An Efficient Video Watermarking Using Color Histogram Analysis and Biplanes Image Arrays

R. Nishanth & Abin john joseph. Int. Journal of Engineering Research and Applications
www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 3, (Part - 6) March 2016, pp.43-48
www.ijera.com 43|P a g e
An Efficient Video Watermarking Using Color Histogram
Analysis and Biplanes Image Arrays
1
R. Nishanth, 2
Abin John Joseph
1
Assistant Professor/ECE Cochin University College of Engineering, Kuttanad
2
Assistant Professor/ECE Cochin University College of Engineering, Kuttanad
ABSTRACT
Video watermarking is a novel technology that has the ability to solve the problem of illegal digital video
manipulation and distribution. In video watermarking, the copyright bits are embedded into video bit streams.
This paper proposes an effective, robust and indiscernible video watermarking algorithm. A video can also
undergo several intentional attacks like frame dropping, averaging, cropping and median filtering and
unintentional attacks like addition of noise and compression which can compromise copyright information,
thereby denying the authentication. In this paper, the design and implementation of SVD and multiple bit plane
image based watermarking is proposed. The developed method embeds 8 bit-plane images, obtained from single
gray scale watermark image, into different frames of a video sequence. In this algorithm, some of the luminous
values in the video pictures are selected and divided into groups, and the watermark bits are embedded by
adjusting the relative relationship of the member in each group. A sufficient number of watermark bits will be
embedded into the video pictures without causing noticeable distortion. The watermark will be correctly
retrieved at the extraction stage, even after various types of video manipulation and other signal processing
attacks.
Keywords— video watermarking, SVD, bit-plane images
I. INTRODUCTION
Music, Images and Video are the digital data
that can be easily reproducible over information
networks and Vulnerable to various types of attacks.
Hence, there is an urgent need of copyright protection
for such forms of data. As a result of these
consequences, creators of digital data are diffident to
provide access to their own digital scholarly property.
Digital watermarking has been propositioned as a
means to identify the owner and dissemination path
of digital data. Creating Digital watermarks for digital
data address this issue by embedding owner
credentials candidly into the digital data.
When there is a problem with ownership, the
information can be extracted to entirely distinguish
the owner or distributor of the data. The main aim of
this paper is to hide the information in the video for
the security purpose.
A. Video Watermarking
As digital video-based solicitation
technologies grow, such as Internet video, wireless
video, videophones, and video conferencing, the
problem of illicit copying and circulation of digital
video rises more and more, thus crafting copyright
predicament for the multimedia industry in
customary, and to the audio-video industry in
specific. Many researches and technologies were
propositioned to offer procedures to elucidate the
problem of illegitimate copying and exploitations of
digital video. An appealing method that has been
suggested a decade ago to instigate copyright
information in multimedia documents is digital
watermarking.
Video watermarking methodologies can be
categorized into two principal categories based on the
technique of hiding watermark information bits in the
host video. The two principal categories are spatial
domain watermarking a transform-domain
watermarking. In spatial-domain watermarking
techniques, embedding and detection are executed on
spatial pixels values (luminance, chrominance, and
color space) or on the complete video frame. Spatial
domain techniques are tranquil to implement,
however they are not robust against customary digital
signal processing operations such as video
compression. Transform-domain techniques, on the
contrary, modify spatial pixel values of the host video
according to a pre-governed transform. Frequently
used transforms are the Discrete Cosine Transform
(DCT), the Fast Fourier Transform (FFT), the
Discrete Wavelet Transform (DWT), and the Singular
Value Decomposition (SVD). Transform-domain
watermarking techniques substantiated to be more
robust and imperceptible contrasted to spatial domain
techniques since diffuse the watermark in the special
domain of video frame, making it very arduous to
remove the embedded watermark. The paper
RESEARCH ARTICLE OPEN ACCESS

