Message Camouflage in an Image using EDGE-Based Steganography

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 03 | Mar 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 942
Message Camouflage in an Image using EDGE-Based Steganography
Akkimsetti Mohana Sai Chandra
Student, Dept of Electronics and Communication Engineering, School of SEEE, SASTRA University, Thanjavur,
Tamil Nadu, INDIA
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - In Modern day world, Data Security is considereda
major concern. Data can be transmitted in numerous ways
over the internet. So, the data needs to be secured betweenthe
sender and receiver. Steganography is one technique, that
hides information inside other mediainsuchawaythatno one
will notice. Generally, inSteganographyImagesareconsidered
as Cover media because they have high redundancy in
representation and most common used over the internet.
Images are popular for cover media as they have small size,
content adaptability, visual resiliencemakethemgoodcarrier
to transmit secret messages over the internet. In this paper,
the data is hidden in an image using Edge based
steganography where the data is hidden in the edge pixels of
the image. Canny edge algorithm is applied foredge detection.
This method considers the amount of data to be embedded as
an important factor on the selection of edges, i.e., more the
amount of data to be embedded, larger the use of weak edges
in the image for embedding
Key Words: Edge Steganography, Canny Edge detection,
Encoding, Decoding, Edge pixel, MATLAB, least
significant bit (LSB), XOR, Modulo Operation
1.INTRODUCTION
Encryption is a process thatencodesinformationormessage
or file so that it can be only be accessed by authorized
people. Encryption uses algorithms to encrypt data at the
transmitter using a key and then the receiver uses same or
different key (based on type of encryption used) to decrypt
the information. In an encryption technique, the secret
information or message, referred to as plaintext, is
encrypted using an encryption algorithm – a cipher–
generating ciphertext that can be read only if decrypted. On
the other hand, Steganography is the practice of hiding a
message within another message or a cover object [1]. In
computing/electronic contexts, a computer file, message,
image, or video is hidden within another file, message,
image, or video. Generally, In Steganography grayscale
images are used. A grayscale image is one in which the RGB
space colors are shades of grey. The reason for opting
grayscale scale image is that less information needs to be
provided for each pixel. In Gray images have red, green and
blue components have all equal intensity in RGB space, so
only a single intensity value is required rather than three
intensity values.
Security ofthesteganographyalgorithmdependson
the pixel selection. Noisy pixels are considered ideal as
modelling them is difficult. Edge pixels are noisybecausethe
intensity suddenly changes at edge. Edge Pixel is a pixel
where the intensity changes suddenly in an image [2].These
sharp variations make the edge pixel difficult to model.
Therefore, edge pixels are better choice for embedding the
data.
Fig 1. Edge Pixel in an image
1.1 Literature Review
As the technology increases the information exchange
became insecure. A breach during thisinformationexchange
can cause a disaster (when it comes to exchange between
military data or confidential data etc.). So, there is a need for
protecting the informationfromthirdparties.Thetool which
helps to protect stored informationandinformationtransfer
is Information Security [3]. Information security is crucial in
safeguarding, securing and maintain the integrity of
information. Steganography is an art of secure transmission
of information from transmitter to receiver by ensuringthat
no other one can reliably conclude on the secret
communication.
Information or secret message can be hidden into cover
medium or cover data through several steganographic
techniques. Steganographic techniques involves embedding
and extracting of information. Image-based steganographic
techniques can be classified as spatial domainandfrequency
(transform) domain [4].
Images are preferred cover medium for
steganographic techniques because of their content
adaptability, resilience, high redundancy and smaller size of
images make them good carrier to transmit secret messages
[5]. A secret message is generally embedded as, the bits of
encrypted message are embedded in pixels of the cover
image. LSB Steganography is the most used and renowned

