Professional Documents
Culture Documents
Abstract— Steganography is a creative work of concealing in the image in a random or a pre-defined order. The
secret data or message within an ordinary, non-secret file. This scrambling order is important and acts like a private key.
helps to prevent detection of the existence secret data.
Steganography takes information hiding one step ahead than As explained in [9] and [18], Spatial domain techniques
data encryption. Data encryption hides the data, but directly deal with pixels of images in order to get the desired
Steganography hides the existence of the data. In this paper, we enhancement by altering them. These techniques work on the
combine both Data Steganography and Data Encryption, thus entire image uniformly, thus producing unacceptable results.
providing double layer security. The paper deals with Also, it denies selective enhancement of edges and other
Multimedia Steganography which enables working successfully specific details effectively. These shortcomings of Spatial
with video files, without obscuring the quality of the resultant Domain techniques can be dealt by Transform techniques by
video. The proposed technique deals with Steganography in ensuring secure and desirable results. Paper [19] explains
Transform domain, thus overcoming the drawbacks of the that Transform techniques are based on manipulation of the
spatial domain techniques. An important drawback of spatial orthogonal transform of the image. They process image
domain technique is that data is prone to different attacks, according to frequency content. Transform or Frequency
whereas Transform domain provides a robust aftermath. domain techniques are majorly based on computing 2-D
Applying the advantages of Transform domain, we have used discrete unitary transform such as DCT, DST, DWT, etc. on
DCT and DST technique on the cover video whilst using
the image. It consists of two components: Magnitude
Encryption algorithm of Scrambling- AES on the message
(comprises of frequency content) and Phase (used to restore
image, ensuring double layer security.
image into spatial domain). These techniques ensure
Keywords—Advanced Encryption Standard, Scrambling enhancement of edges and specific information of an image.
Encryption, Discrete Cosine Transform, Discrete Sine Transform In this paper, we focus on DCT and DST based
Steganography.
I. INTRODUCTION
II. LITERATURE SURVEY
Steganography is an age-old practice. Revolutionary war
spies writing in invisible ink or Da Vinci hiding secret A. Scrambling based Encryption:
meanings in paintings was referred to as Steganography. T. As explained in [3], Scrambling is a process of jumbling
Morkel, J.H.P. Eloff, M. S. Oliver in [1] state that, currently, the pixel values and then re-arranging the values in the
applying this in the digital world, data can be encoded using original form. This technique performs an acceptable
digital keys and decoded by the intended receiver with the encryption on RGB images. The scrambling is performed on
appropriate key. As explained by A. Shaw, S. Ballamudi in all the three planes. Thus, an initiative of being robust.
[2], Multimedia Steganography deals with video, audio and
images all together. It enables hiding multiple messages Steps to perform Image Scrambling understood by
behind video frames in a random scheme. P.Premaratne and M. Premaratne [3] are as follows:
In order to achieve privacy on another level, Step 1: Read the RGB message image
Steganography is combined with Encryption. Thus, if Step 2: Extract the dimensions of the image
Steganography fails, and the message is detected, Encryption
will make sure that the message is unintelligible to the Step 3: Decide on an order that is to be followed to
attacker. The paper [3] identifies Scrambling Encryption perform scrambling of the pixel values
technique previously explained and understood by Step 4: Extract the three planes of the RGB message
P.Premaratne, M. Premaratne. Scramble based encryption image and scramble the values in each plane according to the
changes the message image by changing the pixel positions order decided in the previous step
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.
Step 5: Recreate each of the RGB message image planes ሺଶାଵሻ௩గ
Cv(n)= (v) cos ൫ ଶே
൯
Step 6: Combine the three planes and obtain the
encrypted image. ξଵ
݂ ݒݎൌ Ͳ
(v)= ቐ ே (1)
Step 7: Decryption is performed simply using the ξଶ
݂ݒݎ ൌ ͳǡʹǡ Ǥ Ǥ ǡ ܰ െ ͳ
scrambling order. ே
DCT is a general orthogonal transform for image
This technique has a number of limitations: processing. It has a number of advantages over other
i. Vulnerable to attacks if the Scramble order is known transform techniques:
to the attacker i.High compression ratio
ii. Encrypted image may reveal some information if the
scramble order is not permuted appropriately ii.Less bit error rate
Description of the steps recognized in [6] in each round A standard DCT Steganography in [12-15] depends on
are as follows: the following characteristics:
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.
