You are on page 1of 5

Video steganography Technique in Uncompressed Raw Domain based on Transform

Domain and using SSVLCs

Sushil Kumar Azad


Department of Mathematics, Rajdhani College,
University of Delhi, New Delhi-110015
e-mail: skazad@rajdhani.du.in

Abstract— In this paper, we present a video steganographic


in uncompressed raw domain based on transform domain and I. VIDEO STEGANOGRAPHY TECHNIQUES IN RAW DOMAIN
using self-synchronizing variable length codes. The proposed Unlike the compressed video, raw video steganography
algorithm is tested on uncompressed video files using Matlab
techniques deal with the video as a sequence of frames with
R2015b. Experimental results have shown that Slantlet
transform (SLT), which retain higher percentage of energy after
the same format. First, digital video is converted into 31
compression as compared to the DWT approach, is a better frames as still images, and then each frame is individually
option than the other basic transforms. Also, no noticeable used as carrier data to conceal the hidden information. After
degradation in visual quality is observed the embedding process, all frames are merged together to
produce the stego video. Raw video steganography
Keywords— Video steganography; PSNR; T-codes; DWT; techniques operate in both spatial and transform domains
SLT; DCT In video steganography methods that operate in transform
domain, each video frame is individually transformed into
Introduction
frequency domain using DCT, DWT, and discrete Fourier
The purpose of Steganography is to conceal the message in transform (DFT) and the secret message is embedded
any one of the multimedia carriers such as image, audio and utilizing the low, middle, or high frequencies of the
video files in such a way that no one, apart from the sender transformed coefficients.
and intended recipient, suspects the existence of the message. The advantage of transform based steganography over spatial
In other words, it conceals the existence of data from hackers domain is that the transform domains based schemes improve
and attackers. the robustness against signal processing, noises, and
The work on reversible and irreversible image steganography compression.
techniques in the frequency domain using different discrete A Video steganographic method based on the lazy wavelet
transforms such as DCT, DWT, SLT (Slantlet Transform), transform (LWT) is presented by Patel et al. [14], where first
CTT (Contourlet Transform), and CWT’s (Complex Wavelet each video frame is divided into four sub-bands, separating
Transforms) is described by Sushil Kumar [8-10]. the odd and even coefficients and then the secret information
The three main requirements that any video steganography is embedded into the RGB LWT coefficients.
method must have are imperceptibility, hiding capacity, and [6-7] presented a robust and secure video steganography
robustness against attacks, that is, the main objective of the method in DWT-DCT domain based on MOT and ECC. They
steganography is to provide a good, robust and high observed the high embedding capacity with an average HR of
embedding capacity system of sending data securely and 3.40% and 3.46% for DWT and DCT domains, respectively.
safely to its intended destination. A lot of research work for Also an average PSNR of 49.01 and 48.67 dBs for DWT and
developing an efficient and effective steganographic scheme DCT domains are achieved.
is available, but there is still a lot to do.
Richard Apau [4] has observed that when video
After image steganography, researchers have found that
steganography is combined with Huffman code compression
video is a better option than image, as large amount of data
and asymmetric cryptography, a higher level of security,
can be hidden inside a video. Also, it is difficult for anyone
robustness and capacity are achieved
to observe the distortion in the video, due to the moving
ShengDun Hu, KinTak U [17] have proposed a novel Video
stream of sounds and images.
Steganography which can hide an uncompressed secret video
According to Basheer and Safiya [1], video steganography
stream in a host video stream with almost the same size..
hiding technique is the best since it overcame the capacity
A. Kolakalur, I. Kagalidis, and B. Vuksanovic [3] have
problem of image steganography and alteration problem of
described the algorithm to hide a “secret” color video
text steganography . Also, using video as a cover object didn’t
sequence within another color video sequence in a wavelet
overcome the capacity problem only, but it also enhanced the
transform.
security of the embedded data.
We propose the steganography in the uncompressed (raw)
Like image steganography, video based steganographic
video using transform domain and a self-synchronizing
techniques are also classified into spatial domain and
variable length codes known as T-codes [11]. Before
frequency domain based methods. However, based on
embedding the binary message, and we obtain encoded
Compression, we divide the Video steganogrphy schemes in
message using T-encoding. This not only increase the
two categories: (i) Compressed Domain , (ii) Uncompressed
embedding capacity, but also add another level of security to
Raw Domain. The problem of the Uncompressed Raw
our scheme. Then we try to hide a video stream in another
Domain is how to make the embedded message resist video
video stream with almost the same size. We have also
compression. But because the video basically exists in the
proposed a wavelet-like transform known as Slantlet
format of compression, the research of the latter is more
Transforms in place of DWT, which have already been
significative