Steganographic technique. LSB steganography is technique
in which the least significant bit of the pixels is used for
embedding the secret information. LSB steganography can
be further classified as LSB replacement and LSB matching
[7]. In LSB replacement the least significant bit of each pixel
of cover image is replaced by next bit of secret message,
whereas in LSB matching the pixel value of cover image is
randomly increased or decreased by 1 except at boundary
pixels if any mismatch occurs between least significant bit
and secret message [8]. Edge adaptive image steganography
is another technique based on LSB matching which
calculates the difference between two adjacent pixels. If the
difference is greater than the predefinesthresholdthenboth
pixels are marked as edge pixels, then one bit of data is
hidden in each of them. But the major limitation of this
technique is if the image is smooth then detection of edge
pixels will be difficult.
As mentioned earlier steganography can be
classified into spatial and transform domain, Adaptive
Steganography is a special case ofbothspatial andtransform
domain techniques. In the spatial domain the secret data is
hidden in pixels of cover image using LSBsteganography [6].
This technique is widely used due to its potentiality of
embedding secret information in an image with high
capacity. In transform domain (frequency domain), the
secret data is hidden in the transformed coefficients of the
cover image using Discrete Cosine Transform [9]. Edge
detection is a technique where the edges in an image are
found out by using a suitable algorithm. The edge which can
be defines as a set of pixels positions where a sudden or
abrupt change of intensity values occurs. Edges represent
boundaries betweenobjects andbackground.Edgedetection
involves method of segmenting an image into regions of
discontinuity. Some edge detection techniques are Sobel,
Prewitt, LoG, Robert, Canny etc. [11].
1.2 Classic Edge Detection Techniques
Sobel’s algorithm: When there is a change in the intensity
value of the image pixel, then the edge can detect. There are
several techniques to detect an edge. Sobel is one of the
former edge detection techniques,whichusesthegradientof
the image intensity to detect edge. In this algorithm or
technique, the image intensity gradient is taken at each and
every point in the image. This gives the magnitude and
direction of intensity of the image. Then by comparing the
resultant threshold values, the presence of an edge can be
determined. Sobel operator can be used to find the gradient
of the image intensity values [10]. Basically, Sobel operator
is a discrete differentiation operator. It is very simple and
time efficient computation. Smooth edges can be detected
very easily. But it is very sensitive to noise and not very
accurate in edge detection as it needs comparison with a
threshold. Thick and rough edges do not give appropriate
results.
Prewitt Algorithm: Prewitt algorithmissimilartothe Sobel
algorithm, but the difference is the source image point is
convolved with two 3x3 kernels vectors to obtain the
horizontal and vertical derivatives. By using the “EDGE”
function in MATLAB and the Prewitt algorithm results in a
binary image where bit 1 corresponds to edges and bit 0
corresponds to non-edges. Prewitt operator is the best
operator to detect the orientation of an image, but diagonal
direction points can’t be determined.
Roberts Edge Detection Algorithm: This technique also
detects the edges by computing intensity gradient just like
Sobel and Prewitt. Besides, this algorithm simplifies the
complexity as it uses a simple 2×2 kernel vector for gradient
computation. In addition to this, every pixel gradient
intensity is compared only with that of those diagonally
adjacent pixels. So, its main disadvantage is that sinceituses
such a small kernel, it is very sensitive to noise. Diagonal
direction can be preserved by this technique
Laplacian of Gaussian Algorithm (log) andZeroCrossing
Edge detectionAlgorithm: LaplacianofGaussianAlgorithm
(log) is different from the Sobel, Prewitt as it does not use
the gradient of intensity values(liketheother edgedetection
techniques), instead it uses zerocrossingtechnique.Initially,
Gaussian function is employedtominimizetheeffectofnoise
and smoothness. Then the Laplacian of the resultant is
calculated. If any sign change occurs in the Laplacian,wecan
conclude that an edge is detected, in other words when the
Laplacian crosses zero then an edge is detected and the
resultant binary image has a high value - ‘1’ in that point to
indicate the edge. If the intensity gradient starts increasing
or decreasing then the zero crossing reports an edge, and
this may happen at places which are not true edges. So, the
log method uses other alternativemethosfor edgedetection.
One such method is utilizing threshold of the log value so
that a value higher than the threshold would report an edge.
There is a chance of multiple edges are being detected.
Another method by comparing the pixel log value with that
of its adjacent pixels and choosing those points that are
having lesser log magnitude than that of its fourneighboring
points. But there is a risk of missing few edges in this
technique.
Canny Edge Detection Algorithm: The canny algorithm is
one of the most famous and most used edge detection
techniques as it is less prone to the noise. In this algorithm,
convolution of original image with Gaussian filtereliminates
the effect of noise [12]. The resultant image is then utilized
to calculate the intensity gradient and the direction of the
gradient based on the gradient angle. If the strength of the
intensity at that point is higher than that of its adjacent
points along the direction of the gradient then the edge can
be pointed out on the image. This algorithm employs two
threshold values-the higher valuetoincludethetruegenuine
edges and the lower value to trace the very small details of

