Report on Digital Watermarking Technology

1

CHAPTER 1

INTRODUCTION

The advent of the Internet has resulted in many new opportunities for the creation and

delivery of content in digital form. Applications include electronic advertising, real time

video and audio delivery, digital repositories and libraries, and Web publishing. An important

issue that arises in these applications is the protection of the rights of all participants. It has

been recognized for quite some time that current copyright laws are inadequate for dealing

with digital data. This has led to an interest towards developing new copy deterrence and

protection mechanisms. One such effort that has been attracting increasing interest is based

on digital watermarking techniques.

Digital watermarking is the process of embedding information into digital multimedia

content such that the information (which we call the watermark) can later be extracted or

detected for a variety of purposes including copy prevention and control. Digital

watermarking has become an active and important area of research, and development and

commercialization of watermarking techniques is being deemed essential to help address

some of the challenges faced by the rapid proliferation of digital content.

The recent growth of networked multimedia system has increased the need for the protection

of digital media. This is particularly important for the protection and enforcement of

intellectual property rights. Digital media includes text, digital audio, images, video and

software. Many approaches are available for protecting digital data; these include encryption,

authentication and time stamping.

One way to improve one's claim of ownership over an image, for instance, is to place a low-

level signal directly into the image data. This signal, known as a digital watermark, uniquely

identifies the owner and can be easily extracted from the image.

2

CHAPTER 2

What is watermark?

A distinguishing mark impressed on paper during manufacture; visible when paper is held up

to the light (e.g. $ Bill).

Fig 1

What is digital watermarking?-

In generally digital watermarking is a technique for inserting information into an image. It is

an evolving field that requires continuous effort to find the best possible method in protecting

multimedia content.

It is a process of embedding information into digital multimedia content such that the

information can later be extracted or detected for variety of purposes including identification

and authentication.

Fig 2 Block diagram for watermarking digital image

3

CHAPTER 3

Architecture of digital watermarking

System model of digital watermarking

The process of digital watermarking embeds the special information which stands for the

particular identity of the owner of the copyright by some sort of algorithm to multimedia

data. We can extract the watermark, verify the ownership of the copyright and ensure the

legitimate rights of the copyright owners though the appropriate algorithms.

A complete digital watermarking system is composed of two basic modules: watermark

embedding module and watermark detection and extraction module. Watermark embedding

module is responsible for adding the watermark signal to the original data.

The watermark can be any form of data, such as numeric, text, image, and so on. Key can be

used to strengthen security to prevent unauthorized parties restore and modify the watermark.

The watermark embedding module is as figure 3:

Fig 3 Watermark Embedded module

Watermark detection and extraction module is used to determine whether the data contains

specified watermark or the watermark can be extracted. The module input may be image,

key, watermark or original image, the output is a watermark or some kind of credibility value.

4

It indicates the possibility of the data having a given watermark. The watermark embedding

module is as figure 4.

Fig 4 Detection and Extraction module of watermark

Main algorithms of digital watermarking

In recent years, the study of digital watermarking technology makes great progress. There are

a lot of good algorithms which can be divided into spatial domain algorithm and transform

domain algorithm.

1) Spatial domain algorithms

Spatial domain digital watermarking algorithms directly load the raw data into the original

image.

a) Last significant bit algorithm

The algorithm embeds the information with the form of the least significant bits selected

randomly which can ensure the embedded watermark is invisible. But the algorithm has poor

robustness, and watermark information can easily be destroyed by filtering, image

quantization, and geometric distortion.

b) Patchwork algorithm

Based on the statistics, the algorithm uses the statistical characteristics of pixels to embed the

information into the brightness values of pixel. It can resist lossy compression coding and

malicious attacks. However, the amount of embedded information is limited, in order to

5

embed more watermark information; we can segment the image, and then implement the

embedding operation each image block.

c) Texture mapping coding method

It hides the watermark in the texture part of the original image. The algorithm has strong

resistance ability to attacks for a variety of deformation, but only suitable for areas with a

large number of arbitrary texture images, and cannot be done automatically.