www.ijera.com
discusses two distinct blind, imperceptible and robust
video watermarking algorithms centered on Singular
Value Decomposition (SVD). Each algorithm embeds
the watermark in the transform-domain
II. PROPOSED METHOD
In the method proposed, a gray-scale image
is used as watermark signal. The gray-scale image is
first converted into multiple bit-plane images. These
multiple sets of bit-plane images are embedded in the
original video pictures. Only one set of a single bit-
plane image are used for embedding in all the video
frames of a single scene. When the scene change
occurs, another set of bit-plane image is used to
embed. So all the frames in a scene contain the same
watermark information and different scenes of a
video embed information of different watermarks.
The decomposition of gray scale image into 8 bit
plane image is shown in figure 1.
The first proposed algorithm for video
watermarking is based on transforming the host video
using the SVD operator and then embedding the
watermark information in the S, U, or V matrices in
diagonal-wise manner. The proposed algorithm
consists of two techniques, the former embeds the
watermark into the original video clip, whereas the
latter extracts it form the watermarked version of the
digital video clip. Foreground pixels are only
embedded into the watermark.
A. Watermark Embedding
The video watermarking algorithm proposed
will embed the each bit plane image of gray scale
watermark into the Y channel of each video picture of
same scene. Different bit-plane image of watermark
is embedded into different scene of the video.
The details of the watermark embedding processes
are the follows:
• Use the frame change detection algorithm and
find the different frames in a video.
• Get the luminous channel of the video picture
and divide it into non-overlapping 8×8 blocks.
In this, 7 triple groups of luminous values
are selected, and are used to embed the watermark.
These groups are presented in TABLE 1 and they are
marked from A to G in alphabetical order, and each
group contains 3 luminous values and the number
next to the letter represents the order in its group. Get
7 bits watermark from the binary watermark image, if
there are no 7 bits, and then get the rest. For each 8×8
blocks, embeds 1 bit watermark frame to each
selected luminous value in the order group A to group
G.
Suppose that group A is the current group,
and m is the current watermark bit, m can be 0 or 1.
When m=0, then A2 should be the absolute maximum
of group A, if A2 is not maximum, exchange the
value of A2 with the maximum of group A. When
m=1, then A2 should be the absolute minimum of
group A, if A2 is not minimum, exchange the value
of A2 with the minimum of group A.
Figure 1: Bit decomposition of 8-bit watermarking
image signal [P0 is the least-significant bit (LSB) and
P7 is the most-significant bit (MSB)].
TABLE 1: GROUP OF 8×8 LUMINOUS VALUES
After the 8×8 blocks has been embedded
watermark bits rebuilt the video picture, and
reconstruct the entire video using each embedded
video pictures.
A.Frame Change Detection
The color histogram of an image can be
computed by dividing a color space, e.g., RGB, into
discrete image colors called bins and counting the
number of pixels falling into each bin. The difference
between two images Ii and Ij based on their color
histograms Hi and Hj can be formulated as
The above relation denotes the difference in
the number of pixels of two image that fall into same