III. PROPOSED MODEL
(2)
A. Image Scrambling-AES Encryption:
By referring paper [3] based on Image Scrambling and
DST Steganography is performed as follows: paper [6] based on AES, a new technique that is a
i. Select the RGB message image and the RGB cover combination of the two is proposed. Image Scrambling
image Encryption technique is vulnerable to information theft.
Thus, combining the Image scrambling with AES will
ii. The cover image has to be a 2n image as DST works
overcome this disadvantage. In this procedure, initially the
only on 2n sized images.
three planes of the RGB message image are scrambled
iii. Convert the cover image into DST by applying the according to a Scramble order. The scrambled image then
formula: D * CoverImage * D’, where D is the DST undergoes AES encryption, where ‘n’ rounds are performed
matrix for n calculated using the above formula, D’ on the message image. The sub steps in each AES round are
is inverse DST explained previously.
iv. Replace the values in the HH frequency band of the
DST cover image by the values of the message
image
v. Perform inverse DST by applying the formula:
D’ * CoverImage * D
E. Diffie-Hellman Key exchange technique:
Articles [10, 11] state a technique that ensures secure
transmission of private keys between the sender and the
receiver. Most practical and reliable approach. Transfers keys
over a hostile network in a suspicious manner.
Consider the following example in terms of keys.
x The sender and receiver will have a common public
key that can be known to all.
x The sender performs some calculations using this
public key and its own private key and sends the
results to the receiver.
x The receiver performs some calculations using the
same public key and its own private key and
transfers results to the sender.
x Sender uses the newly obtained value and performs
calculations again using its own private key.
x Receiver does the same with its own private key.
x The final resultant value is then the actual agreed
upon private key between the sender and the
receiver.
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.
The proposed algorithm combines all the above- Transmission:
mentioned techniques to provide high security.
Step 15: The video is then sent to the receiver.
The model follows the following steps:
Step 16: The key for identifying random frames
Embedding: containing message image is then sent to the receiver using
Step 1: Select a cover Video behind which the message is Deffie-Hellman Key exchange
to be hidden
Extraction:
Step 2: Select the message image that is to be hidden
Step 17: On the receiver’s side, decryption is performed
Step 3: Extract each frame of the cover video and store by initially extracting the video frames
them in a cell.
Step 18: The key identifying the random frames with
Step 4: Extract the size of the frame images message is used for extracting the required frames.
Step 5: Crop the image in a 2n size because both DCT and Step 19: DCT/ DST is performed on each frame
DST can be performed only on 2n sized images. sequentially and then the message image part for each frame
Step 6: For each cropped frame, divide the image into 4 will be visible in the bottom right of the DCT/ DST image
parts and identify the bottom right part of the division. Step 20: These parts are then initially decrypted using the
Step 7: The bottom right part is selected because, as we AES decryption key and then again decrypted using the
are performing DCT and DST Steganography, the bottom Scrambling order as the key
right part contains High frequency components with minimal Step 21: Original message image part is obtained
information. Thus, this band ensures that a message can be
hidden without being suspected. Step 22: All the image parts are combined and the entire
message image is obtained successfully.
Step 8: Now, divide the message image into parts
depending on the size of the bottom right band. i.e. if the size IV. INFERENCES
of the bottom right part is 512*512, divide the message The proposed model enables efficient hiding of the
image into parts, each of size 512(Add padding if size is message image behind a cover video.
less).
The message image is divided into parts and each part is
Step 9: Now concentrating on the message image, encrypted twice and then hidden using DCT steganography.
perform Image Scrambling-AES encryption algorithm as This ensures multi layered security on the message to be
explained earlier on each part of the message image. shared.
Step 10: Obtain the encrypted message image parts.
TABLE I. KEY FEATURES OF THE MODEL
Step 11: Now, using a random function, choose the
frames in which the message parts are to be hidden, one part Entity Description
in each frame. Encryption Image Scrambling and Advanced
Step 12: For each selected frame and message parts, techniques Encryption Standard
perform the following steps- No. of keys 1 for AES encryption + 1 for Scrambling
used based encryption + 1 for selecting random
a) Convert the frame into DCT/ DST and obtain the
frames = 3 keys in total.
