Conference PaperPDF Available

A Novel Video Steganography Algorithm in the Wavelet Domain Based on the KLT Tracking Algorithm and BCH Codes

Abstract and Figures

Recently, video steganography has become a popular option for a secret data communication. The performance of any steganography algorithm is based on the embedding efficiency, embedding payload, and robustness against attackers. In this paper, we propose a novel video steganography algorithm in the wavelet domain based on the KLT tracking algorithm and BCH codes. The proposed algorithm includes four different phases. First, the secret message is preprocessed, and BCH codes (n, k, t) are applied in order to produce an encoded message. Second, face detection and face tracking algorithms are applied on the cover videos in order to identify the facial regions of interest. Third, the process of embedding the encoded message into the high and middle frequency wavelet coefficients of all facial regions is performed. Forth, the process of extracting the secret message from the high and middle frequency wavelet coefficients for each RGB components of all facial regions is accomplished. Experimental results of the proposed video steganography algorithm have demonstrated a high embedding efficiency and a high embedding payload.
Content may be subject to copyright.
978-1-4577-1343-9/12/$26.00 ©2015 IEEE
A Novel Video Steganography Algorithm in the Wavelet
Domain Based on the KLT Tracking Algorithm and BCH Codes
Ramadhan J. Mstafa, IEEE Student Member
Department of Computer Science and Engineering
University of Bridgeport
Bridgeport, CT 06604, USA
rmstafa@my.bridgeport.edu
Khaled M. Elleithy, IEEE Senior Member
Department of Computer Science and Engineering
University of Bridgeport
Bridgeport, CT 06604, USA
elleithy@bridgeport.edu
Abstract—Recently, video steganography has become a popular
option for a secret data communication. The performance of any
steganography algorithm is based on the embedding efficiency,
embedding payload, and robustness against attackers. In this
paper, we propose a novel video steganography algorithm in the
wavelet domain based on the KLT tracking algorithm and BCH
codes. The proposed algorithm includes four different phases.
First, the secret message is preprocessed, and BCH codes (n, k, t)
are applied in order to produce an encoded message. Second, face
detection and face tracking algorithms are applied on the cover
videos in order to identify the facial regions of interest. Third, the
process of embedding the encoded message into the high and
middle frequency wavelet coefficients of all facial regions is
performed. Forth, the process of extracting the secret message
from the high and middle frequency w avelet coefficients for each
RGB components of all facial regions is accomplished.
Experimental results of the proposed video steganography
algorithm have demonstrated a high embedding efficiency and a
high embedding payload.
Keywords- Video steganography; DWT; BCH codes; face
detection; KLT tracking
I. INTRODUCTION
In today's world, people communicate over the Internet and
share private information. This secret information should be
protected through a secure technique that blocks the data from
intruders and hackers. Steganography is a technique that
protects any secret message from an unintended recipient’s
suspicion within any data form. However, many steganalytical
detectors have been invented that detect a secret message from
an unsecure steganography algorithm. In order to avoid the
secret data from being detected by steganalytical tools, the
steganography algorithm must be efficient. Every successful
steganography algorithm should contain an embedding
efficiency, an embedding payload, and robustness in order to
work against attackers.
The steganography algorithm containing a high embedding
efficiency will reduce a hacker’s suspicion of finding the
hidden data, and will be difficult to detect through steganalysis
detectors. In addition, accurate visual quality of the stego data
and a low modification rate of the cover data will improve the
embedding efficiency [1]. The embedding efficiency includes:
perceptual quality, complexity, and security. The embedding
efficiency is directly affected by the security of the
steganographic algorithm. However, if any obvious distortion
of the cover data after the embedding process occurs, then the
result may be an increase in the attention of the hackers [2].
Another component of a successful steganography algorithm is
a high embedding payload. The embedding payload is defined
as the amount of secret information that is required to be
embedded inside the cover data. Furthermore, an algorithm that
contains a high embedding payload will have an extensive
capacity to hide a secret message. In traditional steganographic
algorithms, an embedding efficiency and an embedding
payload are opposites. Increasing the capacity of the secret
message will decrease the visual quality of stego videos
resulting in a weakened embedding efficiency. Both factors
should be considered. The deciding factors depend on the
steganography algorithm and the user’s requirements [2, 3]. In
order to increase the embedding payload with the low
modification rate of the cover data, many steganography
algorithms have been proposed using alternative methods.
These algorithms use block codes and matrix encoding
principles such as BCH codes, Hamming codes, cyclic codes,
Reed-Solomon codes, and Reed-Muller codes [4].
The contribution of this paper introduces a novel video
steganography algorithm in the wavelet domain based on the
Kanade-Lucas-Tomasi tracking algorithm and BCH codes (n,
k, t). Our proposed steganography algorithm offers a reasonable
trade-off between the perceptual quality and an embedding
payload. The remainder of this paper is organized as follows:
Section 2 presents some related works. Section 3 explains BCH
codes. Section 4 discusses discrete wavelet transform. Section
5 reviews face detection and tracking algorithms. Section 6
presents the proposed steganography algorithm. Section 7
illustrates and explains the experimental results. Section 8 is
the conclusion.
II. RELATED WORK
Patel et al. proposed a new video steganography based on
the lazy wavelet transform method. Here, after the video stream
is separated into the frames, the lazy wavelet transform will be
applied to each frame to obtain four different sub-bands. The
main achievement of using the lazy wavelet method is to divide
the odd coefficients from the even coefficients. Then, the LSB
substitution will be used in each of the four sub-bands. Every
nine bits of the secret message will be embedded into the three
coefficients of a single pixel (each coefficient of Red, Green,
and Blue has 3 bits of message). The total length of the secret
message is embedded into the LSB coefficients of the audio
component. This scheme has used both visual frames and audio
components. The authors have claimed that the steganography
algorithm has a high embedding payload due to the usage all
video frames and the audio components. However, the lazy
wavelet is not a true mathematical wavelet. As a result, the
algorithm will not prevent attackers from locating the secret
message, because of the similarity to the spatial domain [5].
Zhang et al. proposed an efficient embedder using BCH codes
for steganography. The embedder conceals the secret message
into a block of cover data. The embedding process is completed
by changing various coefficients in the input block in order to
make the syndrome values null. The efficient embedder
improves both storage capacity and computational time
compared with other algorithms. According to the system
complexity, Zhang’s algorithm improves the system
complexity from exponential to linear [6]. The flexibility for
both embedding payload and efficiency in the previously
mentioned algorithms exists. This flexibility can still be used to
improve the capacity of a secret message even further.
III. BCH CODES
BCH are multiple-error-correcting codes which were
invented by Bose, Chaudhuri, and Hocquenghem. BCH codes
are a class of cyclic codes that are generated based on finite
fields. One of the key features of BCH codes is different from
the Hamming codes in number of correctable errors, in which
BCH codes can correct multiple bit errors. Another key feature
of BCH codes is the simplicity, in which these codes can be
decoded using syndrome decoding. A binary BCH codes (n, k,
t) can correct errors of a maximum t bits for codewords of the
length n ,,,…, and message length
k ,,,…,. Encoded codewords and messages can
both be interpreted as polynomials, where 
, and 
. When m and t are any positive integers where
3and   2, there will be a binary BCH codes
with the following properties:
Block codeword length 21
Message length
k
Maximum correctable error bits
t
Minimum distance  21
Parity check bits 
The inventors of BCH codes decided that the generator
polynomial will be the polynomial of the lowest degree in the
Galois field GF (2), with ,,,…, as roots on the
condition that is a primitive of GF2 . In the proposed
algorithm, the BCH codes is applied in order to encode the
secret message in the preprocessing phase. The parity-check
matrix H of the BCH codes is described as follows [4, 7]:

1  
1 
1 
. . . . .
. . . . .
. . . . .
1   
(1)
IV. DISCRETE WAVELET TRANSFORM (DWT)
DWT is a recognized method that transfers the signal from
the time domain to the transform domain at different frequency
bands with different resolutions [8]. DWT separates high,
middle, and low frequencies and their boundaries from one
another, while other methods, such as DCT, group the various
frequencies into estimated regions. DWT utilizes two groups of
functions, wavelet functions and scaling functions, which are
related with both low pass filter and high pass filter. The first
level of the 2D-DWT signal decomposition is applied to the
cover video frame. It analyzes the frame into four sub-bands,
LL (approximation), LH (horizontal), HL (vertical), and HH
(diagonal), using both a low pass filter and a high pass filter for
the decomposition process. LL is a low frequency sub-band,
which is an approximation of the original frame reduced to a
quarter of its size. The LH, HL, and HH sub-bands are middle
and high frequencies that contain detailed information about
any frame. In the second level of frame decomposition, the 2D-
DWT is applied to the LL sub-band, producing four new sub-
bands[9]. In the proposed algorithm, the LH, HL, and HH
coefficients of the first level are used as cover data to embed
the BCH encoded secret message. Fig. 1 shows the third level
of the 2D-DWT frame decomposition process.
Figure 1. Third Level of the 2D-DWT Frame Decomposition
V. FACE DETECTION AND TRACKING ALGORITHMS
In this section, we introduce how the facial region is
detected from the first video frame. Then, we explain how the
facial regions are tracked throughout the remaining video
frames. The facial region is detected in the first video frame by
the Viola-Jones algorithm which is one of the fastest and
strongest algorithms in object detection. The Viola-Jones
detector has three main contributions. The first contribution is
to introduce a new image representation, which is referred to an
integral image. The representation of the integral image can
compute the selected features much faster than other detectors.
The second contribution is to build a feature based classifier
utilizing an AdaBoost algorithm. The third contribution is to
combine many complex classifiers, which is called a cascade
structure. This cascade structure focuses on the important
regions of the frame that contain a facial object [10].
The process of facial detection in all video frames require a
high computation time. In addition, based on the classifier, the
detector will fail if a person moves fast or tilts his head.
Therefore, it is necessary to have an alternative algorithm
which tracks the face throughout the video frames. This
alternative algorithm is called a Kanade-Lucas-Tomasi (KLT)
tracking algorithm. The KLT algorithm operates by searching
through good feature points called Harris Corners in the facial
region from the first frame. These Harris Corners are tracked
throughout all the video frames. Each feature point will have a
corresponding point between two successive
v
displacement of the corresponding point pairs
as motion vectors. The process of tracking t
h
based on the movement of the centers of th
e
consecutive video frames. The following eq
u
process of face tracking across the video fram
e
R
R