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE


proved by researchers a better option to Wavelet transform
for image steganography [10]. In the next subsection we
summarize the latest trends in video coding to know the
importance of Video steganography.

II. WAVELET-LI KE TRANSFORMS


We know that the Wavelet transforms produces much less li
artifacts than the DCT and they are adopted in JPEG2000.
However, they do not represent 2D singularities effectively.
Also they do not work well in retaining the directional edges
in the image, and not sufficient in representing the contours
not horizontally or vertically.
The Slantlet transforms (SLTs) have been proposed as an Fig. 3. 2-level slantlet image of ‘lena.bmp’
improvement over the classical DWT, which have shown
better time localization [19].In comparison to DWT, SLT
filters are of shorter length and satisfy orthogonality and zero III. TRENDS IN VIDEO CODING
moments conditions. G. Panda et al [18] have shown that SLT With the higher coding efficiency of the
provides improves time localization than the DCT and DWT. H.264(AVI)/H.265(HEVC)/MPEG-4, one might think that
there is no further research required in video coding. But, it
is not true. The Joint Video Experts Team (JVET) was formed
in October 2017 to develop a new video standard that has
capabilities beyond HEVC. Looking at the recent trend of
online video services and competition where Amazon Prime,
HULU Plus, HBO, Crackle, Youtube, Netflix and others are
in a race, it appears that the area is more active than ever
before. Video experts from ITU-T VCEG and ISO/IEC
MPEG are actively working on the next-generation standard
VVC. To foster groundbreaking innovation in video and
image coding is the today’s demand.
Today, video stats account for 80 percent of global consumer
internet traffic.And as per survey predictions, by 2021 video will be
responsible for an unprecedented 85% of total internet consumption
worldwide (https://www.muvi.com/blogs/best-video-codec-for-
streaming.html). The characteristics of some of the most commonly
used Video Codecs are summarized in Table 1:
Table 1. Video Codecs
Video Codecs Design Benefits Applications
a) Motion- Distribution Streaming
H.264/AVC
compensation- of 50% market, Cable
Fig. 1 Two-scale SLT based data compression/reconstruction based video smaller files Broadcasting
scheme. compression as compared and Blu-Ray
standards to previous disk
generation distributions
codecs –
MPEG-2
and H.263.

b) - successor of - Delivers - Supports 8K


H.265/HEVC AVC 25% to 50% UHD
(High-Efficiency better resolution, but
Video Coding) -Advanced
video coding compression comparatively
layers, parallel efficiency delivers
processing with smaller files
Fig. 2. Two-scale SLT based data compression/reconstruction
tools, motion improved or that require
scheme.
vector same-level low
predictions, of video bandwidth for
and other quality; streaming
essential -
coding Step2. Obtain the text message or a message-video,i.e., the
extensions. video to be hidden. In case the message is video, then
2.1 Divide each frame of message-video into three
planes. For each plane
2.2 Apply DCT
2.3 Pick the coefficients in zig-zag pattern,
2.4 Convert the coefficients into binary form.
Step3. Encode the binary message using T-encoding to
c) AV1 an open, - delivers WebRTC
obtain the message to be hidden
royalty-free highest- and HTML5
video codec quality real- Web Video
time videos,
scalable to
any modern
device at
any
bandwidth.

d) VP9 - royalty-free - offers Every chrome


DCT
alternative to better video browser,
HEVC, quality at Android T-
DWT/SLT
developed by the same bit phones, and ENCOD
Google rate as even ING
HEVC and Google’s own
thus is video
highly streaming
effective for platform
delivering YouTube,
4K HD supports VP9
videos for seamless
online. delivery of
videos to end-
users.

