Designing a Fragile cum Robust

Digital Watermarking Scheme for

Big Data Application
by

Name Roll No. Registration No:

SOHINI MONDAL 11700313096 131170110280
ADVAITAA BISWAS 11700313002 131170110186
ARUNOTHPOL DEBNATH 11700313024 131170110208
ANIRBAN DASGUPTA 11700313008 131170110192

A comprehensive project report has been submitted in partial fulfillment of the

requirements for the degree of

Bachelor of Technology
in
ELECTRONICS & COMMUNICATION ENGINEERING
Under the supervision of

Dr. Tirtha Sankar Das

Professor

Department of Electronics & Communication Engineering

RCC INSTITUTE OF INFORMATION TECHNOLOGY
Affiliated to Maulana Abul Kalam Azad University of Technology, WestBengal
CANAL SOUTH ROAD, BELIAGHATA, KOLKATA – 700015

May, 2017
 CERTIFICATE OF APPROVAL

This is to certify that the project titled “Designing a Fragile cum Robust Digital Watermarking
Scheme for Big Data Application” carried out by
Name Roll No. Registration No:
SOHINI MONDAL 11700313096 131170110280
ADVAITAA BISWAS 11700313002 131170110186
ARUNOTHPOL DEBNATH 11700313024 131170110208
ANIRBAN DASGUPTA 11700313008 131170110192

for the partial fulfillment of the requirements for B.Tech degree in Electronics and

Communication Engineering from Maulana Abul Kalam Azad University of

Technology, West Bengal is absolutely based on his own work under the

supervision of Dr. Tirtha Sankar Das. The contents of this thesis, in full or in parts, have

not been submitted to any other Institute or University for the award of any degree or

diploma.

Optional in case of External Supervisor .........................................................

.........................................................
Dr./Mr./Ms./Mrs. Dr./Mr./Ms./Mrs.
Designation and Department Professor , Dept. of ECE
Institute RCC Institute of Information Technology

..........................................................

Head of the Department (ECE)

RCC Institute of Information Technology
DECLARATION

“We Do hereby declare that this submission is our own work conformed to the

norms and guidelines given in the Ethical Code of Conduct of the Institute and that, to the

best of our knowledge and belief, it contains no material previously written by another

neither person nor material (data, theoretical analysis, figures, and text) which has been

accepted for the award of any other degree or diploma of

the university or other institute of higher learning, except where due

acknowledgement has been made in the text.”

.......................................................... ..........................................................
Registration No: Registration No:
Roll No: Roll No:

.......................................................... ..........................................................
Registration No: Registration No:
Roll No: Roll No:

Date:

Place:
 CERTIFICATE of ACCEPTANCE

This is to certify that the project titled “Designing a Fragile cum Robust Digital
Watermarking Scheme for Big Data Application” carried out by
Name Roll No. Registration No:
SOHINI MONDAL 11700313096 131170110280
ADVAITAA BISWAS 11700313002 131170110186
ARUNOTHPOL DEBNATH 11700313024 131170110208
ANIRBAN DASGUPTA 11700313008 131170110192

is hereby recommended to be accepted for the partial fulfillment of the requirements for

B.Tech degree in Electronics and Communication Engineering from Maulana Abul Kalam

Azad University of Technology, West Bengal

Name of the Examiner Signature with Date

1. ……………………………………………………………………

2.……………………………………………………………………..

3.…………………………………… ………………………………

4. ……………………………………. ………………………………
 Contents
• Abstract
• Introduction
• Literature Survey
• Problem Statement
• Working Principle & Block Diagrams
• Results
• Attacks
• Conclusion
• Future Scope
• Reference
• Annexure
 Abstract