www.ijera.com
bin. In the RGB color space, the formula (1) can be
written as
Color histogram is used for detecting the changes in
the successive video frames. Each video picture
frame is converted into RGB image. For each RGB
image dRGB(Ii, Ij) is calculated using equation (2). The
average of dRGB(Ii, Ij) is compared with a predefined
threshold value. A value of dRGB(Ii, Ij) greater than
threshold is selected. If average of dRGB(Ii, Ij) is
greater than threshold, it means that a new scene has
appeared by replacing the previous scene [8]. When a
new scene is detected, a new watermark image is
selected from the list of bit-plane images for
embedding in all the host image frames. In this
process, an identical watermark is used for each
frame within a motionless scene. In order to make the
watermarking process more robust, the selection of
the bit-plane image is controlled in a specific order.
So only a predetermined detector can recover the
original watermarks and its sequences [2].
C. Watermark Extracting
Proposed system is a blind video watermarking. At
the extraction process original video is not required
only watermarked video is used. The extracting
process is as follows:
• Calculate the luminous value of each video frames.
• For each 8×8 blocks, extracting the watermark bit
from the luminous group in the alphabetical order as
TABLE I presents. The extracting algorithm is as
follows. Suppose current group is A. If the coefficient
of A2 is maximum than other two coefficients A1 and
A3, then 0 has to be extracted.
Else if the coefficient of A2 is minimum than other
two coefficients A1 and A3, then 1 has to be
extracted. Extracting process is continued until the
entire watermark bits have been extracted.
General overview of the watermarking method is
shown in figure 2.
Figure 2: General overview of the watermarking
method
III. SECOND PROPOSED SVD-BASED
VIDEO WATERMARKING
ALGORITHM
The second video watermarking algorithm is
centered on transforming the host video by means of
the SVD operator and then embedding the watermark
information in the U, or V matrices in a block-wise
manner. Foreground pixels are only embedded into
the watermark.
A. Watermark Embedding Procedure:
The embedding procedure of the second proposed
algorithm, with its three possible disparities, is
described in detail through the following steps:
Step 1: Split the video clip into video scenes Vsi.
Step 2: Sort out the frames of each video scene
utilizing SVD described in steps 3 ~ 8 below.
Step 3: Translate every video frame F from RGB to
YCbCr color matrix structure.
Step 4: Calculate the SVD for the Y matrix in each
frame F. This operation generates 3 matrices (U, S,
and V).
Step 5: Rescale the watermark image so that the size
of the watermark will counterpart the size of the
matrix which will be utilized for embedding U or V.
Embedding procedure in Matrix U Block-wise:
Y'= UY'SYVY
T
Embedding procedure in Matrix V Block-wise:
Y'= UY' SY VY
T
Where Y' is the updated luminance in the YCBCR
color representation.
Step 6: Translate the video frames F' from YCBCR
to RGB color matrix.
Step 7: Reconstruct frames into the conclusive
watermarked Video scene Vsi.
Step 8: Reconstruct watermarked scenes to get the
conclusive watermarked Video.
B. Watermark Extraction Procedure:
The algorithm is a blind watermarking
algorithm, and there by it does not require the original
video clip during extraction procedure. Hence
watermark image can be extracted from watermarked
video by employing the following steps:
Step 1: Split the watermarked Video clip V' into
watermarked scenes Vsi'.
Step 2: Route the watermarked frames of every
single watermarked video scene using SVD as
described in steps 3 ~ 8.
Step 3: Translate the video frame F' from RGB color
matrix to YCBCR.
Step 4: Calculate the SVD for the Y matrix in frame
F'. , this process generates 3 matrices (U, S, and V).
Step 5: Inverse the pixel value for 5 pixels in each
odd block used in the embedding process in the U
matrix if it is used in the embedding, or in the V

www.ijera.com
matrix, similar to x= (1/pixel value).
Step 6: Extract the embedded watermark from the
integer part of x.
Step 7: Build the image watermark WVsi by gushing
all watermark bits extracted from all frames.
Step 8: Reiterate the equivalent procedure for all
video scenes.
IV. EXPERIMENTAL RESULTS
In this experiment, we implemented the proposed
video watermarking algorithm under MATLAB 7.7
environment. The size of the uncompressed AVI
video used is 144x176 and 7 watermark bits can be
embedded into each 8×8 block, so the maximum
capacity of each video frame in the proposed
watermarking algorithm is 2772 watermark bits at
most. Therefore, gray scale image of size 49×56 is
chosen, and the watermark is shown in Figure 3. The
quality of the watermarked video is compared with
the quality of the original video by calculating the
PSNR value of the same video picture [3]. The
calculated PSNR of different videos are presented in
TABLE II. Images I and K where one of the images is
considered a noisy approximation of the other is
defined.
Figure 3: Original
gray scale
watermark
image
The PSNR is defined as:
Here, MAXI is the maximum possible pixel value of
the image. When the pixels are represented using 8
bits per three RGB values per pixel, the definition of
PSNR is the same except the MSE is the sum over all
squared value differences divided by image size and
by three different color space.
VIDEO AVERAGE PSNR
A 38.96db
B 42.47db
C 38.29db
Table II : Average PSNR Value Of Watermarked
Videos
To verify the similarity between the watermark W,
and the extracted watermark after addition of
noise is calculated using normalized correlation
coefficient [4].
NC =
A video sequence of 400 frames with 8 scenes is
used to test under frame dropping, noise addition and
temporal shift attack.To investigate the robustness of
the embedded watermark, different percentages of
frames are dropped from the video and PSNR value is
calculated for each extracted watermark.