IV. PROPOSED ALGORITHM


The proposed algorithm is an increased payload video
steganography algorithm in SLT domain based on SSVLC,
known as T-codes. In this algorithm, we use uncompressed Fig. 4 Embedding
video sequences based on the frames as still images. This
method is illustrated by using two phases: 1) Data embedding Step4. Select one of the frames of cover-video, and
phase and 2) Data extracting phase. 4.1 apply DWT/SLT and get one low and three high
We first extract all the frames from a cover video , and frequency subbands (cA, cH, cV, cD)
message video, respectively. If the cover video stream is C, 4.2 Divide the subband cH into three planes
and the message video stream is M, we assume that the frame Step5. Embed the secret message of step3 into one of the
length of C is longer than or equal to that of M. We convert planes obtained in step4.2, using a random key, k
each message frame into binary string by choosing zig zag Step6. Combine all planes with hidden messages, and apply
pattern and apply T-encoding to get encoded message frame inverse DWTSLT to/ obtain Stego-frame.
to embed in the cover stream using transform based Step7. Combine all the stego-frames and obtain the
steganography technique. Then, combining all the stego- Steganographic-video/Stego-video.
frames we create a stego video. A number of ways exist to
hide information in video. We have concentrated on Extraction Algorithm:
Frequency Domain method. The encoding and decoding This is just the reverse embedding process.
process can be described as follows: Step1. Extract all the frames from Stego-video
Step2. Select the frame as selected in Step4 of embedding
Embedding Algorithm: and apply DWT/SLT to obtain four subbands (cA,
Step1. Extract the frames from the cover- cH, cV, cD).
video.
Step3. Divide the cH into three planes, and using the random- 576
key,k, select the plane
Step4. Extract the secret message from these planes
KB,
Step5. Decode the secret message using T-decding. 12
Step6. In case the hidden message is video, convert the frame
decoded message into DCT/DTT coefficients and s,
arrange them in zig-zag pattern.
Step7. Obtain the secret frame on applying inverse DCT. 15 fps
Step8. Finally obtain the message-video by combining all
secret frames.

V. RESULTS AND ANALYSIS


We have presented a noval video steganography technique in Table III. The message frames vs extracted frames
uncompressed video streams, which uses uncompressed AVI
video file as the cover, based on transform domain and using Message frame extracted Details
SSVLCs. To measure the video quality, the average PSNR is message frame
used. Pod.avi,
146 KB,
The Table II shows the test cover frames and the
corresponding stego-frames obtained after emdbedding the 15 fps
message frame in it. We can see that there no obvious
distortion in stego-frames.

Table II. The conver video frames vs Stego video frames


Cover frame Stego frame details 726 KB
Quetza 30 fps
l.avi, 901
255 frames
KB, 15
fps
Counter.
avi
37 KB,
fp.avi, 8 fps
311
KB, 17 Rupee18
frames 5.mp4,
, 186 KB,
15 fps 30 fps