In the past few years, digital multimedia distribution over the Internet has grown rapidly due to
the latest developments in technologies. As obvious, digital data can be easily shared, processed
or used without any degradation in quality. On the other hand, this leads to the serious security
problems. Therefore, protection of the digital data is becoming an utmost important issue. As a
solution to this problem, different authentication and protection techniques are being used.
Digital watermarking is one of the major technologies to embed additional information into the
host signal to ensure the security and protection of multimedia data as well. Here, a Fragile cum
Robust Digital Watermarking scheme for big data purpose is proposed. The primary objective
results are found satisfactory.
Introduction
The increase in the availability of digital data (e.g. video, audio, and images) on the World Wide
Web has led to large scale unauthorized copying and increased the opportunity for violation of
copyright and tampering with (or the modification of) content. The reason is simple — digital
representation of media facilitates access and potentially improves the portability, efficiency, and
accuracy of the information presented. As a result, there is a pressing need to manage and protect
visual material against manipulation and illegal duplication. One approach to address this
problem involves embedding an invisible structure into a host multimedia data to mark
ownership of them. To accomplish this, a large number of information hiding techniques have
been proposed in the literature. The results achieved in the last 6 years, in a number of
application areas involving audio, video, and digital images, have pointed to information hiding
as one important topic related to the area of information security. The information hiding area
brings together researchers with very different backgrounds: electrical engineering, signal and
image processing, computer science, and cryptography. The main disciplines studied so far have
focused on covert channels, steganography, anonymity, and watermarking. This classification of
information hiding techniques was first proposed by Bauer. However, other researchers consider
steganography, digital watermarking, and fingerprinting at the same level. Recent research has
pointed to steganography and digital watermarking as two areas which are generally referred to
as information hiding.

 Historical Roots of Information Hiding:

Although information hiding seems to some as though it is a new science, some of the first
documented examples can be found in the Histories of Herodotus, in which the father of history
relates several stories from the times of ancient Greece. One classical example refers to
Histiaeus, who wished to inform his allies when to revolt against the enemy (Persians). To do so,
Histiaeus shaved the head of a trusted servant and then tattooed a message on his scalp. When the
servant’s hair grew back, he was sent through Persian territory to the allies. The slave appeared
to be a harmless traveller passing through the area, but upon his arrival, the servant reported to
the leader of the allies and indicated that his head should be shaved to reveal the message. This
method was still used by some German spies in the beginning of the 20th century. A large
number of techniques were invented or reported by Aineias the Tactician, including letters
hidden in messengers’ soles or women’s earrings, text written on wood tablets and then white-
washed, and notes carried by pigeons. He also proposed hiding text by changing the heights of
letter strokes or by making very small holes above or below letters in a cover—text.
This technique was still in use during the 17th century. Later, Wilkins improved it by using
invisible ink to print very small dots instead of making holes.
Recent research has yielded more advanced techniques in information hiding. Such techniques
have been used for centuries to prove the authenticity of physical materials, and recently they
have been used to protect digital property as well.

 A Taxonomy of Information Hiding Techniques

A classification of information hiding techniques was proposed by Bauer. The main
subdisciplines of information hiding are covert channels, steganography, anonymity, and
watermarking. Next, we give a brief definition of each of these sub-disciplines.

Covert Channels: Lampson introduced and illustrated many covert channels in his Confinement
Problem paper. This technique was defined in the context of multilevel secure systems (e.g.
military computer systems), in which communication paths which were not designed to transfer
information at all. Instead, these channels were used by untrustworthy programs to leak
information to their owner while performing a service for another program. For instance, let us
suppose that a program A (service) processes some information for a program B (customer). The
customer program will want to ensure that the service program cannot access (read or modify)
any of its data except those items to which it explicitly grants access. In addition, the service
program must be protected from intrusion by the customer program, since the service program
may have its own private data. Such communications channels have been studied to find ways to
confine these programs, i.e., some secure systems need to safeguard data from unauthorized
access or modification or programs from unauthorized execution.

