Watermarking for Video Authentication
Content
Watermarking For Video Authentication
By
Pathak Utsav Atulkumar
Prof. Ketki Pathak
(Assistance Professor)
A Thesis Submitted to
Gujarat Technological University in Partial
Fulfillment of the Requirements for
the Degree of Engineering in Electronics & Communication
December 2014
Department
Of
ELECTRONICS & COMMUNICATION ENGINEERING
Sarvajanik College of Engineering & Technology
Dr. R.K. Desai Road,
Athwalines, Surat - 395001, India.
Certificate
This is to certify that research work embodied in this thesis entitled “Watermarking For
Video Authentication” was carried out by Mr. Pathak Utsav Atulkumar (Enrollment No.
130420704023) studying at Sarvajanik College of Engineering and Technology (042) for
partial fulfillment of Master of Engineering degree to be awarded by Gujarat
Technological University. This research work has been carried out under my guidance
and supervision and it is up to my satisfaction.
Date:
Place: Sarvajanik College of Engineering and Technology, Surat
Prof. Ketki Pathak
Assistance Professor
Electronics & Communication department
Sarvajanik College of Engineering &
Technology
Dr. Vaishali Mungurwadi
Principal
Faculty of Engineering,
Sarvajanik College of Engineering &
Technology
Seal of Institute
i
Compliance Certificate
This is to certify that research work embodied in this thesis entitled “Watermarking For
Video Authentication” was carried out by Mr. Pathak Utsav Atulkumar (Enrollment No.
130420704023) at Sarvajanik College of Engineering and Technology (042) for partial
fulfillment of Master of Engineering degree to be awarded by Gujarat Technological
University. He has complied to the comments given by the Dissertation phase – I as well
as Mid Semester Thesis Reviewer to my satisfaction.
Date:
Place: Sarvajanik College of Engineering and Technology, Surat
Signature of Student:
Signature of Guide:
Pathak Utsav Atulkumar
Prof. Ketki Pathak
ii
Declaration Of Originality
We hereby certify that we are the sole authors of this thesis and that neither any part of
this thesis nor the whole of the thesis has been submitted for a degree to any other
University or Institution.
We certify that, to the best of our knowledge, the current thesis does not infringe upon
anyone’s copyright nor violate any proprietary rights and that any ideas, techniques,
quotations or any other material from the work of other people included in our thesis,
published or otherwise, are fully acknowledged in accordance with the standard
referencing practices. Furthermore, to the extent that we have included copyrighted
material that surpasses the boundary of fair dealing within the meaning of the Indian
Copyright (Amendment) Act 2012, we certify that we have obtained a written permission
from the copyright owner(s) to include such material(s) in the current thesis and have
included copies of such copyright clearances to our appendix.
We declare that this is a true copy of thesis, including any final revisions, as approved by
thesis review committee.
We have checked write up of the present thesis using anti-plagiarism database and it is in
allowable limit. Even though later on in case of any complaint pertaining of plagiarism,
we are sole responsible for the same and we understand that as per UGC norms,
University can even revoke Master of Engineering degree conferred to the student
submitting this thesis.
Date:
Place: Sarvajanik College of Engineering and Technology, Surat
Signature of Student:
Signature of Guide:
Name of Student: Pathak Utsav Atulkumar
Name of Guide: Prof. Ketki Pathak
Enrollment No: 130420704023
Institute Code:042
iii
Acknowledgement
I would like to express my deep sense of gratitude to my guide, Prof. Ketki Pathak,
for imparting me valuable guidance and priceless suggestions during the dissertation
and in creating such an excellent report and also for her full dedication and devotion
of time.
I would further like to thank our Head of Department, Prof. Niteen B. Patel and all the
faculty members for giving me this opportunity. I also wish to communicate my deep
sense of gratitude and thanks to the Almighty God.
I would like to express thanks, gratitude and respect to my parents for giving me
valuable advice and support at all times and in all possible ways. Last but not least,
Acknowledgement will not be over without mentioning a word of thanks to all my
friends & my family members who have provided immeasurable support throughout
this journey.
Pathak Utsav Atulkumar
iv
Table of Content
Certificate ............................................................................................................................. i
Compliance Certificate ....................................................................................................... ii
Declaration Of Originality ................................................................................................. iii
Acknowledgement ............................................................................................................. iv
Table of Content ..................................................................................................................v
List of Figures ................................................................................................................... vii
Abstract ............................................................................................................................ viii
1.
Introduction ..................................................................................................................1
1.1 Requirement of Digital Watermarking ..................................................................... 2
1.2 Basic Watermarking Principle .................................................................................. 3
1.3 Purpose of the Study ................................................................................................. 5
2.
Video Watermarking ....................................................................................................6
2.1 Introduction: ............................................................................................................. 6
2.2 The principle of video watermarking: ...................................................................... 6
2.3 Characteristics of video watermarking ..................................................................... 7
2.4 Video watermarking model ...................................................................................... 7
2.5 Video watermarking algorithm ................................................................................. 9
2.5.1 Original video watermark ....................................................................................9
2.5.2Compressed video watermarking .......................................................................11
3.
Video tampering attacks .............................................................................................12
3.1 Introduction............................................................................................................. 12
3.2 Spatial tampering .................................................................................................... 12
3.2.1 Object removal attack ........................................................................................12
3.2.2 Object addition attack ........................................................................................13
3.2.3 Object modification attack.................................................................................13
3.3 Temporal tampering................................................................................................ 14
3.3.1Frame addition attack .........................................................................................14
3.3.2 Frame removal attack ........................................................................................14
3.3.3 Frame shuffling attack .......................................................................................15
v
4.
Parameters to check authentication ............................................................................16
4.1 PSNR (Peak Signal to Noise Ratio)........................................................................ 16
4.2 BER (Bit Error Rate) .............................................................................................. 16
4.3 NC (Normalized Correlation) ................................................................................. 16
5.
Implementation and Planning .....................................................................................17
5.1 Algorithm for DWT and SVD based watermarking technique .............................. 17
5.2 Results obtained from DWT and SVD based watermarking technique ................. 17
5.3 Work done till DP1 ................................................................................................. 19
References ..........................................................................................................................20
vi
List of Figures
Figure 1.1 watermark embedding process[5]................................................................. 4
Figure 1.2 watermark extraction process[5] .................................................................. 4
Figure 1.3 different types of watermarking methodologies[5] ...................................... 4
Figure 2.1 principle of video watermarking[4] .............................................................. 6
Figure 2.2 video watermarking model[4] ...................................................................... 7
Figure 2.3 video watermarking algorithm[4] ................................................................. 9
Figure 2.4 Two band decomposition-reconstruction wavelet filter bank[2]................ 11
Figure 3.1 different spatial tampering attacks[5] ......................................................... 12
Figure 3.2 object removal attack[5] ............................................................................. 13
Figure 3.3 object addition attack[5] ............................................................................. 13
Figure 3.4 frame addition attack[5] ............................................................................. 14
Figure 3.5 frame removal attack[5] ............................................................................. 15
Figure 3.6 frame shuffling attack[5] ............................................................................ 15
Figure 5.1 Frame taken from AVI video sequence ...................................................... 17
Figure 5.2 2-D DWT applied on the selected frame to obtain four sub-bands
LL,LH,HL,HH ............................................................................................................. 18
Figure 5.3 Original frame and watermarked frame obtained for proposed method .... 18
vii
Abstract
As the H.264/AVC based video products become more and more popular, issues of
copyright protection and authentication that are appropriate for this standard will be
very important. Here in this dissertation different watermarking techniques for video
authentication are studied. For authentication different parameters are used for video
like PSNR, NC, BER.
viii
1. Introduction
Multimedia production and distribution, as we see it today, is all digital, from the
authoring tools of content providers to the receivers. The advantages of digital processing
and distribution, like noise-free transmission, software instead of hardware processing,
and improved reconfigurability of systems, are all well known and obvious. Not so
obvious are the disadvantages of digital media distribution. For example, from the
viewpoint of media producers and content providers, the possibility for unlimited copying
of digital data without loss of fidelity is undesirable because it may cause considerable
financial loss. Digital copy protection or copy prevention mechanisms are only of limited
value because access to clear text versions of protected data must at least be granted to
paying recipients which can then produce and distribute illegal copies. Technical attempts
to prevent copying have in reality always been circumvented.