2) Transform domain digital watermarking algorithm

Transform domain algorithm is a method of hiding data similar to spread-spectrum

communication technology. Firstly, it does a kind of orthogonal transformation for image,

and then embed watermark information in the transform domain of image, finally use the

inverse transform to recovery the image in spatial domain, the detection an extraction of the

watermark are also realized in transform domain. There are several common used transform

domain methods, such as discrete fourier transform (DFT), discrete cosine transform (DCT),

discrete wavelet transform (DWT), and so on. As a classical mathematical transformation

method, DCT does a very important role in image compressing, coding and other

applications. The watermarking algorithms based on DCT domain are compatible with the

existing international compression standards. The main idea of these methods is to select

middle or low frequency coefficients to superposition watermark in the DCT transform

domain.

6

CHAPTER 4

Types of watermarking techniques

Watermarks and watermarking techniques can be divided into various categories in various

ways. The watermarks can be applied in spatial domain. An alternative to spatial domain

watermarking is frequency domain watermarking. It has been pointed out that the frequency

domain methods are more robust than the spatial domain techniques. Different types of

watermarks are shown in the figure:

Fig 5

Watermarking techniques can be divided into four categories:

1. According to the type of document to

o Image Watermarking

o Video Watermarking

7

o Audio Watermarking

o Text Watermarking

2. According to the human perception

o Visible watermark

o Invisible-Robust watermark

o Invisible-Fragile watermark

o Dual watermark Visible watermark is a secondary translucent overlaid into the primary

image. The watermark appears visible to a casual viewer on a careful inspection. The

invisible-robust watermark is embedded in such a way that alternations made to the pixel

value are perceptually not noticed and it can be recovered only with appropriate decoding

mechanism.

The invisible-fragile watermark is embedded in such a way that any manipulation or

modification of the image would alter or destroy the watermark.

Dual watermark is a combination of a visible and an invisible watermark. In this type of

water mark an invisible watermark is used as a backup for the visible watermark as clear

from the following diagram.

Fig-6 Schematic representation of dual watermarking

3. From application point of view

Source based

8

Source-based watermark are desirable for ownership identification or authentication where a

unique watermark identifying the owner is introduced to all the copies of a particular image

being distributed. A source-based watermark could be used for authentication and to

determine whether a received image or other electronic data has been tampered with.

The watermark could also be destination based where each distributed copy gets a unique

watermark identifying the particular buyer. The destination –based watermark could be used

to trace the buyer in the case of illegal reselling.

4. According to working domain

Spatial watermarking can also be applied using color separation. In this way, the watermark

appears in only one of the color bands. This renders the watermark visibly subtle such that it

is difficult to detect under regular viewing. However, the mark appears immediately when the

colors are separated for printing. This renders the document useless for the printer unless the

watermark can be removed from the color band. This approach is used commercially for

journalists to inspect digital pictures from a photo-stock house before buying unmarked

versions.

Watermarking can be applied in the frequency domain (and other transform domains) by first

applying a transform like the Fast Fourier Transform (FFT). In a similar manner to spatial

domain watermarking, the values of chosen frequencies can be altered from the original.

Since high frequencies will be lost by compression or scaling, the watermark signal is applied

to lower frequencies, or better yet, applied adaptively to frequencies that contain important

information of the original picture. Since watermarks applied to the frequency domain will be

dispersed over the entirety of the spatial image upon inverse transformation, this method is

not as susceptible to defeat by cropping as the spatial technique. However, there is more a

tradeoff here between invisibility and decidability, since the watermark is in effect applied

indiscriminately across the spatial image. Table shows a small comparison between the two

different techniques.