Steganography: Steganography (from the Greek steganos, or “covered,” and graphz'e, or

“writing”) is the hiding of a secret message within an ordinary message and its extraction at a
destination. Steganography takes cryptography one step further by hiding an encrypted message
so that no one suspects its existence. Ideally, anyone scanning the data would fail to know that it
contains encrypted data. While cryptography is about protecting the content of messages,
steganography is about preventing the recognition of their very existence. Steganographic
techniques may be roughly classified into linguistic steganography and technical steganography.
The former consists of linguistic or language forms of hidden writing. A widely used method of
linguistic steganography is the acrostic. One of the most famous examples is Giovanni
Boccaccio’s Amorosavistone, which is called “the world’s hugest acrostic.”. The latter method,
such as invisible ink, tries to hide messages physically. The most famous examples go back to
antiquity. However, in recent years, with the move towards digitalization, messages can be
embedded into digital media using steganographic techniques, and transmitted through the
Internet rapidly. There are many papers in the literature that cover the main steganographic
techniques and the basic concepts as well.

Anonymity:The main purpose of this technique is to find ways to hide the meta-content of
messages, i.e., the sender and the recipients of a message. The basic idea is that one can obscure
the trail of a message by using a set of remailers or routers as long as the intermediaries do not
collude. As a result, trust remains the cornerstone of these tools. There are different varieties,
depending on who is anonymz'zed: the sender, the receiver, or both. Web applications have
focused on receiver anonymity, whereas e-mail users are concerned with sender anonymity.

Watermarking: Digital watermarking, as opposed to steganography, has the additional

requirement of robustness against possible attacks. The purpose of digital watermarking is to
provide evidence that can be used within the legal system to prove that some copyright violation
has occurred. On the other hand, digital watermarking alone is not sufficient protection against
any kind of attack, and should not be considered a panacea for protecting copyrights on digital
content. In this sense, the meaning of robustness is still not very clear, as it depends on the
application. Robustness has strong implications in the overall design of a watermarking system.
Braudaway et al. pointed out that watermarks do not need to be hidden, as some systems use
visible digital watermarking; however, most of the literature has focused on imperceptible
(invisible, transparent, or inaudible) digital watermarks, which have wider applications. There
are three fundamental differences between steganography and watermarking:

(a) the information hidden by a watermarking system is always associated with the digital
object to be protected or its owner, whereas steganographic systems simply hide any
information;

(b) the robustness criteria are also different, since steganography is mainly concerned with
the detection of the hidden message, while watermarking is concerned with removal by a pirate;

(c) steganographic communication is usually point-to-point (between the sender and the
receiver), whereas watermarking techniques are usually one-to-many.

 Difference between Cryptography and Watermarking

Cryptography is defined as the art and science of secret writing. The word itself comes from
Greek where the words kruptos and graphen mean secret and writing, respectively. The focus in
cryptography is to protect the content of the message and to keep it secure from unintended
audiences.The purpose of cryptography is to create schemes or protocols which can still
complete the intended tasks even in the presence of an adversary. Cryptography’s main task is to
ensure users able to communicate securely over an insecure channel. This communication
however must ensure the transmission’s privacy and authenticity.

Watermarking is the process of embedding a message on a host signal. Watermarking, as

opposed to steganography, has the additional requirement of robustness against possible attacks.
A watermark can be either visible or invisible.Using digital watermarking, copyright information
can be embedded into the multimedia data. This is done by using some algorithms. Information
such the serial number, images or text with special significance can be embedded. The function
of this information can be for copyright protection, secret communication, authenticity
distinguish of data file, etc.

 Classification:
  Types of Digital Watermarks:
Watermarks and watermarking techniques can be divided into various categories in various ways.
  Why Digital Watermarking for Video?
Digital watermarking is defined as imperceptibly altering a work in order to embed information
about that work. In the recent years copyright protection of digital content became a serious
problem due to rapid development in technology. Watermarking is one of the alternatives to
copyright-protection problem.
The most common and easily available source of big data is a video. Video watermarking
introduces some issues not present in image watermarking. Due to large amounts of data and
inherent redundancy between frames, video signals are highly susceptible to pirate attacks,
including frame averaging, frame dropping, frame swapping, statistical analysis, etc.

Video Watermarking Applications: The application of video watermarking can be grouped in

6 categories.