One remaining method for the protection of intellectual property rights (IPR) is the
embedding of digital watermarks into multimedia data. The watermark is a digital code
unremovably, robustly, and imperceptibly embedded in the host data and typically
contains information about origin, status, and/or destination of the data. Although not
directly used for copy protection, it can at least help identifying source and destination of
multimedia data and, as a “last line of defense,” enable appropriate follow-up actions in
case of suspected copyright violations.
While copyright protection is the most prominent application of watermarking
techniques, others exist, including data authentication by means of fragile watermarks
which are impaired or destroyed by manipulations, embedded transmission of value
added services within multimedia data, and embedded data labeling for other purposes
than copyright protection, such as data monitoring and tracking. An example for a datamonitoring system is the automatic registration and monitoring of broadcasted radio
programs such that royalties are automatically paid to the IPR owners of the broadcast
data.
The development of watermarking methods involves several design tradeoffs.
Watermarks should be robust against standard data manipulations, including digital-to
analog conversion and digital format conversion. Security is a special concern, and
watermarks should resist even attempted attacks by knowledgeable individuals. On the
other hand, watermarks should be imperceptible and convey as much information as
1
possible. In general, watermark embedding and retrieval should have low complexity
because for various applications, real-time watermarking is desirable.
1.1 Requirement of Digital Watermarking
1) A watermark shall convey as much information as possible, which means the
watermark data rate should be high.
2) A watermark should in general be secret and should only be accessible by authorized
parties. This requirement is referred to as security of the watermark and is usually
achieved by the use of cryptographic keys.
3) A watermark should stay in the host data regardless of whatever happens to the host
data, including all possible signal processing that may occur, and including all hostile
attacks that unauthorized parties may attempt. This requirement is referred to as
robustness of the watermark. It is a key requirement for copyright protection or
conditional access applications, but less important for applications where the watermarks
are not required to be cryptographically secure, for example, for applications where
watermarks convey public information.
4) A watermark should, though being unremovable, be imperceptible.
In the following, a few of the mentioned requirements and the resulting design issues are
highlighted in more detail.
1) Watermark Security and Keys: If security, i.e., secrecy of the embedded
information, is required, one or several secret and cryptographically secure keys have to
be used for the embedding and extraction process. For example, in many schemes,
pseudorandom signals are embedded as watermarks. In this case, the description and the
seed of the pseudorandom number generator may be used as key. There are two levels of
secrecy. In the first level, an unauthorized user can neither read nor decode an embedded
watermark nor can he detect if a given set of data contains a watermark. The second level
permits unauthorized users to detect if data are watermarked, however, the embedded
information cannot be read without having the secret key. Such schemes can, for
example, embed two watermarks, one with a public key and the other with a secret key.
Alternatively, a scheme has been proposed which combines one or several public keys
with a private key and embeds one combined public/private watermark, rather than
several watermarks. When designing an overall copyright protection system, issues like
secret key generation, distribution, and management (possibly by trusted third parties), as
well as other system integration aspects have to be considered.
2
2) Robustness: In the design of any watermarking scheme, watermark robustness is
typically one of the main issues, since robustness against data distortions introduced
through standard data processing and attacks is a major requirement. Standard data
processing includes all data manipulation and modification that the data might undergo in
the usual distribution chain, such as data editing, printing, enhancement, and format
conversion. “Attack” denotes data manipulation with the purpose of impairing,
destroying, or removing the embedded watermarks. Although it is possible to design
robust watermarking techniques, it should be noted that a watermark is only robust as
long as it is not public, which means as long as it cannot be read by everyone.
3) Imperceptibility: One of the main requirements for watermarking is the perceptual
transparency. The data embedding process should not introduce any perceptible artifacts
into the host data. On the other hand, for high robustness, it is desirable that the
watermark amplitude is as high as possible. Thus, the design of a watermarking method
always involves a trade-off between imperceptibility and robustness. It would be optimal
to embed a watermark just below the threshold of perception. However, this threshold is
difficult to determine for real-world image, video and audio signals.
4) Watermark Recovery With or Without the Original Data: Watermark recovery is
usually more robust if the original, unwatermarked data are available. Further, availability
of the original data set in the recovery process allows the detection and inversion of
distortions which change the data geometry. This helps, for example, if a watermarked
image has been rotated by an attacker. However, access to the original data is not possible
in all cases, for example, in applications such as data monitoring or tracking. For other
applications, like video watermarking, it may be impractical to use the original data
because of the large data volume, even if it is available. It is, however, possible to design
watermarking techniques that do not need the original for watermark extraction. Most
watermarking techniques perform some kind of modulation in which the original data set
is considered a distortion. If this distortion is known or can be modelled in the recovery
process, explicitly designed techniques allow its suppression without knowledge of the
original. In fact, most recent methods do not require the original for watermark recovery.
1.2 Basic Watermarking Principle
The basic idea in watermarking is to add a watermark signal to the host data to be
watermarked such that the watermark signal is unobtrusive and secure in the signal
3
mixture but can partly or fully be recovered from the signal mixture later on if the correct
cryptographically secure key needed for recovery is used.
Watermark embedding process is shown below in Fig.1.1
watermark
Digital
Data
Watermarking
Watermarked data
Secret or public key
Figure 1.1 watermark embedding process[5]
And the watermark extraction process is shown below in Fig.1.2
Watermark
and/or original image
Digital
image
Watermarked data
Watermarking
Watermark
image
Secret or public key
Figure 1.2 watermark extraction process[5]
Different types of Watermarking methodologies are given below
Figure 1.3 different types of watermarking methodologies[5]
4
1.3 Purpose of the Study
Here in this dissertation purpose is to apply different watermarking techniques for video
authentication for compressed and uncompressed video and to find PSNR,SSIM and BER
parameters for watermarked video to check authentication of video.
5
2. Video Watermarking
2.1 Introduction:
Video watermarking is loaded in the digital watermark on the video, which uses video
data redundancy and embeds the copyright information into the original video data,
thereby protecting the copyright of digital products or the legitimate interests of copyright
holders. In video watermarking, the movement of video properties and the defect of
human visual model make video watermarking technology fall behind the image
watermark technology. Current video coding technologies (such as H. 264), some special
attacks, the blind video watermarking detection and real-time features, have brought new
challenges to video watermarking.
