A Compressed Video Steganography using Random Embedding Scheme

Document Sample
A Compressed Video Steganography using Random Embedding Scheme Powered By Docstoc
					                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                      Vol. 8, No. 4, July 2010




   A Compressed Video Steganography using
        Random Embedding Scheme
               Sherly A P                             Sapna Sasidharan                          Amritha P P
    TIFAC CORE in Cyber Security               TIFAC CORE in Cyber Security              TIFAC CORE in Cyber Security
     Amrita Vishwa Vidyapeetham,                Amrita Vishwa Vidyapeetham                Amrita Vishwa Vidyapeetham
          Coimbatore, India                          Coimbatore, India                         Coimbatore, India
        sherlyram@gmail.com                        sapnapv@gmail.com                         ammuviju@gmail.com


Abstract−Steganography is the art of hiding while the                and have been considered widely in various fields
communication is taking place, by hiding information in              like covert communication, copyright protection, and
other information. Many different carrier file formats can           broadcast monitoring and military communication.
be used, images, videos, audios, image etc. This paper
                                                                     Steganography is the art of hiding information in
proposes a Compressed Video Steganographic Scheme. In
                                                                     such a way that no one can realize a hidden message
this scheme, data hiding operations are executed entirely in
                                                                     in the data except the sender and the intended
the compressed domain. Here data are embedded in the
macro blocks of I frame with maximum scene change. To                recipient. Steganography is also known as ‘covered
enlarge the capacity of the hidden secret information and to         writing’ which includes methods of transmitting
provide an imperceptible stego-image for human vision,               secret messages through inoffensive cover mediums
random embedding scheme (Pixel Value Differencing) is                in such a manner that the survival of the embedded
used. Decompression process is not required in this                  messages is undetectable. It can also be viewed as a
scheme.   Experimental    results   demonstrate   that   the         tradeoff between detectability, robustness, and bit
proposed algorithm has high imperceptibility and capacity.
                                                                     rate. Detectability is the apprehension of clandestine

    Keywords- Video Steganography; MPEG-4; PVD                       transmission and is often used in combination with
                                                                     encryption. It is robust to all types of processing such
              I.         INTRODUCTION                                as transformations, filtering, truncation, and scaling.
          The quick growth of the Internet and                       Finally, bit rate or the maximum amount of data that
multimedia communication systems in the past                         can be transmitted. This paper considers data
decade has enabled users to send digital data over                   embedding in videos. A video can be viewed as a
network suitably. However, transmission of data in                   sequence of still images and data embedding in
an open network is not secure, and data can be easily                images seems very similar to videos. However, there
tampered by illegal users. Consequently, shielding                   are many differences between data hiding in images
data during transmission is an important task.                       and videos, where the first important difference is the
Although cryptographic techniques can be used for                    size of the host media. Since videos contain more
this purpose, they are not secure enough because                     sample number of pixels or the number of transform
encryption can provide secure delivery of digital                    domain coefficients, a video has higher capacity than
content, but       when the content        is decrypted,             a still image and more data can be embedded in the
encryption no longer provides any security. To solve                 video. Also, there are some characteristics in videos
this problem, data hiding techniques were proposed



                                                               263                               http://sites.google.com/site/ijcsis/
                                                                                                 ISSN 1947-5500
                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                  Vol. 8, No. 4, July 2010




which cannot be found in images as perceptual                                   II.           ARCHITECTURE
redundancy in videos is due to their temporal
features. Here data hiding operations are executed
entirely in the compressed domain [1] [2]. On the
other hand, as a really higher amount of data must be
embedded in the case of video sequences, there is a
more demanding constraint on real-time effectiveness
of the system. Furthermore, with the development of
multimedia and stream media on the Internet,
transmitting video on the Internet will not incur
suspicion.       Image-based        and    video-based
steganographic techniques are mainly classified into                                      Fig.1: Block Diagram
spatial domain and frequency domain based methods.
The former embedding techniques are LSB, matrix,                           As shown in the Fig. 1, the architecture
embedding [5] etc. Two important parameters for                  consists of four functions: I frame extraction, the
evaluating the performance of a steganographic                   scene change detector, the data embedder. The first
system are capacity and imperceptibility. Capacity               section explains the extraction of I frames from
refers to the amount of data that can be hidden in the           MPEG video. In the second module, scene change
cover medium so that no perceptible distortion is                detector analyzes the frames with maximum scene
introduced.      Imperceptibility    or   transparency           change. I frames in MPEG standard is coded in intra
represents the invisibility of the hidden data in the            frame manner, we can obtain the DC picture with
cover media without degrading the perceptual quality             abstracting the DC coefficients from the DCT
by data embedding. Security is the other parameter in            coefficient codes. Eq.1 describes the compare method
the steganographic systems, which refers to an                   between two conjoint I frames
unauthorized person’s inability to detect hidden data.
                                                                             