the image. Hence, the canny edge detectionalgorithmgivesa
more accurate, elaborate and detailed binary image.
Algorithm Total Pixels Edge Pixels (Average)
Canny 262,144 23,818
Sobel 262,144 8,451
Prewitt 262,144 8,407
LOG 262,144 18,220
Table1. Comparison of detected pixels values of
different algorithms
Fig 2.1 RGB Image Fig 2.2 Grayscale Image
2. METHODOLOGY
The image in which the secret message will be embed ca be
referred as Cover Image. In this technique, the preprocesses
all will be done on the cover image. At first, boundary will be
traced for the cover image and then the edge pixels will be
detected using Canny algorithm. The Cover image is initially
converted into grayscale image. The secret message or
information will be embedded into the edge pixels of the
cover image, so an edge detection algorithm will be applied
to cover image so as to detect the edges in the cover image.
Canny edge detection will be considered among other
algorithm (from Table 1) since it can detect a greater
number of edges compared to other algorithms. Canny
detection uses convolution of original image with Gaussian
filter eliminates the effect of noise, so it is more resistant to
noise.
Fig.3 Message Hiding into Edge pixel process
Fig.4 Message Retrieving process from the Stego Image
To embed the data into image, the data must be in binary to
perform XOR operation. So, initially the data will be
converted into binary data using ASCII values for characters
and then decimal to binary conversion, which gives binary
data for the secret message. XOR operation will be
performed between the edge pixel value of cover image and
binary data of secret message. The resultant XOR value will
be added to the pixel value of Stego image. Stego image will
be formed form the resultant of cover image pixel value and
secret message data.
1. LSB=mod (cover (i, j),2);
2. a=str2double(binary_all(count));
3. temp=double (xor (LSB, a));
4. stego (i, j) =cover (i, j) +temp;
The stego image will be sent over internet from transmitter
to receiver. At the receiver’s end the secret message should
be retrieved back. The message will be retrieved back by
computing modulo 2 operations (remainder) between the
stego image pixels values. This will result in binaryformat of
the secret message, which can be converted into characters
using ‘bin2dec’ function.
2.1. Algorithm Implementation
1. % Read in original RGB image.
2. rgbImage = imread('godavari1.jpg');
3. subplot (2, 2, 1);
4. imshow(rgbImage)
5. axis ('on', 'image');
6. title ('Original Image')
7.
8. % Convert to gray scale.
9. grayImage = rgb2gray(rgbImage);
10. subplot (2, 2, 2);

11. imshow(grayImage)
13. title ('Grey Scale Image')
14.
15. % Get edges
16. Canny_img = edge (grayImage, 'Canny');
17. %Canny_img= cannydetector(grayImage);
18. subplot (2, 2, 3);
19. imshow (Canny_img, [])
21. title ('Edge Detected Image')
Code 1. Canny Edge Detection on the Cover Image
1. %====================================
======
2. %Encoding the message
3. %====================================
======
4. ascii=uint8(message);
5. binary_separate=dec2bin(ascii,8);
6. binary_all='';
7. for i=1: strlength(message)
8. binary_all=append (binary_all,binary_separate
(i, :));
9. end
10. count=1;
11. for i=1: row
12. for j=1: column
13. if count<=len
14. LSB=mod (cover (i, j),2);
15. a=str2double(binary_all(count));
16. temp=double (xor (LSB, a));
17. stego (i, j) =cover (i, j) +temp;
18. count=count+1;
19. end
20. end
21. end
22. imwrite(stego,'Stego_Image.jpg');
Code 2. Message Hiding into the edge pixels
Implementation
1. %====================================
======
2. %Decoding the message
3. %====================================
======
4. message_in_bits='';
5. for i=1: row
6. for j=1: column
7. if count<=len
8. LSB=mod (stego (i, j),2);
9. message_in_bits=append (message_in_bits,
num2str (LSB));
10. count=count+1;
11. end
12. end
13. end
14. i=1;
15. original_message='';
16. while i<=len
17. original_message=append(original_message,char
(bin2dec (message_in_bits (1, i: i+7))));
18. i=i+8;
19. end
Code 3. Message decoding from the edge pixels
Implementation