9

Fig 7 Comparison between watermarking technique

10

CHAPTER 5

Basic characteristic of digital watermarking

The basic requirement of digital watermarking is closely related to its purpose of

applications, different application has different demand. In general, the characteristics of

digital watermarking are as follows.

Robustness refers to that the watermark embedded in data has the ability of surviving after a

variety of processing operations and attacks. Then, the watermark must be robust for general

signal processing operation, geometric transformation and malicious attack. The watermark

for copyright protection does need strongest robustness and can resist malicious attacks,

while fragile watermarking; annotation watermarking do not need resist malicious attacks.

-perceptibility

Watermark cannot be seen by human eye or not be heard by human ear, only be detected

through special processing or dedicated circuits.

Watermark should be able to provide full and reliable evidence for the ownership of

copyright-protected information products. It can be used to determine whether the object is to

be protected and monitor the spread of the data being protected, identify the authenticity, and

control illegal copying.

Watermark information owns the unique correct sign to identify, only the authorized users

can legally detect, extract and even modify the watermark, and thus be able to achieve the

purpose of copyright protection.

11

CHAPTER 6

Watermark Attacks

-- Hacker attempts to remove or destroy the watermark.

• Hacker tries to find if a watermark is present.

• Removal of watermark is not an aim.

detector can accept modified media.

12

CHAPTER 7

Digital image processing

Digital images are represented using matrices:

Types of Watermarking

Wavelet Transform Domain (DWT)

13

Spatial domain Watermarking

The watermark message is added in the spatial Domain:

IM (x, y) = I (x, y) + kM(x, y)

14

Spatial Domain Watermark Embedding Scheme

One approach is to embed watermarks in the spatial domain of an image

The idea is that the number of bits you wish to embed, the watermark size, is stored in the

pixels of the image to be marked. Let us for example want to embed 40 bits of watermark

information in a host image. Then we supply a seed to a PRNG (Pseudo Random Number

Generator) and use it to provide us with 40 sets of (x,y) with respect to the illustration, in

other words we select 40 pixels.

Each of these pixels are now altered, exchanging a chosen bit of color information with our

watermark bit. The choice of bit involves a trade-off:

● If we choose one of the lesser significant bits, our alteration can be held invisible in the

marked picture, but image compression and other things might be attacking these less

significant bits, which could mean that the watermark is not very robust.

● If a more significant bit is chosen, we would achieve much better robustness, but our

alteration would be visible. It would look weird if the sunset picture had 40 black spots in it.

When you would attempt to extract the watermark at a given time, the PRNG would then be

started with the same seed, and again produce the 40 pixels, from where we can extract our

bits.

This method is easy to use, but has many drawbacks. Even small alterations to our image

would destroy at least some of our watermark information.

15

Viterbi Decoding:

It estimate v1 a sequence v that maximizes P(r/ v). Where r is sequence, Probability p and the

starting and ending state is predetermined to be the zero-state.

Why Viterbi Decoding:

•A highly satisfactory bit error performance

•High speed of operation

•Ease of implementation

•Low cost.

•Fixed decoding time.

Quad Tree Region Splitting Image segmentation Method:

Region Based Image Segmentation

Let R represent the entire image region, then we may view region based segmentation as a

process that partitions R into n sub regions, R1,R2……..Rn, such that

(a) U Ri=R.

i=1

(b) Ri is a connected region, i=1,2………n. i=1

(c)Ri ∩Rj=Φ for all i and j, i≠j.

(d) P (Ri) =TRUE for i=1,2….n.

Quad Tree Approach:

16

Fig 8 Quad tree

Advantage of Quad Tree Decomposition

Small regions represent the presence of critical information of the image and hence are the

good place for the watermark insertion.

Watermark insertion:

Fig 9 watermark insertion

17

Watermark Insertion Algorithm:

Apply QUAD TREE decomposition on color image I (x, y) and select all 4x4 blocks in blue

channels.