                                                                                N
                                                                 H(Ii,Ii+1)=           (Hi(k )  Hi 1(k ))^2 /(Hi(k )  Hi 1(k ))^2
Previous work in data hiding field cared little about                           k 1


capacity and had low embedding capacity.                                                                th
                                                                 where I and, Ii+1 means the i and i+1th I frames, Hi
          In this paper,       we propose a secure               and Hi+1are histograms of DC pictures from the ith and
compressed video steganographic architecture taking              i+1th I frames. If the HD (Ii, Ii+1) is the peak value the
account of video statistical invisibility. This paper is         two I frames are from different scenes, therefore the
organized as follows: Section II describes the framework         scene change point is found. Also the variances var ( i)
of our video steganography system. In Section III, the           U of each DC picture from I frame will be calculated.
embedding mechanism is described in detail. We give the
                                                                 With the third module, data embedder, secret message
experimental results in Section IV. In Section V, a
                                                                 M is hidden into the compressed video sequence
conclusion is drawn finally.
                                                                 without bringing perceptive distortion. To increase the
                                                                 capacity of the hidden secret information and to
                                                                 provide an imperceptible stego-image for human




                                                           264                                    http://sites.google.com/site/ijcsis/
                                                                                                  ISSN 1947-5500
                                                  (IJCSIS) International Journal of Computer Science and Information Security,
                                                  Vol. 8, No. 4, July 2010




vision, here pixel-value differencing (PVD) is used for           message into the I-frames of an MPEG compressed
embedding.                                                        video sequence.


         III.     EMBEDDING MECHANISM                             B. Compressed Video Steganographic Algorithm

A. Embedding Position Selection
                                                                           Here a novel steganographic approach called
         Compressed      video     sequence     achieves
                                                                  pixel-value differencing algorithm (PVD) is used for
compression through the elimination of temporal,
                                                                  embedding. Images are more easily noticed by human
spatial and statistical redundancies with the use of
                                                                  eyes. In the PVD embedding method, the cover image
motion compensation, block quantization inside a
                                                                  (I frame) is simply divided into a number of non-
discrete cosine transform (DCT), and Huffman run-
                                                                  overlapping     two-pixel     blocks.      Each     block      is
level encoding. While selection of the embedding
                                                                  categorized according to the difference of the gray
block calculate the maximum scene change of each
                                                                  values of the two pixels in the block. A small
block of the conjoint I frames. Select the block with
                                                                  difference value indicates that the block is in a smooth
which maximum scene change occurs by using
                                                                  area and a large one indicates that it is in an edged
threshold value. Selection of proper color channel is
                                                                  area. The pixels in edged areas may, as mentioned
another issue in Video Steganography. The frames of
                                                                  previously, tolerate larger changes of pixel values than
video sequence is split into Y,Cb,Cr channels in the
                                                                  those in the smooth areas. So, in this method more
MPEG coding stage. According to different color
                                                                  data is embedded in edged areas than in the smooth
resampling rules, it is possible for the ratio Y:Cb:Cr to
                                                                  areas. And it is in this way that the changes in the
be set to 4:2:2 or 4:2:0. Under these the only
                                                                  resulting stego-image are kept unnoticeable.
unchanged channel is the Y channel. So here Y
channel is preferred as the host channel. In addition to
                                                                  C. Data Embedding and Extraction
the above selections, choosing inappropriate frame
type among I-frame, P-frame or B-frame for hiding                 1) Calculate the difference value di between two
message is also a crucial issue. Usually, a conventional          consecutive pixels pi and pi+1for each block in the
video consists of a number of GOPs. Each GOP is                   cover image. The value is given by di= gi+1− gi
composed of one I-frame and several B-frames and P-               2) Using di to locate a suitable Rk in the designed
frames. A typical I-frame adopts intra coding, which              range table, that is to compute j = min (uk − |di |)
means it does not refer to any other frames. Different            where uk>=di for all 1<=k<=n. Then Rj is the located
from an I-frame, a P-frame only refers to its nearest             range.
preceding I- or P-frame. As for a B-frame, it refers to           3) Compute the amount of secret data bits t that can
the nearest preceding and succeeding I-frame or P-                be embedded in each pair of two consecutive pixels
frame. In a conventional MPEG format, the content of              by Rj. The value t can be estimated from the width w
a B- or P-frame is the so-called residual error between           of Rj, this can be defined by t=log2wj
the current frame and the frame to which it refers.               4) Read t bits from the binary secret data and
Therefore, only an I-frame can hold complete                      transform the bit sequence into a decimal value b. For
information. In this paper, we choose to embed




                                                            265                               http://sites.google.com/site/ijcsis/
                                                                                              ISSN 1947-5500
                                                      (IJCSIS) International Journal of Computer Science and Information Security,
                                                      Vol. 8, No. 4, July 2010