C
C

C
1
|f
|f
i
C

1
|f

|f

i
Where R
and R

represent the face area
video frames, respectively. C
and C

are th
of features in two consecutive frames, re
s
f
and f

are the feature points in curr
e
frames, respectively [14]. Fig. 2 shows the pr
o
the facial region in the first video frame, a
n
facial regions throughout the remaining vide
o
both Viola-Jones and KLT algorithms, respect
i
Figure 2. Face detection and tracking using Viola-Jones
a
VI. THE PROPOSED STEGANOGRAPHY A
L
In this section, we introduce a novel vid
e
algorithm in the wavelet domain based on t
h
algorithm and BCH codes (15, 11, 1).
steganography is divided into the following fo
u
A. Secret Message Preprocessing Phase
In this work, a large size text file has bee
n
message, and it is preprocessed before the e
m
Here, the whole characters in the text file a
r
ASCII codes in order to generate an array of
b
for security purposes, the binary array is enc
r
key (Key
1
) that represents the size of the sec
r
p
rocess will encode the message and protect
i
Since the binary linear block of BCH code
s
used, the encrypted array is divided into 11-
b
every block is encoded by the BCH codes (15,
15-
b
it blocks. The size of the message is ex
t
four parity bits into each block. Another key
to generate randomized 15-
b
it numbers, an
d
XORed with the 15-
b
it encoded block. Th
e
p
roposed algorithm will be improved by usin
g
codes, and XOR operation.
B. Face Detection and Tracking Phase
As we previously mentioned, the process
facial regions in the video frames must be
v
ideo frames. The
can be computed
h
e facial region is
e
features in two
u
ations show the
e
s [11-13]:
2
3
4
s in two adjacent
e
position centers
s
pectively. Also,
e
nt and previous
o
cess of detecting
nd then tracking
o
frames by using
i
vely.
a
nd KLT algorithms.
L
GORITHM
e
o steganography
h
e KLT tracking
Our proposed
u
r phases:
n
used as a secret
mbedding phase.
r
e converted into
b
inary bits. Then,
r
ypted by using a
r
et message. This
i
t from attackers.
s
(15, 11, 1) are
b
it blocks. Then,
11, 1) producing
t
ended by adding
(Key
2
) is utilized
d
each number is
e
security of the
g
two keys, BCH
of extracting the
identified when
facial regions are used as cover dat
a
algorithm is applied in order to de
t
first video frame. Then, the KLT
throughout the remaining video fr
a
facial regions.
C. Data Embedding Phase
In each video frame, the co
v
algorithm is the facial region of int
e
be extracted from the video frames
tracked. The region of interest chan
g
the size of the facial bounding bo
x
2D-DWT is applied on each R, G,
the facial region producing LL, L
H
Then, the secret message is embe
d
HH coefficients of the facial regio
n
this stage is completed, the secret k
e
hidden into the non-facial areas.
b
ounding box in each frame are
e
non-facial area known by both tr
a
length of each facial box is 80 bits
X-axis and Y-axis). Moreover, in
o
receiver party, both keys will be e
m
region of the first video frame.
U
frames will be reconstructed in orde
format that transmits via the co
m
receiver party. Fig. 3 illustrates
t
embedding phase.
D. Data Extraction Phase
The process of the data extracti
n
4. In order to retrieve a secret m
video is divided into frames throug
h
are extracted from the non-facial
addition, in each video frame, the f
o
box are extracted from the non-fa
c
facial region is identified in each
v
DWT is applied on each of the R,
G
of the facial box in order to genera
t
sub-
b
ands. Then, the process of ex
t
is conducted by taking out the sec
and HH coefficients of each RGB
c
region. The extracted bits from all
v
binary array. The binary array is di
v
Each block will be XORed with t
h
generated by Key
2
. The results of t
h
using the BCH (15, 11, 1) decode
r
Since the sender has encrypted the
s
array is decrypted using Key
1
. The
f
8-bit code (ASCII) in order to gen
e
the original message.
VII. EXPERIMENTAL RESUL
T
A dataset of three different vi
d
Video3) with the format of Audio
V
used. The implemented video st
r
objects which are taken by a lapt
o
videos were captured with almost t
h
different people in front of a stable
c
a
. The Viola-Jones detector
t
ect the facial region in the
tracking algorithm is used
a
mes in order to track the
v
er data of the proposed
e
rest. The facial regions will
after they are detected and
g
es in every frame based on
x
. In each video frame, the
and B color components of
H
, HL, and HH sub-bands.
d
ded into the LH, HL, and
n
in all video frames. Once
e
ys and facial box edges are
Every four edges of the
e
mbedded into the specific
a
nsmitter and receiver. The
per frame (40 bits for each
rder to transmit keys to the
m
bedded into the no
n
-facial
U
pon completion, the stego
r
to produce the stego video
m
munication channel to the
t
he block diagram of data
n
g phase is illustrated in Fig.
essage correctly, the stego
h
the receiver, and two keys
area of the first frame. In
o
ur edge points of the facial
c
ial region. Thus, the exact
v
ideo frame. Then, the 2D-
G
, and B color components
t
e the LL, LH, HL, and HH
t
racting the hidden message
r
et message from LH, HL,
c
olor channels of the facial
v
ideo frames are stored in a
v
ided into the 15-bit blocks.
h
e 15-bit number randomly
h
e 15-bit blocks are decoded
r
to produce 11-bit blocks.
s
ecret message, the obtained
f
inal array is divided into an
e
ra
t
e the right characters of
T
S AND ANALYSIS
eos (Video1, Video2, and
V
ideo Interleave (AVI) are
r
eams contain human face
o
p camera. In addition, all
h
e same movement of three
c
amera.
Figure 3. Block diagram for the data embedding phase.
Figure 4. Block diagram for the data extracting phase.
Experimental results are obtained by u
s
version of MATLAB software program. The
640x480 pixel resolution at 30 frames per s
e
rate of 8856 kbps. Each cover video contains
3
video frames, the secret message appears as
split in accordance with the size of the facial r
e
A. Visual Quality
The visual quality of the proposed algorith
m
applying different metrics. The Peak Signa
l
(PSNR) is a non-
p
erceptual objective metri
difference between the original and the distor
t
is calculated as follows:
 10  