1.Transaction Tracking: Transaction tracking is used to track how content was distributed
through a system or transmitted between multiple points. A unique identifier is embedded into
the media at the time of playback, which can later be extracted. In the case of illegal distribution
of the content, it should ideally be possible to identify the source from where the distribution
occurred, possibly identifying the misappropriating party.

2.Broadcast Monitoring: Broadcast monitoring enables broadcasters or content owners to track

or verify the transmission of media in a broadcast system. The watermarks can automatically be
extracted to verify if a commercial has successfully been aired or whether a certain segment of
material was used in a broadcast. The content is usually watermarked by the content owner,
while detection can be done by a monitoring site in the broadcast chain or a third party at the
receiving end.

3.Copy Control: Copy control aims to disable the duplication of copyrighted material on
devices equipped with special watermark detectors. The watermark is used to indicate copy
control information, such as copy_never, copy_once or copy_freely. By implementing
watermark extraction and embedding in devices, the user can be allowed or denied permission to
duplicate content.

4. Content Authentication : Content authentication is a method that attempts to ensure the

integrity of media by detecting attempted tampering of the original content. At creation, the
content is usually watermarked with a semi-fragile watermark, which is designed to be affected
by signal transformations. Tampering with the content should destroy or alter this semi-fragile
watermark, which could then be used to determine that the content is not authentic .
5.Ownership Identification:In this application, watermarks can be used to identify the rightful
owner or creator of content. After the original content was watermarked, disputed ownership can
be resolved by extracting the original watermark. Resolving rightful ownership can, however, be
challenging as pirates may also embed their own watermark, in which case it can be difficult to
determine which is the original watermark. This is known as an ownership deadlock problem.

6.Fingerprinting: This category is only included for clarity, as there exist at least two definitions
of fingerprinting, each with specific characteristics and applications. The first definition of media
fingerprinting is “the art, or algorithm, of identifying component characteristics of a source and
then reducing it into a fingerprint that can uniquely identify it.”These techniques do not add any
additional information to the media, but rather generates a compact signature based on the unique
properties of the content.

 Properties of Watermarking:
Robustness
A digital watermark is called "fragile" if it fails to be detectable after the slightest modification.
Fragile watermarks are commonly used for tamper detection (integrity proof). Modifications to
an original work that clearly are noticeable commonly are not referred to as watermarks, but as
generalized barcodes.
A digital watermark is called semi-fragile if it resists benign transformations, but fails detection
after malignant transformations. Semi-fragile watermarks commonly are used to detect malignant
transformations.
A digital watermark is called robust if it resists a designated class of transformations. Robust
watermarks may be used in copy protection applications to carry copy and no access control
information.
Imperceptibility
A digital watermark is called imperceptible if the original cover signal and the marked signal are
perceptually indistinguishable.
A digital watermark is called perceptible if its presence in the marked signal is noticeable (e.g.
Digital On-screen Graphics like a Network Logo, Content Bug, Codes, Opaque images). On
videos and images, some are made transparent/translucent for convenience for consumers due to
the fact that they block portion of the view; therefore degrading it.
This should not be confused with perceptual, that is, watermarking which uses the limitations of
human perception to be imperceptible.
Capacity
The capacity or payload size of a watermark is an indication of the amount of information that
can be embedded with a watermark. The payload size varies with the application, but in general
an identifier packet of 64 bits is considered appropriate for most applications.
  Attacks On Watermarks:
A successful attack on a watermarking technique refers to a case where the watermark has been
removed or modified to prevent successful extraction, without degrading the quality of the
watermarked content significantly. A successful attack on a watermark does not necessarily
mean that the content was restored to the original, un-watermarked state, but instead that the
watermark detection and extraction processes were defeated. Intentional attacks are deliberate
attempts to prevent successful extraction of embedded watermarks. Unintentional attacks, on the
other hand, are caused by normal signal conversions and compression that may be introduced in
a distribution chain. Normal signal conversion operations.
 Literature Survey
Maher El’Arbi, Chokri Ben Amar and Henri Nicolas has proposed a video watermarking scheme
based on multi resolution motionestimation and artificial neural network. The results show that
embedding watermark where picture content is moving is less perceptible.The proposed scheme
is also shown to be robust against common video processing attacks. [1]