3. RESULTS AND CONCLUSIONS
Fig 5 is the given input image, the corresponding grayscale
image is generated and referenced as Cover image in Fig 6.
After the embedding the message into the Cover image the
resultant Stego Image, Fig 7 is generated which is used for
transmission over the internet. At the receiversidetheStego
image is take as input and the message is retrieved using the
decryption process. In this present implementation, two
images godavari1.jpg of 2816 x 2112 pixels and lena512.jpg
of 512 x 512 pixels color digital images has been taken as
cover image, tested for full embedding capacity and the
results are given. The effectiveness of the proposed
Steganography technique has been studied by calculating
MSE and PSNR for the message ‘SASTRA UNIVERSITY’
The MSE is calculated by using the equation,
Fig 5. Cover Image
Fig 6. Edge Detection Image
Fig 7. Final Image with Message hidden
where Xij is Stego value and Yij is the cover object.
The PSNR is calculated using the equation
where “I” is the intensity value of each pixel whichisequal to
255 for 8-bit gray scale images. Higher the value of PSNR
better the image quality
Table 2.1 Test Image – godavari1.jpg (2816 x 2112)
Metrices Results
Entropy 0.9125
SNR 23.4931
MSE 7.4037
PSNR 39.4363
The message is hidden/camouflaged into an image using
edge detection and XOR operation. Here Canny edge
detection algorithm is used for edge pixel detection. Output
Image (Stego image) pixels values are modified basedonthe
XOR operation between cover image pixels and secret
message data. Large data can also be embedded into an
image using this technique.
REFERENCES
[1]. Rengarajan Amritharajan, Benita Bose, Sasidhar
Imadbathuni and John Bosco Balaguru Rayappan. “Security
Building at the Line of Control of Image Stego”, International
Journal of Computer Applications, 2010.
[2]. R. Amritharajan, Akhila R, P. Deepikachowdaarapu. “A
computer Analysis of Image Steganography”, International
Journal of Computer Applications, May 2010.
[3]. Rupali Bharadwaj, Vaishali Sharmab. “Image
Steganography Based on Complemented Message and
Inverted bit LSB Substitution”.

[4]. Jhilik Guha, Orchi Saha, Riya Roy. “Improved edge-based
Image Steganographic Technique”.
[5]. Saiful Islam, Mangat R Modi, PhalguniGupta.Edge-based
Image Steganography. EURASIP Journal of Information
Security, 2014.
[6]. Ismail KICH, El Bachir AMEUR, Youssef Taouil. “Image
Steganography on Edge-Detection Algorithm, IEEE, 2018
International Conference.
[7]. Ramandhan J. Mastafa, Chrostian Bach. “Information
hiding in Images using Steganography Techniques”,
Northeast Conference of the American Society of the
American Society for Engineering Education (ASEE), 2013.
[8]. Rupali Roy. “Image Steganography usingPython”,May7,
2020.
[9]. Reem Abdulrahman Alomirah. “An Edge-based
Steganography Algorithm for Hiding text into images”,
Unitech Institute of Technology, Auckland, New Zealand,
2019.
[10]. M Saritha, Vishwanath M Khadabdi, M Sushravya.
“Image and text Steganography with Cryptography using
MATLAB”, International Conference on Signal Processing,
Communication, Power and Embedded System (SCOPES),
2016.
[11]. Larry S. Davis. “A survey of edge detection techniques”,
Computer Graphics and Image Processing,Volume4, Issue 3.
[12]. Paul bao, Lei Zhang and Xiaolin Wu. “Canny Edge
Detection Enhancement by Scale Multiplication”, IEEE
Transactions on pattern analysis and Machine Intelligence,
Vol 27, No 9, September 2005

Message Camouflage in an Image using EDGE-Based Steganography

More Related Content

Similar to Message Camouflage in an Image using EDGE-Based Steganography (20)

More from IRJET Journal (20)

Recently uploaded (20)

Message Camouflage in an Image using EDGE-Based Steganography