 
And Mean Square Error (MSE) is calculat
e
  ∑∑,,


,,


C and S refer to the cover frame a
n
respectively. In addition, m and n are d
resolutions, and h indicates the R, G, and
B
(k=1, 2, and 3).
The PSNR is not correlated with the
p
quality because the HVS has a nonlinea
r
PSNR_HVS (PSNRH) and the PSNR_H
V
metrics have improved the visual quality
o
PSNRH and PSNRM are perceptual subjectiv
e
b
ased on the frequency domain’s coeffic
i
PSNRM improves the quality of the stego v
i
PSNRH.
 10  





 10  



_

MSE

and MSE
_
are dependent o
n
coefficients of the original and distorted block
s
Fig. 5 shows the PSNR comparison of the
t
stego videos’ visual qualities are close to th
e
visual qualities due to the values of the P
N
Fig. 6 illustrates the PSNRH compariso
n
experiments. The PSNRH metric has imp
r
quality of all of the stego videos better than
t
The PSNRH values range between 60-70 dB
PSNRM comparison of the three experimen
t
obtained PSNRM values of the three videos
r
82 dBs. Fig. 8 summarizes the visual qu
a
between the PSNR, PSNRH, and PSNRM m
e
average of the three experiments. The PS
N
improved the visual quality of all three stego
v
the other two metrics. Overall, due to the hig
h
PSNRH, and PSNRM, the proposed algorit
h
visual qualities for stego videos. These visua
l
same as the original videos’ visual qualities.
s
ing the R2013a
videos contain a
e
cond, and a data
3
00 frames. In all
a large text file
e
gions.
m
is measured by
l
to Noise Ratio
c measuring the
t
ed videos. PSNR
5
e
d as follows:
6
n
d stego frame,
d
efined as video
B
color channels
p
erceptual visual
r
behavio
r
. The
V
S_M (PSNRM)
o
f stego videos.
e
metrics that are
i
ents. Moreover,
ideos better than
7
 8
n
the 8x8 DCT
s
[15, 16]
.
t
hree videos. The
e
original videos’
N
SR (41-50 dBs).
n
of the three
r
oved the visual
t
he PSNR metric.
s. In Fig. 7, the
t
s is shown. The
r
ange from 67 to
a
lity comparison
e
trics totaling the
N
RM metric has
v
ideos better than
h
values of PSNR,
h
m has excellent
l
qualities are the
Figure 5. The PSNR comparison
o
Figure 6. The PSNRH comparis
o
Figure 7. The PSNRM comparison
of
the three experiments.
o
n of the three videos.
of the three experiments.
Figure 8. The PSNR, PSNRH, and PSNRM comparison
fo
three experiments.
B. Embedding Payload
The proposed algorithm has a high em
b
according to the reference [17]. The average
o
ratios for three videos is 4.4%. Moreover, t
h
embedded secret message in each Video1, Vi
d
experiments is 12.39, 12.31, and 12.43 Mega
b
The hiding ratio can be calculated as follows:
     