Mr Mohan A Chimanna , Prof. S.R. Khot has proposed a compressive approach for digital video
watermarking ,where watermark image is embedded into the video frame and each video frame
is decomposed into sub images using 2 level Discrete Wavelet Transform (DWT) and Principal
Component Analysis (PCA) Transform is applied for each block in the two bands LL & HH.The
results show no visible difference between watermark frame and original video frame, it shows
the robustness against a wide range of attack such as Gaussian noise, salt & pepper Noise
,median filtering , rotation, cropping etc. The proposed algorithm is however,a non blind
algorithm. [2]

Han-Seung Jung, Young-Yoon Lee, Sang Uk Lee proposed a scene-based temporal

watermarking algorithm. In each scene, segmented by scene-change detection schemes, a
watermark is embedded temporally to one-dimensional projection vectors of the log-polar map,
which is generated from the DFT of a two-dimensional feature matrix. Each column vector of the
feature matrix represents each frame and consists of radial projections of the DFT of the
frame.The proposed algorithm is not robust against random temporal attacks. [3]

Tahani Al-Khatib, Ali Al-Haj, Lama Rajab and Hiba Mohammed proposed an algorithm based
on a cascade of two powerful mathematical transforms; Discrete Wavelets Transform (DWT)
and Singular Value Decomposition (SVD).In the proposed hybrid algorithm, the watermark bits
are not embedded directly on the wavelet coefficients, but rather on the elements of singular
values of the frames' DWT sub-bands. [4]

Jaya Jeswani, Dr. Tanuja Sarode proposed an algorithm in which cover video is divided into
frames and watermark is inserted into selected frames. For selected video frames twodimensional
8×8 discrete cosine transform is carried out on luminance component. Finally binary watermark
is embedded into mid frequency DC coefficients by adjusting coefficients DCT (4, 3) and DCT
(5,2).This algorithm does not embed watermark in all the frames. Also, the result against attacks
like Gaussian Noise, Salt and Pepper Noise etc does not show satisfactory results. [5]

Past Challenges:

The challenges yet to be resolved by the existing work are as follows:

• Most watermarking algorithms are yet to resolve robustness issues completely
• In order to maintain the imperceptibility, the capacity of the system is often
compromised.
• The existing algorithms are prone to failure against common attacks like geometric
attacks, random temporal attacks etc.  Non-blind detection scheme.
Present Challenges:

Each watermarking application has its own set of attributes of the watermarking system which
determines the choice of techniques used for embedding and detecting the watermark.
Commonly discussed attributes of real world systems include:
• First issue is to maintain balance between imperceptibility, robustness and capacity as
increasing one factor adversely effect on other and a good digital watermarking system
possess all three features. To achieve good imperceptibility, watermark should be
embedded in high frequency component whereas robustness occurs in low frequency
component.
• Another one is in fragile watermarking, where data recovery against cropping is a
challenging issue. In fragile watermarking, slight distortion results in destruction of
watermark.
• In the age of Big Data, capacity of the watermarking system is a very important
factor.More capacity means imperceptibility is compromised. So, equilibrium needs to be
maintained.
• Next issue is robustness.The watermarking technique should be universal so that it is
robust against various kinds of attacks.
• Other issue is computational cost i.e. cost of inserting and detecting watermark that
should be minimized.

Problem Statement
Design a fragile cum robust digital watermarking scheme for Big Data application. In our
project we have tried to resolve the issues of

• copyright protection
• tampering detection
 • capacity

Working Principle
Although existing literature on watermarking embedding techniques deals with image, video and
audio files, most of publications refer to images. Based on recent publications, we roughly
divided watermarking techniques into two classes taking, into account the domain in which the
watermark signal is embedded. The first class refers to spatial domain techniques which embed
the data by directly modifying the pixel values of the original image. The other class lays on
transform domain methods which embed the data by modulating the transform domain
coefficients.