2.2 The principle of video watermarking:
A complete digital watermarking system should include three basic parts: watermark
generation, watermark embedding and watermark extraction or detection. Watermark
embedding algorithm uses the symmetric key or public key to make the watermark
information embed into the original carrier to get concealed carrier. Watermark detection
/ extraction algorithm using the corresponding key vector from the hidden watermark is
detected or recovered without the key. The attacker is very difficult to find and modify
the hidden watermark vector. Block diagram of watermark embedding and extraction is
shown in Figure 3:
Figure 2.1 principle of video watermarking[4]
6
2.3 Characteristics of video watermarking
Video watermarking not only has the characteristics of digital image watermarking, but
also has its unique characteristics.
High real-time. Three-dimensional video signal has more the amount of data than the
image does. So calculation quality is larger and embedding / detection needs more
times. The procession of embedding, using video compression standard for these
specific structures such as motion vector coding, VLC code word, etc., can achieve
efficient algorithms.
Random detection. The watermark is detected in any position of the video rather
than the position according to the video playback order to detect the watermark.
The combination of video codec standard. Video is compressed formats in the
storage or transmission procession, so information hiding technology research cannot
do anything without specific video codec standard. Only combined with the encoding
and decoding standard, video information hiding technology can make the real-time
requirements fulfilment.
Better robustness. Video watermarking scheme must ensure it can resist almost
kinds of processing or attacks.
Blind detection scheme. Non-blind detection needs the original host signal, but it is
very inconvenient to use the original data, because of the huge video data. Blind
detection does not need the original host signal.
2.4 Video watermarking model
According to the embedded strategy, video watermarks algorithm can be divided into
three types of solutions: the original uncompressed video, video codec and the
compressed video stream. The course of the solutions is show in the figure 2.2.
Figure 2.2 video watermarking model[4]
7
Solution 1: Non-compressed domain video watermarking. Watermark is directly
embedded into the original encoded video sequences and then the video containing
watermark image is encoded. Such programs can take advantages of still images
watermarking technology and combine with the structural characteristics of video frames
to form a solution for video watermarking. The advantage of this watermarking algorithm
is relatively mature. Many still image watermarking methods, such as spread spectrum,
the human vision model, image adaptive watermarking irreversible, synchronous
detection mechanism, etc, can be applied to this video watermarking system. But the
solution also has obvious disadvantages: it will increase the video bit rate data streams
and effect video rate of constancy; After compressing and encoding the video data, the
watermark may be lost; For compressed video, we need to decode, and then re-encoding
after embedding watermarking. This process increases the complexity of the calculation
and reduces the video quality.
Solution 2: Watermark embedding and detection module are introduced in the encoder.
Today's video compression standards include ISO / IEC of MPEG-1, MPEG-2, MPEG-4
and ITU-T of H. 261, H. 263, etc. Their basic ideas are motion compensation prediction
coding and block-based transforms coding. The watermark is embedded in the video
encoding phase. We can use the characteristics of encoded data and the principle of video
data compression (such as transformation to the spatial redundancy, quantization and
entropy coding, the motion compensation, motion estimation, etc). The process can be
simple to achieve watermark embedding and extracting real-time processing. This
solution, which is a relatively simple process of watermark embedded in the transform
domain coefficients, does not increase the bit rate video data stream; In addition, because
the watermark is embedded in the transform domain and combines with the encoding
process, we can design a watermarking algorithm against multiple attacks. But the
solution needs to modify the encoder and decoder and the video codec cannot perform
common watermark embedding and detection, so there is GOP error accumulation.
Solution 3: Embed watermark in the compressed domain, i.e. the watermark is directly
embedded into the compressed encoded bit stream. The advantages of this solution are no
decoding and re-encoding process, which will not cause a decline in video quality, while
the computational complexity is lower. The disadvantage is that the compressed bit-rate
constraints the size of the watermarked data. The strength of watermark embedded is
limited by the error of the video decoder and the embedding strategy is constrained by the
8
coding standard and the video compression algorithms. This algorithm takes advantage of
video compression coding standard without the compressed video stream re-encoding and
full decoding. It has a lower computational complexity and a higher rate of embedded
watermark. But it’s disadvantages are that the capable of deal channel interference is bad;
the watermark can be destroyed by adding random bits labeled the least bit amplitude in
the VLC code using the same algorithm; traditional filtering, resampling and time-domain
scaling processing will also effect the extraction of the watermark. Some algorithms are
also proposed to embed watermark in motion vector or in the range of larger and small
phase angle change motion vectors. In the compressed video sequences, most of the
frames are the frame motion compensation coding, so the motion vector hide the
watermark information can be more effective to use the information of video bit stream.
2.5 Video watermarking algorithm
There are a lot of embedded algorithms in the video watermarking flied. In my views, we
can divide them into two categories: embedded watermarks in the original video and
embedded watermarks in the compressed video stream. It is show in the figure 2.3.
DCT
Original
video
video
Compressed
video
Spatial
domain
DWT
Frequency
domain
DFT
DCT
coefficient
DHT
Motion
vector
SVD
Inter-frame
correlation
Figure 2.3 video watermarking algorithm[4]
2.5.1 Original video watermark
The original video watermarks are deal with the uncompressed video data and have
nothing to do with video code format. So there are two categories in this flied:
It can obtain the original video data. It can embed/extract watermarks into the
original video and the process is quite simple.
9
It can only get the code video data. So the encode video is decoded and then
embedded/extracted the watermarks.
a) Spatial domain watermarking:
Spatial domain watermarks are directly embedded into the original video data. The
embedded watermark signals are usually added to the luminance component, and
sometimes some color components are added to, or added to the color components in all.
The advantage is simple and has a lower complexity, but it cannot be perceived in the
robustness.
The main purpose of embedding or extracting watermarking on the pixel values directly
is to reduce the complexity of watermarking, but as the processor speeds continue to
increase, watermark embedding and extraction has become the most common method in
the frequency domain.
b) Frequency domain watermarking:
The watermarking in transform domain is that watermarking is embedded / extracted in
some transform domain of the original video. The common transforms are DCT, DFT,
DWT, DHT, etc. There are three corresponding methods: (1) Video stream is viewed as
three-dimensional signals, two dimensions in space, one in time. (2) Video stream is seen
as still image. The watermarks are embedded into the single frame image. (3) Video is
processed according to the blocks in transform domain. Most of methods are combined
with the video encoder.
Osama S. Faragallah[2], proposed an efficient DWT-based SVD digital video
watermarking method. The proposed method makes use of two powerful mathematical
transforms; DWT and SVD using the additive method. The two transforms were
combined and their mixtures exploit their attractive features such as frequency
localization of the DWT, and the compact capturing of the semi-global features and the
geometric information of images. The proposed DWT-based SVD video watermarking
method worked by applying two-level decomposition DWT to convert video frames into
bands of different frequency. Then, the high and middle frequency bands HH, LH and HL
are SVD transformed and a watermark is embedded in the diagonal matrix of each band.
The idea of the DWT is to represent a signal as a series of approximation and details.
Figure 2.4 shows the two band decomposition reconstruction wavelet filter bank. The
signal is lowpass filtered with H0(z) to give an approximation signal and highpass filtered
with H1(z) to give a detail signal. The basis function is chosen such that a perfect
reconstruction can be achieved.