Repeat (for each selected 4x4 block (H) of blue channel)

Step 1: Compute the average, Imean, minimum, Imin, and maximum, Imax, of the pixels in

H.

Step 2: Classify each pixel into one of two categories, based on whether its intensity value is

above or below the mean intensity of the block, i.e., the ijth pixel, bitij is classified depending

on its intensity, I, as

bitij ∈ YH if I >Imean

bitij ∈ YL if I ≤ Imean

where YH and YL are the high and low intensity classes, respectively.

Step3: Compute the means, meanL and meanH, for the two classes, YL and YH.

Step 4: Define the contrast value of block H as

CB = max(Cmin, β(Imax-Imin))

where β is a constant and Cmin is a constant which defines the minimal value a pixel's

Intensity can be modified.

Step5: Select a watermark bit (bitw ) randomly depending on the key value.

Step 6: Given the value of bitw is 0 or 1, modify the pixels in H according to:

18

if bitw = 1,

I new = Imax + λ if I > meanH

I new =Imean + λ if meanL ≤ I < Imean

I new =I + δ otherwise

if bitw = 0,

I new = Imin - λ if I < meanL

I new = Imean - λ if Imean ≤I < meanH

I new = I - δ otherwise

Where I new is the new intensity value for the pixel which had original intensity value I and δ

is a random value between 0 and CB and λ is the watermark strength.

Step 7: The modified block of pixels, Hnew, is then positioned the watermark image in the

same location as the block, H, of pixels from the original host image. Until all watermark bits

are inserted.

Step 8: Marge red, green and blue channel.

Watermark detection:

Fig 10 Watermark detection

19

Watermark Extraction Algorithm:

Apply QUAD TREE decomposition on original image I (x, y):

Repeat {

Step 1: take one 4x4 blocks of the host image and Corresponding 4x4 block of watermarked

Image using the same coordinate value as of 4x4 block of host image.

Step2: a watermark bit is decoded by making the comparison of the two resultant values:

If Average w > Average o, then bit w = 1

If Average w ≤ Average o, then bit w = 0

Where Average o and Average w are the averages for the 4x4 blocks of the host and

Corresponding 4x4 blocks of watermarked Images, respectively. }

Until all watermark bit are extracted. The decoded bits are then arranged in order using same

key, which was used during embedding. Then, the encoded watermark is exclusive order by

128 bit key and then decoded by viterbi decoding.

Fig 11

20

Transform Domain Watermarking:

Spatial domain methods are less complex as no transform is used, but are not robust against

attacks. Transform domain watermarking techniques are more robust in comparison to spatial

domain methods. This is due to the fact when image is inverse wavelet transformed

watermark is distributed irregularly over the image, making the attacker difficult to read or

modify. Among the transform domain watermarking techniques discrete wavelet transform

(DWT) based watermarking techniques are gaining more popularity because DWT has a

number of advantages over other transform such as progressive and low bit-rate transmission,

quality scalability and region-of-interest (ROI) coding demand more efficient and versatile

image making the attacker difficult to read or modify. Among the transform domain

watermarking techniques discrete wavelet transform (DWT) based watermarking techniques

are gaining more popularity because DWT has a number of advantages over other transform

such as progressive and low bit-rate transmission, quality scalability and region-of-interest

(ROI) coding demand more efficient and versatile image coding that can be exploited for

both, image compression and watermarking applications.

Discrete Fourier Transform (DFT) is superior for shifting attacks

-Shifting in the space domain leads to a phase shift in the frequency domain.

Discrete Cosine Transform (DCT)

DCT does a very important role in image compressing, coding and other applications. The

watermarking algorithms based on DCT domain are compatible with the existing

international compression standards. The main idea of these methods is to select middle or

21

low frequency coefficients to superposition watermark in the DCT transform domain. Image

x(m,n) can be seen as a matrix of M×N and be transformed from the spatial domain DCT.