 Though Spatial Domain approach is easy to implement and relatively computationally

efficient, but not robust against common digital signal processing operations such as video
compression. So we selected to take frequency domain approach, as it disperses the watermark in
the spatial domain of the video frame, hence making it very difficult to remove the embedded
watermark.
 The Discrete Wavelet Transform offers the advantage that it can take advantage of
special properties of alternate domains to address the limitations of pixel-based methods.
Normalization performed in both insertion and detection phase makes the watermark to be robust
to resizing of the video frame. It is robust against format conversions because the watermark is
inserted before compression.
Proposed Algorithm for Video Watermarking

Part 1: Watermark Embedding

Original Video

Extraction of
Frames

RGB to Gray
conversion

Application of
DWT

Generation of
binary pattern

Key=Binary pattern
XOR Watermark

Embedding the
Key in the frames

Application of
IDWT

Watermarked
video
Discrete Wavelet Transform:

The Discrete Wavelet Transform (DWT) yields the fast computation of Wavelet transform. In
this algorithm, DWT is applied on each frame and the video watermarking is done on every
frame to which DWT is applied. For 2-D frame, applying DWT means applying 1-D filter in two
dimensions. The filter then divide the frame into four non overlapping sub-bands called as LL1,
LH1, HL1 and HH1. In this paper, L stands for low pass, H stands for high pass, while, the
number indicates the level of DWT applied. To obtain the next level, the LL1 sub-band is
selected and again it is divided into four non overlapping sub-bands named - LL2, LH2, HL2 and
HH2.In this way we applied level 3 DWT on each frame.
Part 1: Key Generation & Watermark Embedding

The watermarking embedding procedure for each frame can be described in the following steps:

Step 1: Decompose the video in individual frames and convert the frames in Gray scale images.

Step 2: Apply Level 6, 2D DWT on each frame to decompose into various sub-bands using db2
Wavelet. This leads to LL6 frequency band.

Step3: For each frame, calculate mean of low freq coefficient using:

; k = 1,2,...P
where P = total number of frames available in the video
N = the total number of coefficients available in LL6 component of corresponding
frame
ak = low frequency DWT coefficients.
Hence Tk represents a vector consisting of P thresholds for the video frames.

Step 4: Compare the low frequency coefficient with the mean value Tk for k-th frame obtained in
step 3, to generate a binary pattern:
Wk = 1, if ak ≥ Tk
= 0, if ak < Tk; k = frame number

This leads to generation of a binary pattern corresponding to the importance of DWT coefficients
in LL6 part, based on the threshold obtained.This can be termed as weights of the coefficients to
generate the key.
Step 5: Generate the key ' Kk ' by applying XOR operation between Wk and the watermark
image- G, for k-th frame. Let the dimension of the key Kk be m×n.
Kk = Wk (i) XOR G(i)
Now, Kk is the key to extract the watermark during extraction process.

Step 6: Generate Pseudo-Noise sequences.Generate (m×n) number of PN sequences having the

same dimension as the HH coefficient at Level 1.Only the states having their mean value close to
0 are selected. The states are stored for future use in watermark detection.

Step 7: Embedding the key in the frames if

Kk(i)==1

HH’ = HH + g *pns

else

HH’ = HH - g *pns
where k=k-th frame in the video

K=Key matrix of the frame

HH=detail coefficient at Level 1

HH’=modified detail coefficient at Level 1

G= Gain

pns= PN sequence

Step-8: Applying IDWT, reconstruct the video from the modified frames to get the watermarked
video.