10
G 0( z ) ( z 1)
1
G1( z ) ( z 1)
2
H 1( z ) ( z 1 1)
Figure 2.4 Two band decomposition-reconstruction wavelet filter bank[2]
2.5.2Compressed video watermarking
The MPEG coding standards are usually considered in the video watermarking. There are
three kinds of video in MPEG coding standards: I-frame, P- frame and b-frame. I frame
encoding is similar to JPEG, using intra-flied neighbouring pixels of space to compress
redundant information; P Frame need use the previous frame when it is encoding and
current frame can be used as a reference frame for prediction. B frame data compression
affection is obvious. It requires the previous frame and follow-up frame to prediction. B
frame itself cannot be predicted as the other reference frame. P frame and B frames are
adjacent frames using a time-domain redundancy to compress information, while the
prediction error signal can remove spatial redundancy. The process of removing spatialdomain redundancy uses DCT, quantization and entropy coding techniques and the
process of removing time-domain redundancy uses the motion compensation, motion
representation and motion estimation techniques.
In this three techniques are used:
a) Embed watermark in DCT coefficient
b) Embed watermark in motion vector
c) Watermarking using inter-frame correlation
11
3. Video tampering attacks
3.1 Introduction
There are several possible attacks that can be applied to alter the contents of a video data.
Video sequence can be viewed as a collection of consecutive frames with temporal
dependency, in a three dimensional plane. This is called the regional property of the video
sequences. When a malicious alteration is performed on a video sequence, it either attacks
on the contents of the video (i.e. visual information presented by the frames of the video),
or attacks on the temporal dependency between the frames. Therefore based on the
regional property of the video sequences, we can broadly classify the video tampering
attacks into three categories: spatial tampering attacks, temporal tampering attacks and
the combination of these two, spatio-temporal tampering attacks.
3.2 Spatial tampering
In spatial tampering malicious alterations are performed on the content of the frames (XY axis). The operations that can be done as tampering attack in spatial tampering are
cropping and replacement, morphing, content (object) adding and removing etc
Spatial
Tampering
Attacks
Object
Removal Attack
Attack On
Foreground
Object
Attack on
Background
Object
Object
Modification
Attack
Object Addition
Attack
Attack On
Foreground
Object
Attack on
Background
Object
Attack On
Foreground
Object
Attack on
Background
Object
Figure 3.1 different spatial tampering attacks[5]
3.2.1 Object removal attack
In object removal attack, the objects of the frames of the video are eliminated. This kind
of attack is commonly performed where a particular person wants to hide his/her presence
in a certain sequence of frames. This attack can be performed with both kinds of object,
foreground objects and background object.
12
Figure 3.2 object removal attack[5]
3.2.2 Object addition attack
When an object is inserted in a frame or in a set of frames then there is a kind of spatial
tampering attack: say Object addition attack. It can also be performed with both kinds of
objects, foreground objects and background objects.
Figure 3.3 object addition attack[5]
3.2.3 Object modification attack
In Object modification attack, an existing object of the frame(s) can be modified in such a
way that the original identity of that object is lost, and a new object may be in appearance
which is totally different from the original object. The object modification attacks can be
existed in many prospects in the given video. For instance, the size and shape of the
existing object may be changed, the colour of the object may be changed or it may be
discoloured, and with the help of additional effect the nature of the object and it’s relation
with other objects also may be changed.
13
3.3 Temporal tampering
In temporal tampering manipulation is performed on the sequence of frames. The focus is
on the temporal dependency. Temporal tampering attacks are mainly affecting the time
sequence of visual information, captured by video recording devices. The common
attacks in temporal tampering are frame addition, frame removal and frame reordering or
shuffling.
3.3.1Frame addition attack
In frame addition attack, additional frames from another video, which has the same
statistical properties, are intentionally inserted at some random locations in a given video.
Figure 3.4 frame addition attack[5]
3.3.2 Frame removal attack
In frame removal attack the frames of the given video are intentionally eliminated. In this
kind of attack frames or set of frames can be removed from a specific location to a fixed
location or can be removed from different locations. It depends upon the intention.
Commonly this kind of tampering attack is performed on surveillance video where an
intruder wants to remove his/her presence at all.
14
Figure 3.5 frame removal attack[5]
3.3.3 Frame shuffling attack
In frame shuffling attack, frames of a given video are shuffled or reordered in such a way
that the correct frame sequence is intermingled and wrong information is produced by the
video as compared to original recorded video.
Figure 3.6 frame shuffling attack[5]
15
4. Parameters to check authentication
4.1 PSNR (Peak Signal to Noise Ratio)
Equation to find mean PSNR for whole video sequence is given by below equation:
nr _ frames
PSNR
PSNR
i 1
nr _ frames
Where nr_frame = total no of frames
Where PSNR for one frame can be obtained by given formula:
2552
PSNR 10 * log(
)
MSE
y
x
( Ai , j Bi , j ) 2
MSE
x* y
i 1 j 1
Where x=width of frame/image
y=height of frame/image
4.2 BER (Bit Error Rate)
The mean decoding BER for the watermarked videos after they were attacked is given by
below formula:
BER
1 P
wout ( j ) win ( j )
P j 1
Where wout is the extracted watermark, win is the original watermark and P is the size of
the watermark.
4.3 NC (Normalized Correlation)
Normalized correlation is obtained by below formula:
NC ( win, wout )
w (i, j )w (i, j )
[w (i, j )]
i
j
i
in
j
out
in
2
Where win is the original watermark and wout is the extracted watermark.
16
5. Implementation and Planning
5.1 Algorithm for DWT and SVD based watermarking technique
1. Insert video in AVI format
2. Divide video into frames.
3. Take one frame from total frames obtained in step-2.
4. Apply 2-D DWT on the selected frame to obtain four levels HH,HL,LH,LL.
5. Take LL sub-band and apply SVD on that sub-band.
6. Take watermark image.
7. Apply 2-D DWT on that image.
8. Take LL sub-band and apply SVD on that sub-band.
9. Add results obtained from step-5 and step-8 using some scaling factor
10. Apply SVD on the result obtained in step-9.
11. Apply IDWT on that image to obtain watermarked image/frame.
5.2 Results obtained from DWT and SVD based watermarking
technique
Figure 5.1 Frame taken from AVI video sequence
17
LL band of image
LH band of image
HL band of image
HH band of image
Figure 5.2 2-D DWT applied on the selected frame to obtain four sub-bands
LL,LH,HL,HH
Figure 5.3 Original frame and watermarked frame obtained for proposed method
18
5.3 Work done till DP1
Study different watermarking techniques used for video authentication.
Use DWT-SVD based method on the video sequence to study watermarking.
19
References
[1] Dawen Xu, Rangding Wang, Jicheng Wang, “A novel watermarking scheme for
H.264/AVC video authentication ”, Signal Processing: Image Communication(Elsevier)
26 (2011)267–279
[2] Osama S. Faragallah, “Efficient video watermarking based on singular value
decomposition in the discrete wavelet transform domain”, Int. J. Electron. Commun.
(Elsevier) 67 (2013) 189– 196
[3] Po-Chyi Su, Chin-Song Wu, Ing-Fan Chen, Ching-Yu Wu, Ying-Chang Wu, “A
practical design of digital video watermarking in H.264/AVC for content authentication ”,
Signal Processing: Image Communication(Elsevier) 26 (2011)413–426
[4] Xing Chang, Weilin Wang, Jianyu Zhao, Li Zhang, “A Survey of Digital Video
Watermarking”, Seventh International Conference on Natural Computation 2011 IEEE
[5] Richa Singh, Mayank Vatsa, Sanjay K. Singh, Saurabh Upadhyay, “Integrating SVM
classification with SVD watermarking for intelligent video authentication”, Telecommun
Syst(Springer) (2009) 40: 5–15
[6] Hartung, F.; Kutter, M., "Multimedia watermarking techniques," Proceedings of the
IEEE , vol.87, no.7, pp.1079-1107, Jul 1999
20
By
Pathak Utsav Atulkumar
Prof. Ketki Pathak
(Assistance Professor)
A Thesis Submitted to
Gujarat Technological University in Partial
Fulfillment of the Requirements for
the Degree of Engineering in Electronics & Communication
December 2014
Department
Of
ELECTRONICS & COMMUNICATION ENGINEERING
Sarvajanik College of Engineering & Technology
Dr. R.K. Desai Road,
Athwalines, Surat - 395001, India.