www.ijera.com
Since the proposed method uses a series of
watermark signals, one for each successive individual
scene, with the repetition in a time-varying different
scene, this makes it more robust against attack by
frame dropping. In this case, all the frames in a scene
are watermarked using an identical signal. So the
watermark remains active in the rest of the frames
even after deleting all frames in a scene, leaving only
one frame. In order to remove a watermark signal, an
attacker must need to remove the whole frames in a
scene, that is, the complete scene, which would
degrade the video sequences significantly. The
watermark signals remain in all other scenes even
after completely removing a specific scene, which
only destroys a single watermark in a series.
In temporal shift attack, frames of a scene are
removed from a video clip or inserted into a sequence
so that the original scenes are shifted temporally. The
proposed scheme shows robustness against this type
of attack, since, here, a series of independent
watermark signals are used, one for each scene, being
repeated in a predetermined order. Keeping the
original watermark signal indices and comparing
them with the detected watermark signal indices, it
can be readily detected where the two scenes would
be lined up.
TABLE: PERFORMANCE COMPARISON TABLE
Fig: Estimated PSNRValue For Different Noise Density
V. CONCLUSION
With Multimedia becoming more and more
popular and readily accessible, the copyright and
ownership issues also assume significant importance.
The design and implementation of blind
watermarking algorithm for uncompressed video is
proposed. The algorithm successfully embeds bit
plane watermark bits into the luminous pixel value for
each video frames. Scene change detection algorithm
is used for detecting scenes in the video. In each
scene same bit plane image is embedded and different
scene contains different bit plane image. The
extraction process is blind and the watermark can be
extracted without any distortion from the
watermarked frame. Experimental results demonstrate
that the proposed technique is robust against attacks
such as frame dropping, temporal shifts and addition
of noise. Robustness of the scheme can be further
improved by combining video with audio
watermarks. This can considerably increase the
robustness of the watermark since most of the
specified attacks are on the video channel.
REFERENCES
[1]. Ye Dengpan, Dai Yuewei, Wang Zhiquan,
―A Survery of Video Watermarking‖.
Computer Engineering and Applications.
2005, 41(1),pp. 14 – 18
[2]. Hartung F, Girod B. ―Digital watermarking
of MPEG-2 coded video in the bitstream
domain‖ IEEE International Conference on
Acoustic, Speech, and Signal Processing,
Munich, Germany, 1997. 2621~2624.
[3]. Lu Jianfeng; Yang Zhenhua;Yang Fan; Li li,
―A MPEG2 Video Watermarking Algorithm
Based on DCT Domain‖, 2011 Workshop on
Digital Media and Digital Content
Management.
[4]. Satyen Biswas, Sunil R. Das, and Emil M.
Petriu, ―An Adaptive Compressed MPEG-2
Video Watermarking Scheme ‖, IEEE
transactions on instrumentation and
measurement, vol. 54, no. 5, october
2005,pp.1853-1861
[5]. ZHENG Zhen-dong, WANG Pei, LI Li, ―A
Hybrid Video Watermarking Scheme Based
on Motion Vectors and DCT,‖ Journal of
Image and Graphics.2009,14(12),pp.2631-
2634.
[6]. LIU Hong, LIU QUAN. ―Novel MPEG-2
Compression Domain-base Video Digital
Watermarking Algorithm,‖ Journal of
Wuhan University of
Technology.2008,30(7),pp.134-137.
[7]. Raghavendra K, Chetan K.R. ―A Blind and
Robust Watermarking Scheme with
Noise
Density
PSNR MSE NCC AD
0 4.8137
2.1464e
+004
1.6616
-
112.3201
0.1 8.0692
1.0029e
+004
1.0389 -52.8224
0.2 8.0830
1.0143e
+004
1.0391 -53.1698
0.5 8.1000
1.0071e
+004
1.0402 -53.1576
1.0 8.1182
1.0111e
+004
1.0397 -53.2310

www.ijera.com
Scrambled Watermark for Video
[8]. Authentication‖, International Conference on
Internet Multimedia Services Architecture
and Applications (IMSAA), 2009 .
[9]. Haitao Jiang, Abdelsalam (Sumi) Helal,
Ahmed K. Elmagarmid, Anupam
Joshi,‖Scene change detection techniques for
video database systems‖ Journal on
Multimedia systems. 1998.

An Efficient Video Watermarking Using Color Histogram Analysis and Biplanes Image Arrays

More Related Content

What's hot (19)

Similar to An Efficient Video Watermarking Using Color Histogram Analysis and Biplanes Image Arrays (20)

Recently uploaded (20)

An Efficient Video Watermarking Using Color Histogram Analysis and Biplanes Image Arrays