Image Watermarking
Feng CHEN
02430339
159.731 Machine Vision Seminar Report
June 2004
Table of Contents
1. Introduction 3
2. Motivation 4
3. Approaches of Watermarking 5
4. Multi-resolution watermarking 6
5. Applications of watermarking 10
6. Requirements of watermarking 11
7. Conclusions 12
8. References 13
1. Introduction
The rapid expansion of the Internet in the past years has rapidly increased the availability of digital data such as audio, images and videos to the public. As we have witnessed in the past few months, the problem of protecting multimedia information becomes more and more important and a lot of copyright owners are concerned about protecting any illegal duplication of their data or work. Some serious work needs to be done in order to maintain the availability of multimedia information but, in the meantime, the industry must come up with ways to protect intellectual property of creators, distributors or simple owners of such data. This is an interesting challenge and this is probably why so much attention has been drawn toward the development of digital images protection schemes. Of the many approaches possible to protect visual data, digital watermarking is probably the one that has received most interest The idea of robust watermarking of images is to embed information data within the image with an insensible form for human visual system but in a way that protects from attacks such as common image processing operations.
As the computers are more and more integrated via the network, the distribution of digital media is becoming faster, easier, and requiring less effort to make exact copies. One of the major impediments is the lack of effective intellectual property protection of digital media to discourage unauthorized copying and distribution.
1.1 What is a watermark?
A watermark is the hidden information within a digital signal (such as image, video, audio, polygonal model...).It is integrated into the content of host signal itself, and requires no additional file header or conversion of data format as well. Moreover, it is designed to permanently reside in the host data .Finally, unlike encryption; it does not restrict access to the host data.
1.2 What is a digital watermarking?
Also referred to as simply watermarking, a pattern of bits inserted into a digital image, audio or video file that identifies the file's copyright information (author, rights, etc.). The name comes from the faintly visible watermarks imprinted on stationery that identify the manufacturer of the stationery.
The purpose of digital watermarks is to provide copyright protection for intellectual property that's in digital format.
Unlike printed watermarks, which are intended to be somewhat visible, digital watermarks are designed to be completely invisible, or in the case of audio clips, inaudible. Moreover, the actual bits representing the watermark must be scattered throughout the file in such a way that they cannot be identified and manipulated. And finally, the digital watermark must be robust enough so that it can withstand normal changes to the file, such as reductions from lossy compression algorithms.
Satisfying all these requirements is no easy feat, but there are a number of companies offering competing technologies. All of them work by making the watermark appear as noise - that is, random data that exists in most digital files anyway. To view a watermark, you need a special program that knows how to extract the watermark data.
Watermarking is also called data embedding and information hiding.
2. Motivation
2.1 In the past
Conventionally, in analog world, a painting is signed by the artist to attest the copyright, an identity card is stamped by the steel seal to avoid forgery, and the paper money are identified by the embossed portrait. Such kind of hand-written signatures, seals and watermarks have been used from ancient times as a way to identify the source, creator of a document or a picture. For example, a priceless painting of the 11th century in National Palace Museum named "Travelers on a Mountain Path" had not been identified as the genuine work of Fan Kuan until Fan's signature is found between the woods behind a group of travelers of the painting.
Travelers on a Mountain Figure 1 Fan’s signature “范宽”
2.2 In digital World
However, in the digital world, digital technology for manipulating images has make it difficult to distinguish the visual truth. Besides, the characteristics of digitization bring significant hangs in copyright issues, which create an urgent need to intellectual property protection on the digitally recorded information.
Digital watermarking has been proposed as a way to claim the ownership of the source and owner. Unlike encryption, watermarking does not restrict access to the data. Once the encrypted data is decrypted, the intellectual property rights are no longer protected.
Over the past few years, the technology of the digital watermarking has gained prominence and emerged as a leading candidate that could solve the fundamental problems of legal ownership and content authentications for digital multimedia data. A great deal of research efforts has been focused on digital image watermarking in recent years. The techniques proposed so far can be divided into two groups according to the embedding domain.
One group is spatial domain approach.
The other group is frequency domain approach.
3. Approaches of watermarking
3.1 Spatial Domain Approach
The earliest watermarking techniques are mainly this kind and the simplest example is to embed the watermark into least significant bits (LSBs) of the image pixels. However, this technique has relatively low information hiding capacity and can be easily erased by lossy image compression.
3.2 Frequency Domain Approach
Another way to produce high quality watermarked image is by first transforming the original image into the frequency domain by the use of Fourier, Discrete Cosine or Wavelet transforms for example. And it can embed more information bits and is relatively robust to attack. With this technique, the marks are not added to the intensities of the image but to the values of its transform coefficients. Then inverse-transforming the marked coefficient forms the watermarked image. The use of frequency based transforms allows the direct understanding of the content of the image; therefore, characteristics of the human visual system (HVS) can be taken into account more easily when it is time to decide the intensity and position of the watermarks to be applied to a given image.
3.2.1 Examples of Implementation
1. Cox et al. used the spread spectrum communication for digital multimedia watermark.
2. Hsu and Wu embedded an image watermark into selectively modified middle frequency of discrete cosine transform (DCT) coefficients of container image.
3. Joseph et al. developed a digital image watermarking using the Fourier-Mellin transform that is invariant to image manipulations or attacks due to rotation, scaling and translation.
Several other methods used discrete wavelet transform to hide the data to the frequency domain.
4. Multi-resolution Watermarking
4.1 Description
Here, I am going to discuss an example on watermarking method proposed by Ward .This first generation technique called Multi-resolution Watermark for Digital Images works in the wavelet domain.