Certificate
This is to certify that research work embodied in this thesis entitled “Watermarking For
Video Authentication” was carried out by Mr. Pathak Utsav Atulkumar (Enrollment No.
130420704023) studying at Sarvajanik College of Engineering and Technology (042) for
partial fulfillment of Master of Engineering degree to be awarded by Gujarat
Technological University. This research work has been carried out under my guidance
and supervision and it is up to my satisfaction.
Date:
Place: Sarvajanik College of Engineering and Technology, Surat
Prof. Ketki Pathak
Assistance Professor
Electronics & Communication department
Sarvajanik College of Engineering &
Technology
Dr. Vaishali Mungurwadi
Principal
Faculty of Engineering,
Sarvajanik College of Engineering &
Technology
Seal of Institute
i
Compliance Certificate
This is to certify that research work embodied in this thesis entitled “Watermarking For
Video Authentication” was carried out by Mr. Pathak Utsav Atulkumar (Enrollment No.
130420704023) at Sarvajanik College of Engineering and Technology (042) for partial
fulfillment of Master of Engineering degree to be awarded by Gujarat Technological
University. He has complied to the comments given by the Dissertation phase – I as well
as Mid Semester Thesis Reviewer to my satisfaction.
Date:
Place: Sarvajanik College of Engineering and Technology, Surat
Signature of Student:
Signature of Guide:
Pathak Utsav Atulkumar
Prof. Ketki Pathak
ii
Declaration Of Originality
We hereby certify that we are the sole authors of this thesis and that neither any part of
this thesis nor the whole of the thesis has been submitted for a degree to any other
University or Institution.
We certify that, to the best of our knowledge, the current thesis does not infringe upon
anyone’s copyright nor violate any proprietary rights and that any ideas, techniques,
quotations or any other material from the work of other people included in our thesis,
published or otherwise, are fully acknowledged in accordance with the standard
referencing practices. Furthermore, to the extent that we have included copyrighted
material that surpasses the boundary of fair dealing within the meaning of the Indian
Copyright (Amendment) Act 2012, we certify that we have obtained a written permission
from the copyright owner(s) to include such material(s) in the current thesis and have
included copies of such copyright clearances to our appendix.
We declare that this is a true copy of thesis, including any final revisions, as approved by
thesis review committee.
We have checked write up of the present thesis using anti-plagiarism database and it is in
allowable limit. Even though later on in case of any complaint pertaining of plagiarism,
we are sole responsible for the same and we understand that as per UGC norms,
University can even revoke Master of Engineering degree conferred to the student
submitting this thesis.
Date:
Place: Sarvajanik College of Engineering and Technology, Surat
Signature of Student:
Signature of Guide:
Name of Student: Pathak Utsav Atulkumar
Name of Guide: Prof. Ketki Pathak
Enrollment No: 130420704023
Institute Code:042
iii
Acknowledgement
I would like to express my deep sense of gratitude to my guide, Prof. Ketki Pathak,
for imparting me valuable guidance and priceless suggestions during the dissertation
and in creating such an excellent report and also for her full dedication and devotion
of time.
I would further like to thank our Head of Department, Prof. Niteen B. Patel and all the
faculty members for giving me this opportunity. I also wish to communicate my deep
sense of gratitude and thanks to the Almighty God.
I would like to express thanks, gratitude and respect to my parents for giving me
valuable advice and support at all times and in all possible ways. Last but not least,
Acknowledgement will not be over without mentioning a word of thanks to all my
friends & my family members who have provided immeasurable support throughout
this journey.
Pathak Utsav Atulkumar
iv
Table of Content
Certificate ............................................................................................................................. i
Compliance Certificate ....................................................................................................... ii
Declaration Of Originality ................................................................................................. iii
Acknowledgement ............................................................................................................. iv
Table of Content ..................................................................................................................v
List of Figures ................................................................................................................... vii
Abstract ............................................................................................................................ viii
1.
Introduction ..................................................................................................................1
1.1 Requirement of Digital Watermarking ..................................................................... 2
1.2 Basic Watermarking Principle .................................................................................. 3
1.3 Purpose of the Study ................................................................................................. 5
2.
Video Watermarking ....................................................................................................6
2.1 Introduction: ............................................................................................................. 6
2.2 The principle of video watermarking: ...................................................................... 6
2.3 Characteristics of video watermarking ..................................................................... 7
2.4 Video watermarking model ...................................................................................... 7
2.5 Video watermarking algorithm ................................................................................. 9
2.5.1 Original video watermark ....................................................................................9
2.5.2Compressed video watermarking .......................................................................11
3.
Video tampering attacks .............................................................................................12
3.1 Introduction............................................................................................................. 12
3.2 Spatial tampering .................................................................................................... 12
3.2.1 Object removal attack ........................................................................................12
3.2.2 Object addition attack ........................................................................................13
3.2.3 Object modification attack.................................................................................13
3.3 Temporal tampering................................................................................................ 14
3.3.1Frame addition attack .........................................................................................14
3.3.2 Frame removal attack ........................................................................................14
3.3.3 Frame shuffling attack .......................................................................................15
v
4.
Parameters to check authentication ............................................................................16
4.1 PSNR (Peak Signal to Noise Ratio)........................................................................ 16
4.2 BER (Bit Error Rate) .............................................................................................. 16
4.3 NC (Normalized Correlation) ................................................................................. 16
5.
Implementation and Planning .....................................................................................17
5.1 Algorithm for DWT and SVD based watermarking technique .............................. 17
5.2 Results obtained from DWT and SVD based watermarking technique ................. 17
5.3 Work done till DP1 ................................................................................................. 19
References ..........................................................................................................................20
vi
List of Figures
Figure 1.1 watermark embedding process[5]................................................................. 4
Figure 1.2 watermark extraction process[5] .................................................................. 4
Figure 1.3 different types of watermarking methodologies[5] ...................................... 4
Figure 2.1 principle of video watermarking[4] .............................................................. 6
Figure 2.2 video watermarking model[4] ...................................................................... 7
Figure 2.3 video watermarking algorithm[4] ................................................................. 9
Figure 2.4 Two band decomposition-reconstruction wavelet filter bank[2]................ 11
Figure 3.1 different spatial tampering attacks[5] ......................................................... 12
Figure 3.2 object removal attack[5] ............................................................................. 13
Figure 3.3 object addition attack[5] ............................................................................. 13
Figure 3.4 frame addition attack[5] ............................................................................. 14
Figure 3.5 frame removal attack[5] ............................................................................. 15
Figure 3.6 frame shuffling attack[5] ............................................................................ 15
Figure 5.1 Frame taken from AVI video sequence ...................................................... 17
Figure 5.2 2-D DWT applied on the selected frame to obtain four sub-bands
LL,LH,HL,HH ............................................................................................................. 18
Figure 5.3 Original frame and watermarked frame obtained for proposed method .... 18
vii
Abstract
As the H.264/AVC based video products become more and more popular, issues of
copyright protection and authentication that are appropriate for this standard will be
very important. Here in this dissertation different watermarking techniques for video
authentication are studied. For authentication different parameters are used for video
like PSNR, NC, BER.
viii
1. Introduction
Multimedia production and distribution, as we see it today, is all digital, from the
authoring tools of content providers to the receivers. The advantages of digital processing
and distribution, like noise-free transmission, software instead of hardware processing,
and improved reconfigurability of systems, are all well known and obvious. Not so
obvious are the disadvantages of digital media distribution. For example, from the
viewpoint of media producers and content providers, the possibility for unlimited copying
of digital data without loss of fidelity is undesirable because it may cause considerable
financial loss. Digital copy protection or copy prevention mechanisms are only of limited
value because access to clear text versions of protected data must at least be granted to
paying recipients which can then produce and distribute illegal copies. Technical attempts
to prevent copying have in reality always been circumvented.