Crocus.
avi,
368 Table IV. PSNR of Proposed algorithms based SLT vs DWT
KB, Test Videos DWT SLT
12 Quetal.avi (cover) 27.0557 29.5112
frames and
15 fps Pod.avi(message)
Cartoo Crocus.avi(cover) and 27.47 28.346
fp.avi(message)
n.avi,
Earth.avi(cover) and 27.33 29.78
405,01 fp.avi(message)
2 KB, Cartoon.avi(cover)
28 fps and
counter.avi(message)
VI. CONCLUSION [7] R. J. Mstafa and K. M. Elleithy, ``Compressed and raw video
steganography techniques: A comprehensive survey and analysis,''
A new Video Steganographic Scheme based on Multimedia Tools Appl., pp. 1_38, 2016, doi: 10.1007/s11042-016-
uncompressed raw domain is proposed in this paper. Further 4055-1.
t-codes, self-synchronizing variable length codes are used in [8] Sushil Kumar, "Performance analysis of Image Steganography
Schemes based on different Transform Domains", International Journal
the pre-processing step to provided additional layer of of Management, Technology And Engineering (IJMTE), Volume 8,
security. For data hiding, Wavelet (DWT) and Wavelet-like Issue X, OCTOBER/2018, pp. 2810-2828
transforms (SLT) have been used [9] Sushil Kumar , “ Image Steganograpy based on Discrete Chebychev
This algorithm provides high capacity and imperceptible Transform”, International Conference on Strategy Innovation &
stego-image for human vision of the hidden secret Technology (ICSIT2014), Ansal University,Gurgaon,Haryana, 2014
information. The performance of the steganographic [10] Sushil Kumar nee’ Sushil Kumar Azad,
https://scholar.google.co.in/citations?user=RPVXo_QAAAAJ&hl=en
algorithm is studied and experimental results shows that this
[11] M. R. Titchener, “Generalised T-codes: extended construction
scheme can be applied on uncompressed videos (.avi) with no algorithm for self- synchronization codes”, IEE Proc. Commun., Vol.
noticeable degradation in visual quality. 143, No.3, pp. 122-128, 2006
[12] C Xu, X Ping, “A steganographic algorithm in uncompressed video
sequence based on difference between adjacent frames”, In: Fourth
REFERENCES International Conference on Image and Graphics (ICIG) 297–302,
2007.
[13] M Yang, N Bourbakis, “A high bitrate information hiding algorithm
[1] R Basheer & M.K Safiya “Video data hiding in selective pixels of
for digital video content under H.264/AVC compression”, In: 48th
forbidden zone using mapping function”, International Journal of
Midwest Symposium on Circuits and Systems 935–938, 2005.
Advanced Computer Technology (IJACT).ISSN:2319-7900, 2014.
[14] Patel, K., “Lazy Wavelet Transform Based Steganography in Video”,
[2] A. K. Bhaumik, M. Choi, , R. J. Robles , & M. O. Balitanas, “Data
International conference on communication systems and network
hiding in video”, International Journal of Database Theory and
technologies (CSNT) 2013.
Application, 2(2), 9-16, 2009.
[15] Mennatallah M. Sadek, Amal S. Khalifa, Mostafa G. M. Mostafa,
[3] A. Kolakalur, I. Kagalidis, and B. Vuksanovic, “Wavelet Based Color
“Video steganography: a comprehensive review”, Multimed Tools
Video Steganography”, IACSIT International Journal of Engineering
Appl, Springer Science+Business Media New York 2014, DOI
and Technology, Vol. 8, No. 3, June 2016
10.1007/s11042-014-1952-z
[4] Apau Richard,“ Enhancing data security using video steganography,
[16] A P Sherly, P P Amritha, “A Compressed Video Steganography using
RSA and Huffman code algorithms with LSB insertion “, M.Phil.
TPVD”, Int J of Database Manag Syst 2 (3). doi:5121/ijdms.2010.2307
thesis, Kwame Nkrumah University Of Science And Technology,
67, 2010.
Kumasi, Ghana, September, 2016
[17] ShengDun Hu, KinTak U, “A Novel Video Steganography based on
[5] Apau Richard, J. B. Hayfron-Acquah, Frimpong Twum, “A Modified
Non-uniform Rectangular Partition”, IEEE International Conference
High Capacity Video Steganography Technique Based On Spatial
on Computational Science and Engineering CSE/I-SPAN/IUCC 2011
Domain Method, Asymmetric Cryptography and Huffman Code
Algorithms”, Communications on Applied Electronics (CAE), Volume [18] G. Panda and S.K. Meher, “An efficient approach to signal
5, No.10, September 2016 compression using slantlet transform”, IETE Journal of
Research, Vol. 46, No. 5, September, pp. 299-307, 2000
[6] Ramadhan J. Mstafa, Khaled M. Elleithy, Eman Abdelfattah, “A
Robust and Secure Video SteganographyMethod in DWT-DCT [19] Ivan W. Selesnick, “The Slantlet Transform”, IEEE transactions on
Domains Based on Multiple Object Tracking and ECC”, Computer signal processing,Vol. 47, No. 5, May, pp. 1304-1312., 1998
Science & Information Technology, Sacred Heart University, IEEE
Access, May 2017.
oi:10.1109/ACCESS.2017.2691581(http://digitalcommons.sacredhear
t.edu/computersci_fac)

You might also like