Transform image by using the formulae mentioned
above Steganography Discrete Cosine Transform and Discrete
technique Sine Transform where image is hidden in
b) Select the lower right part of the frame and replace bottom right part of cover
its values with the first part of the message image
divided by 255(for normalization) Key sensitivity Low for all
c) After storing the message parts in each frame, Image handling Handles all types of images- RGB, Black
convert the frames back to the Spatial Domain by and white, Grayscale
performing Inverse of the respective DCT/ DST Video format .avi format
d) The result is the DCT-Steganographed/ DST- after
Steganographed frames where the message image embedding
cannot be detected
Step 13: Now, combine all the frames sequentially and “Fig. 5” shows the RGB color image that is divided into
obtain the Steganographed Cover video. parts and then the parts are hidden behind the cover video
Step 14: The resultant video now contains the message frames. Number of parts are decided according the cover
parts that are initially encrypted using Scrambling-AES frames’ size. The message image is quite greater in size.
technique and then hidden behind the cover frame using
DCT/ DST technique
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.
“Fig.8” shows AES encryption performed over the
Scrambled-encrypted image. AES algorithms and rounds are
explained previously in this paper.
“Fig. 9” shows the ith frame of the cover video. The cover
video selected is a shorter one in order to reduce processing
time. The video has frame rate of 25 frames per second.
Fig. 10. Steganographed ith frame. Message part hidden using DCT
Fig. 7. Image scrambling based encrypted image part Steganography
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.
“Fig. 11” shows the steganographed ith frame of the V. EXPERIMENTAL RESULTS
video. The message part is hidden using DST steganography
as discussed above. There is negligible difference between TABLE II. EXPERIMENTAL OBSERVATIONS
the original and steganographed video. Thus, the video Results for message image of size 4000*2662 RGB image and
completely avoids suspiciousness.
cover video of size 3.87 MB, 7 seconds video, 25 FPS
Fig. 11. Steganographed ith frame. Message part hidden using DCT Table II is the evaluation of performance and execution
Steganography time for the embedding and extraction process of the
message image. Less difference in embedding and extraction
“Fig. 12” shows the DCT decrypted image which is prove the process execution under the same time frame.
obtained after reversing the hiding operations. The resultant Workbench 1 works on a system with clock speed of 2.5
image as compared with the original message image contains Gigahertz, whereas workbench 2 works on a system with
noise. But there is negligible data loss. clock speed of 3.5 Gigahertz. The difference in clock speed
of the systems affects model performance to a high extent.
“Fig. 13” shows the DST decrypted image which is Majority of the processing time is occupied by the AES
obtained after reversing the hiding operations. The resultant encryption technique since multiple rounds of encryption
image as compared with the original message image contains process are performed on each message part.
noise. There is more Data loss as compared to the DCT
decrypted image. TABLE III. PERFORMANCE EVALUATION FOR DECRYPTED MESSAGE
OBTAINED
Fig. 12. Message image obtained at receiver's end after DCT decryption.
The Mean Square Error: A performance evaluation
metric used to determine the error rate between original and
decrypted image. Should be as minimum as possible.
The PSNR: Computes the peak signal-to-noise ratio
between two images. Often used as a quality measurement
metric between the original and a compressed image. Should
be as high as possible
Structural Similarity Index (SSIM) is another metric
for measuring image quality. It should be near to 1 for best
results
Thus, considering the above measures, the DCT
Steganography is considered to be better than the DST
Steganography.
Fig. 13. Message image obtained at receiver's end after DST decryption. VI. CONCLUSION
The proposed model is a research based on data hiding.
The proposed model deals with hiding a message image
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.
behind a video, thus providing a secured data transfer [4] Champakamala. B. S, Padmini. K, Radhika. D. K Asst Professors,
between the sender and the receiver. Also, this message is "Least Significant Bit Algorithm For Image Steganography",
Department Of Tce, Ijact, Vol.3, No. 4.
first encrypted twice and then hidden behind the cover video. [5] Federal Information Processing Standards Publications (Fips 197),
The encryption is performed using Scrambling technique and "Advanced Encryption Standard (Aes) ", 26 Nov. 2001.
then performing AES encryption over the Scrambled image. [6] A. Ray, A. Potnis, P. Dwivedy, S. Soofi And U. Bhade, "Comparative
Thus, 2 keys are used: one for Scrambling based encryption Study Of Aes, Rsa, Genetic, Affine Transform With Xor Operation,
And Watermarking For Image Encryption," 2017 International
and one for AES encryption This overlapping of techniques Conference On Recent Innovations In Signal Processing And
ensures that even if the existence of data is known after Embedded Systems (Rise), Bhopal, 2017, Pp. 274-278.
attacking, the attacker will receive only the encrypted data [7] Khayam Sa. 2003,”The Discrete Cosine Transform (Dct): Theory
that is difficult to be predicted or recognized. And Application”, Information Theory And Coding, Seminar 1, Pp.