One remaining method for the protection of intellectual property rights (IPR) is the
embedding of digital watermarks into multimedia data. The watermark is a digital code
unremovably, robustly, and imperceptibly embedded in the host data and typically
contains information about origin, status, and/or destination of the data. Although not
directly used for copy protection, it can at least help identifying source and destination of
multimedia data and, as a “last line of defense,” enable appropriate follow-up actions in
case of suspected copyright violations.
While copyright protection is the most prominent application of watermarking
techniques, others exist, including data authentication by means of fragile watermarks
which are impaired or destroyed by manipulations, embedded transmission of value
added services within multimedia data, and embedded data labeling for other purposes
than copyright protection, such as data monitoring and tracking. An example for a datamonitoring system is the automatic registration and monitoring of broadcasted radio
programs such that royalties are automatically paid to the IPR owners of the broadcast
data.
The development of watermarking methods involves several design tradeoffs.
Watermarks should be robust against standard data manipulations, including digital-to
analog conversion and digital format conversion. Security is a special concern, and
watermarks should resist even attempted attacks by knowledgeable individuals. On the
other hand, watermarks should be imperceptible and convey as much information as
1
possible. In general, watermark embedding and retrieval should have low complexity
because for various applications, real-time watermarking is desirable.
1.1 Requirement of Digital Watermarking
1) A watermark shall convey as much information as possible, which means the
watermark data rate should be high.
2) A watermark should in general be secret and should only be accessible by authorized
parties. This requirement is referred to as security of the watermark and is usually
achieved by the use of cryptographic keys.
3) A watermark should stay in the host data regardless of whatever happens to the host
data, including all possible signal processing that may occur, and including all hostile
attacks that unauthorized parties may attempt. This requirement is referred to as
robustness of the watermark. It is a key requirement for copyright protection or
conditional access applications, but less important for applications where the watermarks
are not required to be cryptographically secure, for example, for applications where
watermarks convey public information.
4) A watermark should, though being unremovable, be imperceptible.
In the following, a few of the mentioned requirements and the resulting design issues are
highlighted in more detail.
1) Watermark Security and Keys: If security, i.e., secrecy of the embedded
information, is required, one or several secret and cryptographically secure keys have to
be used for the embedding and extraction process. For example, in many schemes,
pseudorandom signals are embedded as watermarks. In this case, the description and the
seed of the pseudorandom number generator may be used as key. There are two levels of
secrecy. In the first level, an unauthorized user can neither read nor decode an embedded
watermark nor can he detect if a given set of data contains a watermark. The second level
permits unauthorized users to detect if data are watermarked, however, the embedded
information cannot be read without having the secret key. Such schemes can, for
example, embed two watermarks, one with a public key and the other with a secret key.
Alternatively, a scheme has been proposed which combines one or several public keys
with a private key and embeds one combined public/private watermark, rather than
several watermarks. When designing an overall copyright protection system, issues like
secret key generation, distribution, and management (possibly by trusted third parties), as
well as other system integration aspects have to be considered.
2
2) Robustness: In the design of any watermarking scheme, watermark robustness is
typically one of the main issues, since robustness against data distortions introduced
through standard data processing and attacks is a major requirement. Standard data
processing includes all data manipulation and modification that the data might undergo in
the usual distribution chain, such as data editing, printing, enhancement, and format
conversion. “Attack” denotes data manipulation with the purpose of impairing,
destroying, or removing the embedded watermarks. Although it is possible to design
robust watermarking techniques, it should be noted that a watermark is only robust as
long as it is not public, which means as long as it cannot be read by everyone.
3) Imperceptibility: One of the main requirements for watermarking is the perceptual
transparency. The data embedding process should not introduce any perceptible artifacts
into the host data. On the other hand, for high robustness, it is desirable that the
watermark amplitude is as high as possible. Thus, the design of a watermarking method
always involves a trade-off between imperceptibility and robustness. It would be optimal
to embed a watermark just below the threshold of perception. However, this threshold is
difficult to determine for real-world image, video and audio signals.
4) Watermark Recovery With or Without the Original Data: Watermark recovery is
usually more robust if the original, unwatermarked data are available. Further, availability
of the original data set in the recovery process allows the detection and inversion of
distortions which change the data geometry. This helps, for example, if a watermarked
image has been rotated by an attacker. However, access to the original data is not possible
in all cases, for example, in applications such as data monitoring or tracking. For other
applications, like video watermarking, it may be impractical to use the original data
because of the large data volume, even if it is available. It is, however, possible to design
watermarking techniques that do not need the original for watermark extraction. Most
watermarking techniques perform some kind of modulation in which the original data set
is considered a distortion. If this distortion is known or can be modelled in the recovery
process, explicitly designed techniques allow its suppression without knowledge of the
original. In fact, most recent methods do not require the original for watermark recovery.
1.2 Basic Watermarking Principle
The basic idea in watermarking is to add a watermark signal to the host data to be
watermarked such that the watermark signal is unobtrusive and secure in the signal
3
mixture but can partly or fully be recovered from the signal mixture later on if the correct
cryptographically secure key needed for recovery is used.
Watermark embedding process is shown below in Fig.1.1
watermark
Digital
Data
Watermarking
Watermarked data
Secret or public key
Figure 1.1 watermark embedding process[5]
And the watermark extraction process is shown below in Fig.1.2
Watermark
and/or original image
Digital
image
Watermarked data
Watermarking
Watermark
image
Secret or public key
Figure 1.2 watermark extraction process[5]
Different types of Watermarking methodologies are given below
Figure 1.3 different types of watermarking methodologies[5]
4
1.3 Purpose of the Study
Here in this dissertation purpose is to apply different watermarking techniques for video
authentication for compressed and uncompressed video and to find PSNR,SSIM and BER
parameters for watermarked video to check authentication of video.
5
2. Video Watermarking
2.1 Introduction:
Video watermarking is loaded in the digital watermark on the video, which uses video
data redundancy and embeds the copyright information into the original video data,
thereby protecting the copyright of digital products or the legitimate interests of copyright
holders. In video watermarking, the movement of video properties and the defect of
human visual model make video watermarking technology fall behind the image
watermark technology. Current video coding technologies (such as H. 264), some special
attacks, the blind video watermarking detection and real-time features, have brought new
challenges to video watermarking.
2.2 The principle of video watermarking:
A complete digital watermarking system should include three basic parts: watermark
generation, watermark embedding and watermark extraction or detection. Watermark
embedding algorithm uses the symmetric key or public key to make the watermark
information embed into the original carrier to get concealed carrier. Watermark detection
/ extraction algorithm using the corresponding key vector from the hidden watermark is
detected or recovered without the key. The attacker is very difficult to find and modify
the hidden watermark vector. Block diagram of watermark embedding and extraction is
shown in Figure 3:
Figure 2.1 principle of video watermarking[4]
6
2.3 Characteristics of video watermarking
Video watermarking not only has the characteristics of digital image watermarking, but
also has its unique characteristics.