Extraction of Watermark:

Watermarked video

Extraction of
Frames

Application of
DWT
 Co-relating to
PN sequence

Generation of
Binary Pattern

Key Extraction

Generation of
Watermark

Part 2: Watermark Extraction

The watermarking extraction procedure for each frame can be described in the following steps:

Step 1: Decompose the watermarked video in individual frames and convert the frames in Gray
scale images.
Step 2: Apply Level 1, 2D DWT on each frame to decompose into various sub-bands using db2
Wavelet.
Step3: For each frame, calculate mean of low freq coefficient using:
; k = 1,2,...P
Where P = total number of frames available in the video,
N = the total number of coefficients available in LL3
component of corresponding frame,
ak = low frequency DWT coefficients.
Hence Tk represents a vector consisting of P thresholds for the video frames.

Step 4: Corelate the HH coefficient matrix at Level-1 with the PN sequences generated using the
saved states. The values are saved in the matrix cor_coef(i).
if cor_coef(i) >= 0
cor_value(i) = 1; else
cor_value(i) = 0;
where cor_value matrix is the retrieved key matrix Kr ; r = r-th frame from the reconstructed
video.

Step 5: Apply level 6 DWT.Compare the low frequency coefficient with the mean value Tk for
rth frame, to generate a binary pattern:
Wr = 1, if ar ≥ Tr
= 0, if ar< Tr; r = frame number.
Step-6: Generate the watermark ' G’ ' by applying XOR operation between cor_value and the
binary pattern Wr, for r-th frame.
G’ = Wr (i) XOR cor_value(i) Where
G’ is the reconstructed watermark.

Results
The proposed algorithm was tested for four different videos. The PSNR value is calculated for
the reconstructed frames with respect to the original frame. The PSNR computes the peak
signalto-noise ratio, in decibels, between two images. This ratio is often used as a quality
measurement between the original and a compressed image. The higher the PSNR, the better the
quality of the compressed, or reconstructed image. The MSE represents the cumulative squared
error between the compressed and the original image. The lower the value of MSE, the lower the
error.
How much the extracted watermark from each frame deviates from the embedded watermark is
also calculated by:

mismatch percentage =

The resulting plots are following:

Video File 1:
For specific frames from 'Page 394' video the results are:

# Frame: 4.png
• The binary patterns from original and reconstructed frames are exactly same.
 • The embedded watermark and the extracted watermarks are exactly same, no error.
• For this frame, PSNR = 34.3 dB and MSE = 24.18

#Frame: 79.png
 • The embedded watermark and the extracted watermarks are not exactly same, in this case
two mismatches occurred in the extracted frame, i.e. 0.65% error.
• For this frame, PSNR = 34.2 dB and MSE = 22.43

Video file 2:
For specific frames from 'Walter Mitty' video the results are: #
Frame: 214.png

• The embedded watermark and the extracted watermarks are exactly same, no error.
• For this frame, PSNR = 34.36 dB and MSE = 24.42
For specific frames from 'Lion King' video the results are:
# Frame: 344.png

• The embedded watermark and the extracted watermarks are exactly same, no error.
• For this frame, PSNR = 35.42 dB and MSE = 19.35
For specific frame from 'Cosmos' video the results are:
# Frame: 69.png

• The embedded watermark and the extracted watermarks are not exactly same, two
mismatches occurred in the extracted frame, i.e. 0.65% error.
• For this frame, PSNR = 34.53 dB and MSE = 25.62
Attacks

Three attacks were tested :

1. Salt and Pepper Noise

2. Median Filtering
3. Cropping

The results of these attacks are shown below for a particular frame.

1. Salt and Pepper Noise:

• The extracted watermark doesn't match with the embedded one exactly. One mismatch
occurred when Salt & Pepper noise of noise density 0.05 was introduced.
• For this noise-affected frame, PSNR = 17.53 dB and MSE = 127.13
# Median Filter:

• The extracted watermark doesn't match with the embedded one . 15 mismatches occurred
in newly extracted watermark when 2D median filtering was performed; 5% error.
• For this filter applied frame, PSNR = 41.87 dB and MSE = 4.127

# Cropping:
 • The extracted watermark doesn't match with the embedded one . 40 mismatches occurred
in newly extracted watermark when a portion of the image was cropped; 13% error.
• For this filter cropped frame, PSNR = 17.46 dB and MSE = 1167