The modus operandi is pretty simple since it consists of adding weighted pseudo-random codes to the large coefficients at the high and medium frequency bands of the discrete wavelet transform of an image. Even if it might first appear theoretically straightforward the actual implementation is a little more complicated than it seems and great effort will be made here to clarify it.
First of all, one must know why the wavelets approach is used here. The characteristics of this transform domain are well suited for masking consideration since it is well localized both in time and frequency. Besides wavelet transforms match the multi channel model of the HVS so one can actually set a numerical limit to the wavelets coefficients alteration in order to stay under the HVS just noticeable difference 3 (JND), for which our eyes start to become aware of the modifications in the image . Besides, wavelet transforms is a part of upcoming compression standards (such as JPEG-2000) so wavelet-based techniques would allow a much easier and optimized way to include a copyright protection device in the compression code itself.
A second important aspect of the technique used here is the way to introduce a watermark in an image. A watermark should be introduced in perceptually significant regions of the data in order to remain robust. However, by doing so, we risk to alter the image (i.e. perceivably). The technique described here follows to a certain degree this requirement but tries to make the introduced watermark as invisible as it can while showing good robustness. As the reader will understand, we have chosen to ensure transparency of the watermark and, in the same time keep the robustness by embedding the information within only the high and medium frequencies while keeping third resolution low frequencies, for which the HVS is most sensitive, untouched. The implemented technique uses no oblivious watermark. In copyright protection, it is reasonable to assume that the owner or the user who wants to verify the presence of a particular watermarking key has access to the original un-watermarked image. This type of watermarking scheme is called non-oblivious (or private) watermarking in opposition to oblivious (or public) techniques, such as copy protection, where the original image is not accessible.
4.2 Steps of Implementation
4.2.1 Encoder
The first part of the watermarking process is, of course, the encoder. The first step is to decompose the image into ten frequency bands using three resolutions of Haar wavelets.
Figure 2’s bank of filters represents well the idea of the octave-band structure of Haar wavelets, which gives pyramid structure frequency localization as shown in Figure 3.
Here, I must point out that an under sampling operation is done after every filtering. It must be understood that the choice of the Haar wavelet in our system was one made for simplicity. However, we had in mind to investigate the influence of the choice of the wavelet function in our results but, in order to test the robustness truthfully, we had to give up the idea in favor of the addition of extra robustness testing procedures. The next operation is to add a pseudo random sequence N , in fact a Gaussian distribution of mean zero and variance one, to the coefficients of the medium and high frequency bands (i.e. all the bands except the lowest one which is represented by the top left corner in Figure 3). The normal distribution is used because it has been proven to be quite robust to collusive attacks. In order to weight the watermark according to the magnitude of the wavelet coefficients, we used one of the two following relations between the original coefficients y and ÿ , the ones containing the watermark:
ÿ[m,n] = y[m,n] + alpha (y[m,n])2 N[m,n] ( 1 )
ÿ[m,n] = y[m,n] + alpha . abs(y[m,n]) N[m,n] ( 2 )
It must be pointed out that the relations (1) and (2), even though they are mathematically different, have the exact same goal which is to put more weight to the watermark added to high value wavelet coefficients. The parameter alpha is to control
the level of the watermark; it is in fact a good way to choose between good transparency or good robustness or a tradeoff between the two. Finally, the two-dimension inverse wavelet transform of is computed to form ÿ the watermarked image.
Figure 4 gives a good idea of the main components of the encoder
4.2.2 Decoder
At the other end of the communication channel, a decoder is used to extract the watermarked information from the received image. Upon reception of the supposedly watermarked image, the algorithm first isolates the signature included in this image by comparing the DWT coefficients of the image with those of the original (non-watermarked) one. The following operation consists of taking the identified key to put in contrast with the found signature by computing the cross-correlation at the first resolution level (i.e. highest frequency coefficients). The watermark is called detected if there is a peak in the cross-correlation corresponding to a positive identification. If there is no central peak, the decoder adds the second resolution level (i.e. the bottom left square in the pyramid structure of Figure 3) to the computation aiming at finding for a peak. Once again, if there is a peak, the watermark is called detected and if not, we go to the third resolution… and so on until we reach the ninth resolution limit.
4.3 Advantage of this technique
The main advantage of this technique is that while allowing good detection, even in the presence of corruption, it keeps the level of false positive detection to a minimum since the found signature has to go through detection step of positive identification to be called detected. The detector step aims at ensuring the maximum exactitude in the detection of the owner identification key and, as said previously, minimizing the number of false positive detection. The results presented later on should convince the reader of the performance of our decoder.
5. Applications of Watermarking
5.1 Fingerprinting
In order to trace the source of illegal copies the owner can embed different watermarking keys in the copies that are supplied to different customer. For the owner, embedding a unique serial number-like watermark is a good way to detect customers who break their license agreement by copying the protected data and supplying it to a third party.
5.2 Indexing
Watermarking offers a wide range of new capabilities to multimedia applications. It allows the indexing of video mail by permitting the insertion of comments in video content as well as the indexing of movies or news items by making available the utilization of markers that can be exploited in search engines. As the number of images and video contents online increases a lot faster than the capabilities of today’s search engine, it is important to plan ahead for new ways to allow quick access to multimedia data and watermarking is certainly a promising way to do so.
5.3 Copyright Protection & Owner identification:
To protect its intellectual property, the data owner can embed a watermark representing copyright information of his data. This application can be a really helpful tool in settling copyright disputes in court. It is probably the most widely spread use of digital images watermarking and it is also the application we have worked on in the present project.