best matched motion blocks near the center of the frame which is candidate to add a generated watermark. In the Practical Real-Time Video Watermarking scheme, focus is on the robustness to downscaling attack. The reason for this is more portable devices are spread out and more downscaled and distorted video data are in public. In addition to this attack, the proposed scheme is also robust against not only spatial attacks such as cropping, format conversion, and color-to- gray conversion but also temporal attacks such as frame rate change. Both watermark insertion and the extraction are done in uncompressed domain, so the proposed scheme can be straightforwardly applied to all video coding if there are corresponding video coder and decoder The novel video watermarking scheme based on motion estimation for color video sequence in a frequency domain is also tested on compressed (taken from DVD high quality film) and uncompressed (taken by digital camera) video movies. The watermark is the random Gaussian distribution which is embedded into the motion regions between frames (HL, LH bands). This proposed scheme has a higher degree of invisibility against the attack of frame dropping, adaptive quantization, and frame filtering than other schemes in spatial domain.