-based transform (8x8),(16x16), (32x32), …

d in the intermediate frequency coefficients.

e to the intermediate frequency of each block.

Fig 12

The image data is transformed to transform domain data which is also transformed back to

the image data by inverse transform. These transformations are complete by DCT.

22

Fig 13

DCT Domain Watermark Embedding Scheme:

Overview:

In this process, first, a DCT transform operation is performed on each 8x8 pixel block of the

achromatic component. Then, the watermark is embedded into the DCT coefficients, and

after performing the reverse DCT operation to return to the spatial domain, the watermark is

once again redirected to the saturation component of the image. Like in the Spatial Domain

Watermark Embedding Scheme, the original image, f is a color image of size MxN and will

be represented by. The watermark will be a grayscale image of size M/8xN/8. It is

represented by.

Step 1: Find the achromatic component of the host image

Step 2: Break the achromatic image into blocks

Step 3: Convert to DCT Domain

Apply a DCT operation on each block of the achromatic component of the original image.

Once in the DCT domain, the first pixel of each block will be the average value of the other

blocks and the rest of the coefficients are arranged in a zigzag scan starting from the pixel

directly to the right of the first pixel, which will have the lowest frequency, and ending with

the last pixel, which will have the highest frequency.

23

Step 4: Pseudorandom Permutation of the Watermark

Step 5: Embed the Permuted Watermark Embed each watermark pixel into the first 2x2 pixel

block, in each 8x8 pixel block.

Step 6: Convert back to Spatial Domain

Step 7: Redirect the Watermark into the Saturation Component of the Color Image.

DCT Domain Extraction Scheme:

Here both the original image and the watermarked image will be needed to complete the

extraction.

Step 1: Find Achromatic Component of Original Image

Step 2: Reverse the Saturation Adjustment and Find the Achromatic Component of the

Watermarked Image.

Step 3: Break the achromatic components into blocks.

Step 4: Converting to the DCT Domain. Apply the DCT operation on each block of the

achromatic components of both the original image and the watermarked image.

Step 5: Extract the permuted Watermark.

Step 6: Reverse Pseudorandom Permutation.

Fig 14

24

Discrete Wavelet Transform (DWT):

DWT locally separates the content of the image into low frequency and high frequency

subbands. Most of the energy is concentrated in the low frequency subband. In watermarking

the message is inserted in the high frequency subbands (HL,LH and HH). What is achieved

from this process is a composition of an image in terms of frequencies. The frequency is the

change in contrast over pixels. The faster the change and a higher contrast change, gives a

higher frequency. Large surfaces even in contrast have a low frequency. Therefore most

images consist mainly of low frequency information. The high frequency parts are found

around edges and textures, where contrast or color changes rapidly.

Fig 15

Embedding process

1. The image to be marked is decomposed using DWT, to the required resolution level. These

resolution levels are by suggested to be five for most watermarking applications.

2. The hierarchical decomposition is made into a tree representation, where each node has

four children and is associated with a coefficient in the DWT decomposition.

3. A required amount of coordinates is selected from the DWT decomposition. The selection

is made with a PRNG which is given a seed k. No coordinates from resolution level 1 is

selected, since this will distort the image. To prevent disordering siblings of a node where a

coordinate is selected, can no longer be chosen.

4. A training set for the neural network is prepared. This training set consists of eight input

vectors, and four outputs. The first four inputs correspond to the siblings of chosen

25

coordinate, the last four to the siblings of the chosen coordinate’s parent in the tree. The four

outputs are the coefficients corresponding to the chosen coordinate are four children.

5. The trained neural network can now be used to embed the watermark. The eight input

vectors for each selected coordinate is given to the neural network, resulting in four output

vectors. The watermark information is then embedded by replacing the original coefficients

with the output from the neural network, adjusted by a constant.