High real-time. Three-dimensional video signal has more the amount of data than the
image does. So calculation quality is larger and embedding / detection needs more
times. The procession of embedding, using video compression standard for these
specific structures such as motion vector coding, VLC code word, etc., can achieve
efficient algorithms.
Random detection. The watermark is detected in any position of the video rather
than the position according to the video playback order to detect the watermark.
The combination of video codec standard. Video is compressed formats in the
storage or transmission procession, so information hiding technology research cannot
do anything without specific video codec standard. Only combined with the encoding
and decoding standard, video information hiding technology can make the real-time
requirements fulfilment.
Better robustness. Video watermarking scheme must ensure it can resist almost
kinds of processing or attacks.
Blind detection scheme. Non-blind detection needs the original host signal, but it is
very inconvenient to use the original data, because of the huge video data. Blind
detection does not need the original host signal.
2.4 Video watermarking model
According to the embedded strategy, video watermarks algorithm can be divided into
three types of solutions: the original uncompressed video, video codec and the
compressed video stream. The course of the solutions is show in the figure 2.2.
Figure 2.2 video watermarking model[4]
7
Solution 1: Non-compressed domain video watermarking. Watermark is directly
embedded into the original encoded video sequences and then the video containing
watermark image is encoded. Such programs can take advantages of still images
watermarking technology and combine with the structural characteristics of video frames
to form a solution for video watermarking. The advantage of this watermarking algorithm
is relatively mature. Many still image watermarking methods, such as spread spectrum,
the human vision model, image adaptive watermarking irreversible, synchronous
detection mechanism, etc, can be applied to this video watermarking system. But the
solution also has obvious disadvantages: it will increase the video bit rate data streams
and effect video rate of constancy; After compressing and encoding the video data, the
watermark may be lost; For compressed video, we need to decode, and then re-encoding
after embedding watermarking. This process increases the complexity of the calculation
and reduces the video quality.
Solution 2: Watermark embedding and detection module are introduced in the encoder.
Today's video compression standards include ISO / IEC of MPEG-1, MPEG-2, MPEG-4
and ITU-T of H. 261, H. 263, etc. Their basic ideas are motion compensation prediction
coding and block-based transforms coding. The watermark is embedded in the video
encoding phase. We can use the characteristics of encoded data and the principle of video
data compression (such as transformation to the spatial redundancy, quantization and
entropy coding, the motion compensation, motion estimation, etc). The process can be
simple to achieve watermark embedding and extracting real-time processing. This
solution, which is a relatively simple process of watermark embedded in the transform
domain coefficients, does not increase the bit rate video data stream; In addition, because
the watermark is embedded in the transform domain and combines with the encoding
process, we can design a watermarking algorithm against multiple attacks. But the
solution needs to modify the encoder and decoder and the video codec cannot perform
common watermark embedding and detection, so there is GOP error accumulation.
Solution 3: Embed watermark in the compressed domain, i.e. the watermark is directly
embedded into the compressed encoded bit stream. The advantages of this solution are no
decoding and re-encoding process, which will not cause a decline in video quality, while
the computational complexity is lower. The disadvantage is that the compressed bit-rate
constraints the size of the watermarked data. The strength of watermark embedded is
limited by the error of the video decoder and the embedding strategy is constrained by the
8
coding standard and the video compression algorithms. This algorithm takes advantage of
video compression coding standard without the compressed video stream re-encoding and
full decoding. It has a lower computational complexity and a higher rate of embedded
watermark. But it’s disadvantages are that the capable of deal channel interference is bad;
the watermark can be destroyed by adding random bits labeled the least bit amplitude in
the VLC code using the same algorithm; traditional filtering, resampling and time-domain
scaling processing will also effect the extraction of the watermark. Some algorithms are
also proposed to embed watermark in motion vector or in the range of larger and small
phase angle change motion vectors. In the compressed video sequences, most of the
frames are the frame motion compensation coding, so the motion vector hide the
watermark information can be more effective to use the information of video bit stream.
2.5 Video watermarking algorithm
There are a lot of embedded algorithms in the video watermarking flied. In my views, we
can divide them into two categories: embedded watermarks in the original video and
embedded watermarks in the compressed video stream. It is show in the figure 2.3.
DCT
Original
video
video
Compressed
video
Spatial
domain
DWT
Frequency
domain
DFT
DCT
coefficient
DHT
Motion
vector
SVD
Inter-frame
correlation
Figure 2.3 video watermarking algorithm[4]
2.5.1 Original video watermark
The original video watermarks are deal with the uncompressed video data and have
nothing to do with video code format. So there are two categories in this flied:
It can obtain the original video data. It can embed/extract watermarks into the
original video and the process is quite simple.
9
It can only get the code video data. So the encode video is decoded and then
embedded/extracted the watermarks.
a) Spatial domain watermarking:
Spatial domain watermarks are directly embedded into the original video data. The
embedded watermark signals are usually added to the luminance component, and
sometimes some color components are added to, or added to the color components in all.
The advantage is simple and has a lower complexity, but it cannot be perceived in the
robustness.
The main purpose of embedding or extracting watermarking on the pixel values directly
is to reduce the complexity of watermarking, but as the processor speeds continue to
increase, watermark embedding and extraction has become the most common method in
the frequency domain.
b) Frequency domain watermarking:
The watermarking in transform domain is that watermarking is embedded / extracted in
some transform domain of the original video. The common transforms are DCT, DFT,
DWT, DHT, etc. There are three corresponding methods: (1) Video stream is viewed as
three-dimensional signals, two dimensions in space, one in time. (2) Video stream is seen
as still image. The watermarks are embedded into the single frame image. (3) Video is
processed according to the blocks in transform domain. Most of methods are combined
with the video encoder.
Osama S. Faragallah[2], proposed an efficient DWT-based SVD digital video
watermarking method. The proposed method makes use of two powerful mathematical
transforms; DWT and SVD using the additive method. The two transforms were
combined and their mixtures exploit their attractive features such as frequency
localization of the DWT, and the compact capturing of the semi-global features and the
geometric information of images. The proposed DWT-based SVD video watermarking
method worked by applying two-level decomposition DWT to convert video frames into
bands of different frequency. Then, the high and middle frequency bands HH, LH and HL
are SVD transformed and a watermark is embedded in the diagonal matrix of each band.
The idea of the DWT is to represent a signal as a series of approximation and details.
Figure 2.4 shows the two band decomposition reconstruction wavelet filter bank. The
signal is lowpass filtered with H0(z) to give an approximation signal and highpass filtered
with H1(z) to give a detail signal. The basis function is chosen such that a perfect
reconstruction can be achieved.
10
G 0( z ) ( z 1)
1
G1( z ) ( z 1)
2
H 1( z ) ( z 1 1)
Figure 2.4 Two band decomposition-reconstruction wavelet filter bank[2]
2.5.2Compressed video watermarking
The MPEG coding standards are usually considered in the video watermarking. There are
three kinds of video in MPEG coding standards: I-frame, P- frame and b-frame. I frame
encoding is similar to JPEG, using intra-flied neighbouring pixels of space to compress
redundant information; P Frame need use the previous frame when it is encoding and
current frame can be used as a reference frame for prediction. B frame data compression
affection is obvious. It requires the previous frame and follow-up frame to prediction. B
frame itself cannot be predicted as the other reference frame. P frame and B frames are
adjacent frames using a time-domain redundancy to compress information, while the
prediction error signal can remove spatial redundancy. The process of removing spatialdomain redundancy uses DCT, quantization and entropy coding techniques and the
process of removing time-domain redundancy uses the motion compensation, motion
representation and motion estimation techniques.