802–602, East Lansing, Mi: Department Of Electrical & Computer
The encrypted message image parts are then hidden Engineering (Ece).
behind the cover video frames. The frames which are [8] N. Ahmed, T. Natarajan And K. R. Rao, "Discrete Cosine
steganographed are chosen randomly using the inbuilt Transform," In Ieee Transactions On Computers, Vol. C-23, No. 1,
Pp. 90-93, Jan. 1974.
function in MATLAB. These random frames act as a key for [9] P. Vaishali, P. Bhatt, “Transform Domain Techniques For Image
knowing the steganographed frames. Steganography”, International Journal Of Innovative Research In
Electrical, Electronics, Instrumentation And Control Engineering,
Steganography is performed using Discrete Cosine Vol. 3, Special Issue 1, April 2015
Transform or Discrete Sine Transform based Steganography [10] Http://Www.Cse.Unt.Edu/~Tarau/Teaching/Pp/Numbertheoretical/Dif
techniques. DCT and DST both deal with hiding message fi E–Hellman%20key%20exchange.Pdf
image in the High frequency part of the cover image. The [11] S. Kallam, “Diƥe-Hellman:Key Exchange And Public Key
Cryptosystems”, 2015. [Online] Available
high frequency band of the cover image contains minimum : Http://Cs.Indstate.Edu/~Skallam/Doc.Pdf [Accessed- December
information about the respective image. 2018]
[12] Sharma, Sudhanshi & Kumar, Umesh. (2013). Review Of Transform
Diffie Hellman key exchange technique is used for Domain Techniques For Image Steganography.
transferring of the three keys between the sender and the 10.13140/Rg.2.1.4797.1928.
receiver. [13] M. Gunjal , J. Jha, ” Image Steganography Using Discrete Cosine
Transform (Dct) And Blowfish Algorithm”, Ce Dept, Ijctt, Vol. 11
VII. FUTURE WORK No. 4, May 2014.
[14] E. Walia, P.Jain, Navdeep, “An Analysis Of Lsb & Dct Based
The noise contained in the extracted image is planned to Steganography” , Global Journal Of Computer Science And
be removed and obtain more precise results. Other different Technology, Vol. 10, Issue 1, April 2010.
and more efficient techniques for steganography can be [15] S. Goel, A. Rana, M. Kaur, “A Dct-Based Robust Methodology For
Image Steganography”, Ijcst Vol. 4, Issue 3, July-Sept 2013.
performed. Reduction of processing time is a factor to be [16] P. Sangit, S. Shinde, “Steganography In Videos Using Unique Frame
worked on. Exchanging of keys has to be made more secure Selection Technique”, Ijace, Volume-3, Issue-9, Sept.-2015.
and efficient. [17] U. Sahu, S. Mitra, “A Secure Data Hiding Technique Using Video
Steganography”, International Journal Of Computer Science &
The video obtained after embedding is in .avi format, Communication Networks, Vol 5(5),348-357
supporting only 8-bit images to form a video. We plan on [18] C. Kavadia, A. Lodha, “A Review On Spatial & Transform Domain
improving the performance by using .mj2 video format, Digital Watermarking Techniques”, Ijarsc, Volume 4, No. 3, March
2013.
supporting 16-bit images for viewing and creation of the [19] S. Sharma, U. Kumar, “Review Of Transform Domain Techniques
video For Image Steganography”, Ijsr, Volume 4 Issue 5, May 2015
[20] P. Kanikar, R. Chaturvedi, V. Kashyap, “Image Steganography using
REFERENCES DCT, DST, Haar and Walsh Transform”, IJCA, Volume 65– No.17,
[1] T. Morkel, J.H.P. Eloff, M. S. Oliver, “An Overview Of Image March 2013
Steganography”, Icsarg [21] Madhukar B.N. and S. Jain, "A Duality Theorem for the Discrete
[2] Arnab Shaw, Savitha Bhallamudi, “Image Steganography”, 2015 Sine Transform - IV (DST - IV)," 2016 3rd International Conference
[3] P. Premaratne & M. Premaratne, "Key-Based Scrambling For Secure on Advanced Computing and Communication Systems (ICACCS),
Image Communication," In Emerging Intelligent Computing Coimbatore, 2016, pp. 1-6.
Technology And Applications, P. Gupta, D. Huang, P. Premaratne &
X. Zhang, Ed. Berlin: Springer, 2012, Pp.259-263.
Authorized licensed use limited to: Murdoch University. Downloaded on June 16,2020 at 10:45:52 UTC from IEEE Xplore. Restrictions apply.