6. To obtain the watermarked image, an inverse DWT is performed.

Extraction phase:

1. Transform the watermarked image using DWT.

2. Build the tree representation.

3. Use the seed k to start a sequence with the PRNG. This gives a set of coordinates.

4. The corresponding vectors are fed to the neural network, which results in four output

vectors. The difference between the expected output and the actual output of the neural

network provides information of the hidden watermark.

Non-Regular Transform:

– Spreads the energy content into different subbands.

– The subbands are similar to the original image.

26

Fig 16

Our proposed watermarking scheme is based on Non-Regular Transform:

– It inserts the watermark in the transform domain such that the message is more resilient to

attacks.

– The message contains small frequency and high frequency contents.

Fig 17

Our Results:

– The average correlation coefficient for all the subbands after JPEG2000 compression

27

at different bit rates.

Fig 18

28

CHAPTER 8

Application

Ownership assertion

Watermarks can be used for ownership assertion. To assert ownership of an image, Alice can

generate a watermarking signal using a secret private key, and then embed it into the original

image. She can then make the watermarked image publicly available. Later, when Bob

contends the ownership of an image derived from this public image, Alice can produce the

unmarked original image and also demonstrate the presence of her watermark in Bob’s

image. Since Alice’s original image is unavailable to Bob, he cannot do the same. For such a

scheme to work, the watermark has to survive image processing operations aimed at

malicious removal. In addition, the watermark should be inserted in such a manner that it

cannot be forged as Alice would not want to be held accountable for an image that she does

not own.

Fingerprinting

In applications where multimedia content is electronically distributed over a network, the

content owner would like to discourage unauthorized duplication and distribution by

embedding a distinct watermark (or a fingerprint) in each copy of the data. If, at a later point

in time, unauthorized copies of the data are found, then the origin of the copy can be

determined by retrieving the fingerprint. In this application the watermark needs to be

invisible and must also be invulnerable to deliberate attempts to forge, remove or invalidate.

Furthermore, and unlike the ownership assertion application, the watermark should be

resistant to collusion. That is, a group of users with the same image but containing different

fingerprints should not be able to collude and invalidate any fingerprint or create a copy

without any fingerprint.

Fraud and tamper detection

When multimedia content is used for legal purposes, medical applications, news reporting,

and commercial transactions, it is important to ensure that the content was originated from a

specific source and that it had not been changed, manipulated or falsified. This can be

achieved by embedding a watermark in the data. Subsequently, when the photo is checked,

29

the watermark is extracted using a unique key associated with the source, and the integrity of

the data is verified through the integrity of the extracted watermark. The watermark can also

include information from the original image that can aid in undoing any modification and

recovering the original. Clearly a watermark used for authentication purposes should not

affect the quality of an image and should be resistant to forgeries. Robustness is not critical as

removal of the watermark renders the content inauthentic and hence of no value.

ID card security

Information in a passport or ID (e.g., passport number, person’s name, etc.) can also be

included in the person’s photo that appears on the ID. By extracting the embedded

information and comparing it to the written text, the ID card can be verified.

The inclusion of the watermark provides an additional level of security in this application.

For example, if the ID card is stolen and the picture is replaced by a forged copy, the failure

in extracting the watermark will invalidate the ID card.

30

Conclusion

As described recent developments in the digital watermarking of images in which the

watermarking technique is invisible. Digital watermarking is a rapidly evolving area of

research and development. Digital watermarking technology can provide a new way to

protect the copyright of multimedia information and to ensure the safe use of multimedia

information.

31

REFRENCES

Techniques for data hiding. IBM Systems

Journal, 35(3-4):313–336, 1996.

Tolerant Image Authentication”, Proceedings, Int. Conf. Image Proc.,

Chicago, Oct. 1998.

arking in the Wavelet Transform Domain”Peter Meerwald, Januar 2001

-S. Chiang, C.-P. Chang, and T.-M. Tu, “Robust spatial watermarking technique for colour

images via direct saturation adjustment,”