instance, if bit sequence = 110, then the converted                   of wk Є {8, 8, 16, 32, 64, 128}. Here, PSNR value
value b = 6.                                                          is utilized to evaluate the invisibility of the stego-
5) Calculate the new difference value d to replace the                images. Table I lists the experimental results after
original difference                                                   the secret data are embedded using those two
                                                                      approaches. The hiding capacity (in bytes) and
                                                                      PSNR values achieved by the proposed scheme for I-
                                                                      frames are shown. The listed values are the average
                                                                      results after embedding 100 randomly generated bit
6) Modify the values of pi and pi+1 by the following
                                                                      sequences into the cover images. Two stego-images
formula ( gi’, g’i+1) = (gi − ceil (m), gi +1 + floor (m))
                                                                      are still h a r d l y o b s e r v e d t h a t the secret d a t a
if d is odd
                                                                      is hidden inside. This is because of the high
(gi’, g’i+1) = (gi − floor (m), gi+1+ ceil (m)) if d is
                                                                      variance existed in the pixel values of the I- frame.
even, where m = (d’ − d)/2i+1
                                                                      Therefore, this          demonstrates that the proposed
Repeat Step 1-6 until all secret data are embedded
into the cover image, then the stego-image is                         approach provides a promising performance in
obtained.                                                             increasing the capacity of the stego-images and
                                                                      maintaining           the        imperceptible           quality
During the phase of secret extraction, the original
designed range table is required. In the beginning, the               simultaneously. This table explains the capacity and

same method in the embedding phase is used to                         the PSNR value after embedding.
                                                                                                   Table I
partition the stego-image into pixel pairs (blocks).                                   PSNR and Capacity of stego I-frames
Then the difference value d for each pair of two
                                                                              Cover               Capacity                 PSNR
consecutive pixels pi* and p* the stego-image is
                                                                            I-frames              (bytes)                  (db)
calculated. Next, di* is used to locate the suitable Ri+1
                                                                                I1                 70235                   42.14
in Step 2 during the embedding phase. Therefore, b*
                                                                                I9                 71345                   41.5
is obtained by subtracting lj from di*. If the stego-
                                                                                I16                70356                   43.5
image is not altered, b* is equal to b. Finally, b* is
transformed from a decimal value into a binary
sequence with t bits, where t=log2wj


              IV.     EXPERIMENTAL RESULT


Several experiments were performed to evaluate
the performance of the proposed steganographic
algorithm

A. Capacity and PSNR                                                                  Fig. 4.1: I-frame before embedding
          The secret      binary data sequence S is
generated by pseudo-random           numbers.      We     set
the designed range table with the width in the set




                                                                266                                 http://sites.google.com/site/ijcsis/
                                                                                                    ISSN 1947-5500
                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                           Vol. 8, No. 4, July 2010




                                                                           [5] Y. J. Dai, L. H. Zhang and Y. X. Yang, “A New Method of
                                                                           MPEG     Video    Watermarking     Technology”,     International
                                                                           Conference on Communication Technology Proceedings (ICCT),
                                                                           2003.
                                                                           [6] G. C. Langelaar and R. L. Lagendijk, “Optimal Differential
                                                                           Energy Watermarking of DCT Encoded Images and Video”, IEEE
                                                                           Trans. on Image Processing, 2001, 10(1):148-158.



                Fig. 4.2: I-frame after embedding


                    V.          CONCLUSION


A       new     Video           Steganographic        Scheme
was proposed             in     this    paper,    operating
directly in compressed domain. For data hiding
pixel- value differencing               (PVD)       algorithm
has     been used. This algorithm provides high
capacity and imperceptible stego-image                    for
human vision of the hidden secret information.
Here I-frame with maximum s c e n e                   change
blocks        was        used     for    embedding.      The
performance of the steganographic algorithm is
studied and experimental results shows that this
scheme can be applied on compressed videos
with     no    noticeable         degradation    in    visual
quality.

                         REFERENCES

[1] F Hartung, B. Girod “Watermarking of Uncompressed and
Compressed Video”, Signal Processing, Special Issue on
Copyright Protection and Access Control for Multimedia Services ,
1998, 66 (3): 283-301.
[2] Bin Liu, Fenlin Liu, Chunfang Yang and Yifeng Sun, “Secure
Steganography in Compressed Video Bitstreams”, The Third
International Conference on Availability, Reliability and Security
,2008
[3] Y. K. Lee, L. H. Chen, “High capacity image steganographic
model,” IEE Proceedings on Vision, Image and Signal Processing,
Vol. 147, No.3, pp. 288-294, 2000.
[4] D.C. Wu, and W.H. Tsai, “A Steganographic Method for
Images by Pixel-Value Differencing,” Pattern Recognition Letters,
Vol. 24, pp. 1613–1626, 2003




                                                                     267                                http://sites.google.com/site/ijcsis/
                                                                                                        ISSN 1947-5500

				
DOCUMENT INFO
Description: The International Journal of Computer Science and Information Security is a monthly periodical on research articles in general computer science and information security which provides a distinctive technical perspective on novel technical research work, whether theoretical, applicable, or related to implementation. Target Audience: IT academics, university IT faculties; and business people concerned with computer science and security; industry IT departments; government departments; the financial industry; the mobile industry and the computing industry. Coverage includes: security infrastructures, network security: Internet security, content protection, cryptography, steganography and formal methods in information security; multimedia systems, software, information systems, intelligent systems, web services, data mining, wireless communication, networking and technologies, innovation technology and management. Thanks for your contributions in July 2010 issue and we are grateful to the reviewers for providing valuable comments. IJCSIS July 2010 Issue (Vol. 8, No. 4) has an acceptance rate of 36 %.