    3    
Fig. 9 illustrates the average of the data e
m
of the three experiments for the propose
d
algorithm.
Figure 9. Average of the data embedding payload of th
e
VIII. CONCLUSION
In this paper, we proposed a novel vid
e
algorithm in the wavelet domain based on t
h
algorithm and BCH codes. The proposed al
g
different phases. The first phase is the
preprocessing phase. In this phase, a se
c
f
o
r
the average of the
b
edding payload
o
f obtained hiding
h
e amount of the
d
eo2, and Video3
b
its, respectivel
y
.
100% 9
m
bedding payload
d
steganography
e
three experiments
e
o steganography
h
e KLT tracking
g
orithm has four
secret message
c
ret message is
converted to the binary stream. T
h
and BCH codes are sequentially a
p
The second phase is called the f
a
algorithms phase. In this phase, t
h
video frame is detected by using
Then, the facial region is tracked
video frames using the KLT algori
t
data embedding process
p
hase.
embedding of a secret message int
o
frames. In each video frame, the 2
of the RGB color components of
t
LL, LH, HL, and HH sub-
b
ands.
hidden into the LH, HL, and H
H
regions. The final phase involves t
h
hidden message from LH, HL, an
d
each of the RGB components f
o
p
erformance of the proposed algo
r
series of experiments. Experimenta
that the proposed algorithm has a
and high embedding payload. Our
robustness of our proposed algorit
h
such as video processing attacks an
d
REFERE
N
[1] H. Yuh-Ming and J. Pei-Wun,
schemes," in
I
mage and Signal
International Congress on, 2011,
p
[2] R. J. Mstafa and K. M. Ellei
t
steganography using Hamming
Applications and Technology Co
n
Long Island, 2014, pp. 1-6.
[3] R. J. Mstafa and C. Bach, "Infor
m
Steganography Techniques,"
i
E
ngineering Education Northeast
S
[4] R. J. Mstafa and K. M. Elleit
h
Steganography Algorithm in D
W
Codes (15, 11)," in Wireless Te
(WTS), 2015, 2015, pp. 1-8.
[5] K. Patel, et al., "Lazy Wavelet Tr
in Video," in Communication Syst
e
(CSNT), 2013 International Confe
r
[6] Z. Rongyue, et al., "An Efficient
E
Steganography,"
I
nformation Theo
58, pp. 7272-7279, 2012.
[7] Y. Hoyoung, et al., "Area-E
ff
Architecture for Long BCH Co
d
E
xpress Briefs, IEEE Transactio
n
2013.
[8] E. Prasad, "High Secure Image St
e
Chaotic Neural Network and Wav
e
Journal of Computer Science & N
e
[9] G. Prabakaran and R. Bhavani, "
A
steganography based on Discr
e
Computing, Electronics and Elec
t
2012 International Conference on,
[10] P. Viola and M. Jones, "Rapid o
b
cascade of simple features," in
C
R
ecognition, 2001. CVPR 2001.
P
Computer Society Conference on,
2
h
en, the encryption process
p
plied to the binar
y
stream.
a
ce detection and tracking
h
e facial region in the first
the Viola-Jones algorithm.
throughout the remaining
t
hm. The third phase is the
This phase requi
r
es the
o
the facial regions of video
D-DWT is applied to each
h
e facial region generating
Then, a secret message is
H
coefficients of all facial
h
e process of extracting the
d
HH DWT coefficients of
o
r all facial regions. The
r
i
t
hm is verified through a
l
results have demonstrated
high embedding efficienc
y
future work will verify the
h
m against various attacks
d
artificial noises.
N
CES
"Two improved data hiding
Processing (CISP), 2011 4th
p
p. 1784-1787.
t
hy, "A highly secure video
code (7, 4)," in Systems,
n
ference (LISAT), 2014 IEEE
m
ation Hiding in Images Using
i
n American Society for
S
ection Conferenc
e
, 2013.
h
y, "A High Payload Video
W
T Domain Based on BCH
lecommunications Symposium
a
nsform Based Steganography
e
ms and Network Technologies
r
ence on, 2013, pp. 497-500.
E
mbedder for BCH Coding for
ry, IEEE Transactions on, vol.
f
icient Multimode Encoding
d
es," Circuits and Systems II:
n
s on, vol. 60, pp. 872-876,
ganography based on Hopfield
e
let Transforms," International
e
twork Security, vol. 13, 2013.
A
modified secure digital image
e
te Wavelet Transform," in
t
rical Technologies (ICCEET),
2012, pp. 1096-1100.
b
ject detection using a boosted
C
omputer Vision and Pattern
P
roceedings of the 2001 IEEE
2
001, pp. I-511-I-518 vol.1.
[11] B. D. Lucas and T. Kanade, "An iterative image registration
technique with an application to stereo vision," in IJCAI, 1981,
pp. 674-679.
[12] C. Tomasi and T. Kanade, Detection and tracking of point
features: School of Computer Science, Carnegie Mellon Univ.
Pittsburgh, 1991.
[13] J. Shi and C. Tomasi, "Good features to track," in Computer
Vision and Pattern Recognition, 1994. Proceedings CVPR '94.,
1994 IEEE Computer Society Conference on, 1994, pp. 593-600.
[14] L. Guo-Shiang and T. Tung-Sheng, "A face tracking method
using feature point tracking," in Information Security and
Intelligence Control (ISIC), 2012 International Conference on,
2012, pp. 210-213.
[15] K. Egiazarian, et al., "New full-reference quality metrics based
on HVS," in CD-ROM proceedings of the second international
workshop on video processing and quality metrics, Scottsdale,
USA, 2006.
[16] N. Ponomarenko, et al., "On between-coefficient contrast
masking of DCT basis functions," in Proceedings of the Third
International Workshop on Video Processing and Quality
Metrics, 2007.
[17] L. Tse-Hua and A. H. Tewfik, "A novel high-capacity data-
embedding system," Image Processing, IEEE Transactions on,
vol. 15, pp. 2431-2440, 2006.
Biography:
Ramadhan J. Mstafa is originally
from Dohuk, Kurdistan Region, Iraq. He is
pursuing his PhD degree in Computer
Science and Engineering at the University
of Bridgeport, Bridgeport, Connecticut,
USA. He received his Bachelor’s degree in
Computer Science from the University of
Salahaddin, Erbil, Iraq. Mr. Mstafa
received his Master’s degree in Computer
Science from University of Duhok, Duhok, Iraq. He is IEEE Student
Member. His research areas of interest include image processing,
mobile communication, security, and steganography.
Dr. Elleithy is the Associate Vice
President of Graduate Studies and
Research at the University of
Bridgeport. He is a professor of
Computer Science and Engineering. He
has research interests are in the areas of
wireless sensor networks, mobile
communications, network security,
quantum computing, and formal
approaches for design and verification.
He has published more than three hundred research papers in
international journals and conferences in his areas of expertise. Dr.
Elleithy has more than 25 years of teaching experience. His teaching
evaluations are distinguished in all the universities he joined. He
supervised hundreds of senior projects, MS theses and Ph.D.
dissertations. He supervised several Ph.D. students. He developed
and introduced many new undergraduate/graduate courses. He also
developed new teaching / research laboratories in his area of