In this three techniques are used:
a) Embed watermark in DCT coefficient
b) Embed watermark in motion vector
c) Watermarking using inter-frame correlation
11
3. Video tampering attacks
3.1 Introduction
There are several possible attacks that can be applied to alter the contents of a video data.
Video sequence can be viewed as a collection of consecutive frames with temporal
dependency, in a three dimensional plane. This is called the regional property of the video
sequences. When a malicious alteration is performed on a video sequence, it either attacks
on the contents of the video (i.e. visual information presented by the frames of the video),
or attacks on the temporal dependency between the frames. Therefore based on the
regional property of the video sequences, we can broadly classify the video tampering
attacks into three categories: spatial tampering attacks, temporal tampering attacks and
the combination of these two, spatio-temporal tampering attacks.
3.2 Spatial tampering
In spatial tampering malicious alterations are performed on the content of the frames (XY axis). The operations that can be done as tampering attack in spatial tampering are
cropping and replacement, morphing, content (object) adding and removing etc
Spatial
Tampering
Attacks
Object
Removal Attack
Attack On
Foreground
Object
Attack on
Background
Object
Object
Modification
Attack
Object Addition
Attack
Attack On
Foreground
Object
Attack on
Background
Object
Attack On
Foreground
Object
Attack on
Background
Object
Figure 3.1 different spatial tampering attacks[5]
3.2.1 Object removal attack
In object removal attack, the objects of the frames of the video are eliminated. This kind
of attack is commonly performed where a particular person wants to hide his/her presence
in a certain sequence of frames. This attack can be performed with both kinds of object,
foreground objects and background object.
12
Figure 3.2 object removal attack[5]
3.2.2 Object addition attack
When an object is inserted in a frame or in a set of frames then there is a kind of spatial
tampering attack: say Object addition attack. It can also be performed with both kinds of
objects, foreground objects and background objects.
Figure 3.3 object addition attack[5]
3.2.3 Object modification attack
In Object modification attack, an existing object of the frame(s) can be modified in such a
way that the original identity of that object is lost, and a new object may be in appearance
which is totally different from the original object. The object modification attacks can be
existed in many prospects in the given video. For instance, the size and shape of the
existing object may be changed, the colour of the object may be changed or it may be
discoloured, and with the help of additional effect the nature of the object and it’s relation
with other objects also may be changed.
13
3.3 Temporal tampering
In temporal tampering manipulation is performed on the sequence of frames. The focus is
on the temporal dependency. Temporal tampering attacks are mainly affecting the time
sequence of visual information, captured by video recording devices. The common
attacks in temporal tampering are frame addition, frame removal and frame reordering or
shuffling.
3.3.1Frame addition attack
In frame addition attack, additional frames from another video, which has the same
statistical properties, are intentionally inserted at some random locations in a given video.
Figure 3.4 frame addition attack[5]
3.3.2 Frame removal attack
In frame removal attack the frames of the given video are intentionally eliminated. In this
kind of attack frames or set of frames can be removed from a specific location to a fixed
location or can be removed from different locations. It depends upon the intention.
Commonly this kind of tampering attack is performed on surveillance video where an
intruder wants to remove his/her presence at all.
14
Figure 3.5 frame removal attack[5]
3.3.3 Frame shuffling attack
In frame shuffling attack, frames of a given video are shuffled or reordered in such a way
that the correct frame sequence is intermingled and wrong information is produced by the
video as compared to original recorded video.
Figure 3.6 frame shuffling attack[5]
15
4. Parameters to check authentication
4.1 PSNR (Peak Signal to Noise Ratio)
Equation to find mean PSNR for whole video sequence is given by below equation:
nr _ frames
PSNR
PSNR
i 1
nr _ frames
Where nr_frame = total no of frames
Where PSNR for one frame can be obtained by given formula:
2552
PSNR 10 * log(
)
MSE
y
x
( Ai , j Bi , j ) 2
MSE
x* y
i 1 j 1
Where x=width of frame/image
y=height of frame/image
4.2 BER (Bit Error Rate)
The mean decoding BER for the watermarked videos after they were attacked is given by
below formula:
BER
1 P
wout ( j ) win ( j )
P j 1
Where wout is the extracted watermark, win is the original watermark and P is the size of
the watermark.
4.3 NC (Normalized Correlation)
Normalized correlation is obtained by below formula:
NC ( win, wout )
w (i, j )w (i, j )
[w (i, j )]
i
j
i
in
j
out
in
2
Where win is the original watermark and wout is the extracted watermark.
16
5. Implementation and Planning
5.1 Algorithm for DWT and SVD based watermarking technique
1. Insert video in AVI format
2. Divide video into frames.
3. Take one frame from total frames obtained in step-2.
4. Apply 2-D DWT on the selected frame to obtain four levels HH,HL,LH,LL.
5. Take LL sub-band and apply SVD on that sub-band.
6. Take watermark image.
7. Apply 2-D DWT on that image.
8. Take LL sub-band and apply SVD on that sub-band.
9. Add results obtained from step-5 and step-8 using some scaling factor
10. Apply SVD on the result obtained in step-9.
11. Apply IDWT on that image to obtain watermarked image/frame.
5.2 Results obtained from DWT and SVD based watermarking
technique
Figure 5.1 Frame taken from AVI video sequence
17
LL band of image
LH band of image
HL band of image
HH band of image
Figure 5.2 2-D DWT applied on the selected frame to obtain four sub-bands
LL,LH,HL,HH
Figure 5.3 Original frame and watermarked frame obtained for proposed method
18
5.3 Work done till DP1
Study different watermarking techniques used for video authentication.
Use DWT-SVD based method on the video sequence to study watermarking.
19
References
[1] Dawen Xu, Rangding Wang, Jicheng Wang, “A novel watermarking scheme for
H.264/AVC video authentication ”, Signal Processing: Image Communication(Elsevier)
26 (2011)267–279
[2] Osama S. Faragallah, “Efficient video watermarking based on singular value
decomposition in the discrete wavelet transform domain”, Int. J. Electron. Commun.
(Elsevier) 67 (2013) 189– 196
[3] Po-Chyi Su, Chin-Song Wu, Ing-Fan Chen, Ching-Yu Wu, Ying-Chang Wu, “A
practical design of digital video watermarking in H.264/AVC for content authentication ”,
Signal Processing: Image Communication(Elsevier) 26 (2011)413–426
[4] Xing Chang, Weilin Wang, Jianyu Zhao, Li Zhang, “A Survey of Digital Video
Watermarking”, Seventh International Conference on Natural Computation 2011 IEEE
[5] Richa Singh, Mayank Vatsa, Sanjay K. Singh, Saurabh Upadhyay, “Integrating SVM
classification with SVD watermarking for intelligent video authentication”, Telecommun
Syst(Springer) (2009) 40: 5–15
[6] Hartung, F.; Kutter, M., "Multimedia watermarking techniques," Proceedings of the
IEEE , vol.87, no.7, pp.1079-1107, Jul 1999
20