Result summary:

Frame Name Percentage of error in PSNR MSE

extracted watermark
'4.png' of 'Page 394' video 0% 34.3 dB 24.18
'79.png' of 'Page 394' video 0.65% 34.2 dB 22.43
'214.png' of 'Walter Mitty' video 0% 34.36 dB 24.42
'344.png' of 'Lion King' video 0% 35.42 dB 19.35
'69.png' of 'Cosmos' video 0.65% 34.53 dB 25.62
'4R.png' with Salt & Pepper 0.3% 17.53 dB 127.13
Noise
'4R.png' with Median Filtering 5% 41.87 dB 4.127
'4R.png' with Cropping 13% 17.46 dB 1167
 Conclusion
The performance of the algorithm is measured by PSNR, MSE and percentage of mismatch. The
algorithm achieves high PSNR values and low MSE without attacks. Watermark was recovered
from the frames with no or minimal error as calculated by the percentage of mismatch. It is also
found to be robust to most common attacks like Salt and Pepper Noise, Median Filtering etc.
However, it was not found to be robust against cropping. The watermark is embedded in the
diagonal coefficients . Slightest tampering will be detected as the binary pattern changes. The
algorithm is thus, fragile. Therefore, we can say that the algorithm realises blind watermarking
which can be applied for both copyright protection and tampering detection.

Future Scope
•
The algorithm can be made more robust by embedding watermark in both the
approximation as well as detail coefficients.
•
The gain for embedding the watermark can be made adaptive.
•
The capacity can be improved by multi bit embedding procedure.
•
PN sequences used can be modified so as to get better correlation values.
Reference
[1] M. El'arbi, C. B. Amar and H. Nicolas, "Video Watermarking Based on Neural
Networks" 2006 IEEE International Conference on Multimedia and Expo, Toronto, Ont., 2006, pp.
1577-1580.
doi: 10.1109/ICME.2006.262846
[2] “ Digital Video Watermarking Techniques for Secure Multimedia Creation and
Delivery”, Mr Mohan A Chimanna, Prof.S.R.Khot / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 3, Issue 2, March
April 2013, pp.839-844

[3] “RST-Resilient VideoWatermarking UsingScene-Based Feature Extraction”, Jung, HS.,

Lee, YY. & Lee, S.U. EURASIP J. Adv. Signal Process. (2004) 2004: 358092.
doi:10.1155/S1110865704405046

[4] “A Robust Video Watermarking Algorithm”,Tahani Al-Khatib, Ali Al-Haj, Lama Rajab
and Hiba Mohammed, Journal of Computer Science 4 (11): 910-915, 2008
ISSN 1549-3636,© 2008 Science Publications

[5] “A New DCT based Color Video Watermarking using Luminance Component”,Jaya
Jeswani, Dr. Tanuja Sarode, IOSR Journal of Computer Engineering (IOSR-JCE) e-ISSN:
22780661, p- ISSN: 2278-8727Volume 16, Issue 2, Ver. XII (Mar-Apr. 2014), PP 83-90
www.iosrjournals.org https://in.mathworks.com/help/wavelet/discrete-wavelet-analysis.html
https://pdos.csail.mit.edu/archive/decouto/papers/pickholtz82.pdf
https://en.wikipedia.org/wiki/Discrete_wavelet_transform
https://www.mathworks.com/discovery/matlab-gui.html

“Efficient Video Watermarking in Selected Frames Based on Fibonacci Series for Ownership
Proof”,Mahesh Sanghavi, Rajeev Mathur and Archana Rajurkar

“Video watermarking using wavelet transform and tensor algebra”,Emad E. Abdallah · A.

Ben Hamza ·Prabir Bhattacharya, SIViP (2010) 4:233–245,DOI 10.1007/s11760-009-0114-7
http://nptel.ac.in/courses/117105079/

“Digital Image Processing using MATLAB” Book by Rafael C. Gonzalez and Richard Eugene
Woods
Annexure