expertise. Dr. Elleithy is the editor or co-editor for 12 books by
Springer. He is a member of technical program committees of many
international conferences as recognition of his research qualifications.
He served as a guest editor for several International Journals. He was
the chairman for the International Conference on Industrial
Electronics, Technology & Automation, IETA 2001, 19-21
December 2001, Cairo – Egypt. Also, he is the General Chair of the
2005-2013 International Joint Conferences on Computer,
Information, and Systems Sciences, and Engineering virtual
conferences.
... The region of interest (ROI) is selected using object detection, and tracking techniques and embedding are done in the ROI for improving the robustness and security. A video steganography algorithm was proposed by Mstafa et al. (Mstafa & Elleithy, 2015b) based on Kanade-Lucas-Tomasi (KLT) (Tomasi & Detection, 1991) object tracking algorithm with DWT and BCH codes. The proposed scheme utilized the Viola-Jones algorithm for face detection and the KLT tracking algorithm for tracking through the video. ...
Article
Videos are nowadays the most frequent and easy mode of communication over the internet; the reason behind the growth is the accessibility to video processing software available on the internet. Video steganography is a field where the data is embedded in video keeping the visual quality of the video intact. This study presents a new video steganography scheme with a stable trade-off between robustness and imperceptibility using 2D-DWT (Discrete Wavelet Transform) based on object detection and tracking. The main contribution of this paper includes embedding of secret data in the moving objects after applying object detection for the video frames where the secret bits are embedded in middle frequency sub-bands after applying 2D-DWT. To highlight the effectiveness of the proposed scheme, experimental results are carried out both quantitatively and qualitatively where quantitative analysis is done using different metrics such as PSNR, SSIM, BER and qualitative analysis is done using visual results of the frames. The experimental results illustrated that the proposed approach outperforms existing techniques in terms of qualitative and quantitative evaluation with high imperceptibility and robustness against noise attack. Eventually, the scheme has also been tested against existing steganalysis techniques to ensure the security of the proposed scheme.
... Thus, Steganography technology is truly important in terms of information destiny of internet protection and privacy on open systems inclusive of the network which considered respectable when the faintness is high during secret data transmission while needing communication robustness [5]. Most of the existing methods using the Least Significant Bit (LSB) due to the redundant bits on the cover of the images embeds in the spatial area of the image with less effective in which it occurring clear misrepresentation [6,7]. ...
Article
Arabic language contains a multiple set of features which complete the process of embedding and extracting text from the compressed image. In specific, the Arabic language covers numerous textual styles and shapes of the Arabic letter. This paper investigated Arabic cursive steganography using complex edge detection techniques via compressed image, which comprises several characteristics (short, medium and Long sentence) as per the research interest. Sample of images from the Berkeley Segmentation Database (BSD) was utilized and compressed with a diverse number of bits per pixel through Least Significant Bit (LSB) technology. The method presented in this paper was evaluated based on five complex edge detectors (Roberts, Prewitt, Sobel, LoG, and Canny) via MATLAB. Canny edge detector has been demonstrated to be the most excellent solution when it is vital to perform superior edge discovery overcompressed image with little several facts, but Sobel appears to be better in term of the execution time for Long sentence contents.
... It is obvious that more modifications with a particular embedding algorithm raise the detection probability. Hence in the second stage, Syndrome-Trellis Codes (STC) [4,5,15], Wet Paper Codes (WPC) [4,7,18,19], and BCH codes [32] were introduced and applied to improve the embedding efficiency (number of embedded bits per modification [11]); this results not only in higher security, but also in improved imperceptibility. Based on the idea that MVs manipulations result in shifting the MVs from locally optimal to nonoptimal, "Reversion Based features" [6] and "AoSo features" [48] have been introduced. ...
Article
Full-text available
Despite all its irrefutable benefits, the development of steganography methods has sparked ever-increasing concerns over steganography abuse in recent decades. To prevent the inimical usage of steganography, steganalysis approaches have been introduced. Since motion vector manipulation leads to random and indirect changes in the statistics of videos, MV-based video steganography has been the center of attention in recent years. In this paper, we propose a 54-dimentional feature set exploiting spatio-temporal features of motion vectors to blindly detect MV-based stego videos. The idea behind the proposed features originates from two facts. First, there are strong dependencies among neighboring MVs due to utilizing rate-distortion optimization techniques and belonging to the same rigid object or static background. Accordingly, MV manipulation can leave important clues on the differences between each MV and the MVs belonging to the neighboring blocks. Second, a majority of MVs in original videos are locally optimal after decoding concerning the Lagrangian multiplier, notwithstanding the information loss during compression. Motion vector alteration during information embedding can affect these statistics that can be utilized for steganalysis. Experimental results have shown that our features’ performance far exceeds that of state-of-the-art steganalysis methods. This outstanding performance lies in the utilization of complementary spatio-temporal statistics affected by MV manipulation as well as feature dimensionality reduction applied to prevent overfitting. Moreover, unlike other existing MV-based steganalysis methods, our proposed features can be adjusted to various settings of the state-of-the-art video codec standards such as sub-pixel motion estimation and variable-block-size motion estimation.
... In order to enhance embedding efficiency (the number of confidential message bits embedded per unit modification [13]) and consequently acheive higher level of security, second-generation methods have applied error-correcting codes such as BCH codes [35], Wet Paper Codes (WPC) [7,10,19,20] or Syndrome-Trellis Codes (STC) [9,10,17,51]. The mentioned approaches consider the embedding procedure as a source coding problem. ...
Article
Full-text available
The basic requirement of a steganography approach is security against steganalysis attacks. In other words, a steganography method is reliable as long as it withstands all of the known steganalysis approaches. In order to preserve the security of a steganography method, the statistical features of the embedded and the original media must be as close as possible. To achieve this goal, in this paper steganography is applied by introducing the following contributions. Firstly, a new method is suggested to find the most impalpable embedded motion vector (MV). Also, a novel modification cost function with respect to the MVs’ intra-frame and inter-frame statistical differences before and after embedding is proposed for the syndrome-trellis coder. Furthermore, a pseudo-random generator is introduced for altering the arrangement of motion vectors which are used by the syndrome-trellis coder to improve its efficiency. Experimental results show that the proposed method is the most secure MV-based steganography scheme against the state-of-the-art video steganalysis methods as well as preserving other steganography measurements including imperceptibility and compression ratio. Moreover, the computational cost of the proposed scheme is far less than its main rival.
... These metrics are used to determine whether the secret messages are retrieved from the stego videos successfully or corrupted during the communication by comparing the concealed and extracted covert data. The BER and Sim are computed in the following formulas [59,62]: ...
Article
Full-text available
Steganography has been a great interest since long time ago. There are a lot of methods that have been widely used since long past. Recently, there has been a growing interest in the use of sparse representation in signal processing. Sparse representation can efficiently model signals in different applications to facilitate processing. Much of the previous work was focused on image and audio sparse representation for steganography. In this paper, a new steganography scheme based on video sparse representation (VSR) is proposed. To exploit proper dictionary, KSVD algorithm is applied to DCT coefficients of Y component related to video (cover) frames. Both I and Q components of video frames are used for secure message insertion. The aim is to hide secret messages into non-zero coefficients of sparse representation of DCT called, I and Q video frames. Several experiments are performed to evaluate the performance of the proposed algorithm, in case of some metrics such as pick signal to noise ratio (PSNR), the hiding ratio (HR), bit error rate (BER) and similarity (Sim) of secret message, and also runtime. The simulation results show that the proposed method exhibits appropriate invisibility and robustness.
Article
Full-text available
Recent developments in the speed of the Internet and information technology have made the rapid exchange of multimedia information possible. However, these developments in technology lead to violations of information security and private information. Digital steganography provides the ability to protect private information that has become essential in the current Internet age. Among all digital media, digital video has become of interest to many researchers due to its high capacity for hiding sensitive data. Numerous video steganography methods have recently been proposed to prevent secret data from being stolen. Nevertheless, these methods have multiple issues related to visual imperceptibly, robustness, and embedding capacity. To tackle these issues, this paper proposes a new approach to video steganography based on the corner point principle and LSBs algorithm. The proposed method first uses Shi-Tomasi algorithm to detect regions of corner points within the cover video frames. Then, it uses 4-LSBs algorithm to hide confidential data inside the identified corner points. Besides, before the embedding process, the proposed method encrypts confidential data using Arnold's cat map method to boost the security level. Experimental results revealed that the proposed method is highly secure and highly invisible, in addition to its satisfactory robustness against Salt & Pepper noise, Speckle noise, and Gaussian noise attacks, which has an average Structural Similarity Index (SSIM) of more than 0.81. Moreover, the results showed that the proposed method outperforms state-of-the-art methods in terms of visual imperceptibility, which offers excellent peak signal-to-noise ratio (PSNR) of average 60.7 dB, maintaining excellent embedding capacity. INDEX TERMS Arnold's Cat map, corner detector, embedding capacity, imperceptibility, robustness, security, video steganography.
Conference Paper
Full-text available
Video steganography has become a popular topic due to the significant growth of video data over the Internet. The performance of any steganography algorithm depends on two factors: embedding efficiency and embedding payload. In this paper, a high embedding payload of video steganography algorithm has been proposed based on the BCH coding. To improve the security of the algorithm, a secret message is first encoded by BCH(n,k,t) coding. Then, it is embedded into the discrete wavelet transform (DWT) coefficients of video frames. As the DWT middle and high frequency regions are considered to be less sensitive data, the secret message is embedded only into the middle and high frequency DWT coefficients. The proposed algorithm is tested under two types of videos that contain slow and fast motion objects. The results of the proposed algorithm are compared to both the Least Significant Bit (LSB) and [1] algorithms. The results demonstrate better performance for the proposed algorithm than for the others. The hiding ratio of the proposed algorithm is approximately 28%, which is evaluated as a high embedding payload with a minimal tradeoff of visual quality. The robustness of the proposed algorithm was tested under various attacks. The results were consistent.
Conference Paper
Full-text available
people are becoming more worried about information being hacked by attackers. Recently, many algorithms of steganography and data hiding have been proposed. Steganography is a process of embedding the secret information inside the host medium (text, audio, image and video). Concurrently, many of the powerful steganographic analysis software programs have been provided to unauthorized users to retrieve the valuable secret information that was embedded in the carrier files. Some steganography algorithms can be easily detected by steganalytical detectors because of the lack of security and embedding efficiency. In this paper, we propose a secure video steganography algorithm based on the principle of linear block code. Nine uncompressed video sequences are used as cover data and a binary image logo as a secret message. The pixels' positions of both cover videos and a secret message are randomly reordered by using a private key to improve the system's security. Then the secret message is encoded by applying Hamming code (7, 4) before the embedding process to make the message even more secure. The result of the encoded message will be added to random generated values by using XOR function. After these steps that make the message secure enough, it will be ready to be embedded into the cover video frames. In addition, the embedding area in each frame is randomly selected and it will be different from other frames to improve the steganography scheme's robustness. Furthermore, the algorithm has high embedding efficiency as demonstrated by the experimental results that we have obtained. Regarding the system's quality, the Pick Signal to Noise Ratio (PSNR) of stego videos are above 51 dB, which is close to the original video quality. The embedding payload is also acceptable, where in each video frame we can embed 16 Kbits and it can go up to 90 Kbits without noticeable degrading of the stego video's quality.
Article
Full-text available
In this paper we propose a simple and effective model of visual between-coefficient contrast masking of DCT basis functions based on a human visual system (HVS). The model operates with the values of DCT coefficients of 8x8 pixel block of an image. For each DCT coefficient of the block the model allows to calculate its maximal distortion that is not visible due to the between-coefficient masking. A modification of the PSNR is also described in this paper. The proposed metric, PSNR-HVS-M, takes into account the proposed model and the contrast sensitivity function (CSF). For efficiency analysis of the proposed model, a set of 18 test images with different effects of noise masking has been used. During experiments, 155 observers have sorted this set of test images in the order of their visual appearance comparing them to undistorted original. The new metric, PSNRHVS- M has outperformed other well-known reference based quality metrics and demonstrated high correlation with the results of subjective experiments (Spearman correlation is 0.984, Kendall correlation is 0.948).
Conference Paper
Full-text available
Image registration finds a variety of applications in computer vision. Unfortunately, traditional image registration techniques tend to be costly. We present a new image registration technique that makes use of the spatial intensity gradient of the images to find a good match using a type of Newton-Raphson iteration. Our technique is taster because it examines far fewer potential matches between the images than existing techniques Furthermore, this registration technique can be generalized to handle rotation, scaling and shearing. We show how our technique can be adapted tor use in a stereo vision system.
Article
Full-text available
In this paper, we present a novel data-embedding system with high embedding capacity. The embedding algorithm is based on the quantized projection embedding method with some enhancement to achieve high embedding rates. In particular, our system uses a random permutation of the columns of a Hadamard matrix as projection vectors and a fixed perceptual mask based on the JPEG default quantization table for the quantization step design. As a result, the data-embedding system achieves 1/167 (1 bit out of 167 raw image bits) to 1/84 hiding ratios with a BER of around 0.1% in the presence of JPEG compression attacks, while maintaining visual distortion at a minimum.
Article
Full-text available
No feature-based vision system can work unless good features can be identified and tracked from frame to frame. Although tracking itself is by and large a solved problem, selecting features that can be tracked well and correspond to physical points in the world is still hard. We propose a feature selection criterion that is optimal by construction because it is based on how the tracker works, and a feature monitoring method that can detect occlusions, disocclusions, and features that do not correspond to points in the world. These methods are based on a new tracking algorithm that extends previous Newton-Raphson style search methods to work under affine image transformations. We test performance with several simulations and experiments. 1 Introduction IEEE Conference on Computer Vision and Pattern Recognition (CVPR94) Seattle, June 1994 Is feature tracking a solved problem? The extensive studies of image correlation [4], [3], [15], [18], [7], [17] and sum-of-squared-difference (SSD...
Article
The factorization method described in this series of reports requires an algorithm to track the motion of features in an image stream. Given the small inter-frame displacement made possible by the factorization approach, the best tracking method turns out to be the one proposed by Lucas and Kanade in 1981. The method defines the measure of match between fixed-size feature windows in the past and current frame as the sum of squared intensity differences over the windows. The displacement is then defined as the one that minimizes this sum. For small motions, a linearization of the image intensities leads to a Newton-Raphson style minimization. In this report, after rederiving the method in a physically intuitive way, we answer the crucial question of how to choose the feature windows that are best suited for tracking. Our selection criterion is based directly on the definition of the tracking algorithm, and expresses how well a feature can be tracked. As a result, the criterion is optima...
Conference Paper
In the paper, we proposed a face tracking method using feature point tracking. The proposed face tracking method is composed of two parts: face detection and face tracking. To evaluate the proposed method, several videos containing different kinds of face movement are captured by using low-cost webcam. Experimental results show that our proposed method can track facial regions well. In addition, the detection rate of our face tracking is more than 91%. These results demonstrate that our proposed method can achieve face tracking in real-world noisy videos.
Lazy Wavelet Tr in Video
  • K Patel
K. Patel, et al., "Lazy Wavelet Tr in Video," in Communication Syste (CSNT), 2013 International Confer