International Journal of Computer Applications (0975 – 8887)

Volume 98– No.14, July 2014

An Image Encryption Technique based on

Concatenating Images of Same Dimensions

Mohit Kumar Akshat Agrawal Ankit Garg

Research Scholar Assistant Professor Assistant Professor
Amity University Haryana Amity University Haryana Amity University Haryana
India India India

ABSTRACT Correlation coefficient assesses the correlation between two

Images are widely used in many areas for various purposes. adjoining pixels in an image [5, 6, 7]. Generally, correlation
Some images may have confidential information, so security measures the degree of similarity between two pixels. An
of secrete images is a major issue. Various image encryption encrypted image should have low correlation between two
algorithms provide security to images. However, traditional adjoining pixels [5, 6, 7], so that it becomes difficult to guess
encryption techniques encode one image at a time, which the value of neighbors of a pixel. For example, xi and yi are
leads to weak security in the encrypted image due to small two pixel pair then the correlation coefficient can be obtained
size of the image. In this paper, an image encryption by the equation (5) [5, 6, 7].
technique is imparted that concatenate four images of the 1 N
same dimension into one and then produce single encrypted E ( x) 
N

i
xi , (2)
image. In result, encrypted image is more secure due to the 1

large size. This encrypted image has high entropy and low N
1
correlation coefficient, so this encrypted image is more
difficult to understand compared to four separated images.
D( x ) 
N

i1
( xi - E ( x)) 2 , (3)

General Terms 1 N

Image security and image processing cov( x, y) 

N

i1
( xi - E ( x)) ( yi - E ( y)) , (4)

Keywords cov(x,y)
Correlation coefficient, concatenation, entropy, histogram and rxy  , (5)
image D( x ) D( y )

1. INTRODUCTION where D(x)  0 and D(y)  0

Multimedia technologies are developing swiftly and it is an
effective way to communicate, so users use this technology Where xi and yi are gray level value of two adjacent pixels, N
widely. Images are a part of multimedia and tremendously is the number of pairs (xi, yi) and E(x) is the mean of xi and
used in various areas, for example in communication, E(y) is the mean of yi [5, 6].
business, entertainment etc. Some images may be highly
confidential; they need security so that confidentiality can be In experiments, it is observed that correlation coefficient and
maintained. Encryption is an effective way to provide entropy in an encrypted image depend on the area of the
protection from illegal access [1, 2, 3]. Many image image [8]. Encrypted images of large area have a more
encryption techniques provide security to images. The main reduced correlation coefficient as compared to images of less
issue is to confirm the capability of an image encryption area [8]. Moreover, encrypted images of large region have
technique. So, diverse parameters are used to measure the more entropy compared to images of less area [8]. An image
capability of an image encryption algorithm by analyzing the encryption algorithm can take the benefit of this concept.
encrypted image. Entropy and correlation coefficient are two In this paper, a technique is proposed that joins four images of
critical parameters to measure the security level and capability
the same dimension and produce a single encrypted image.
of a technique.
This encrypted image has more entropy and low correlation
Entropy calculates the degree of randomness and uniform coefficient, so this encrypted image is more secure than four
distribution in the system [4, 5]. Thus, an encryption different encrypted images.
technique should show uncertainty and uniform distribution in
Rest of paper has the following section: proposed technique,
the encryption procedure [5]. Information entropy is
decryption, experiment and results, conclusion and future
calculated by the equation (1) [5, 6].
work.
2 N 1
H(m)    P(mi ) log 2  P(mi ) (1) 2. PROPOSED TECHNIQUE
i 0 Proposed technique works in two main phases; joining phase
and encryption phase.
Where p (mi) defines the probability of a pixel and N is the
number of bits in each pixel [5]. For a gray level image, each In joining phase, imparted technique takes four different
pixel has 8 bits, so the probability of a pixel is 1∕ 28 [5]. images of same dimension and joins them to make a single
Hence, information entropy of the gray level image is image. This single image is provided to second phase for
H (m) = 8 [5]. However, practically it is intricate to obtain encryption. Figure 1 shows the block diagram of this
ideal entropy; so slight difference is also tolerable [5]. technique.

30
 International Journal of Computer Applications (0975 – 8887)
Volume 98– No.14, July 2014

Step 5: obtain the complement of key

Four different images of same dimension Step 6: operate again XOR function on the result of step 4 and
the complement of key.
Step 7: decompose the resulting 400 bits of step 6 into 50
segments, where each segment will have 8 bits.
Joining Phase
Step 8: convert each segment to equivalent decimal number.
These decimal numbers are encrypted pixels.
Join four images and make a single image
Step 9: replace these encrypted pixels with the pixels in one-
dimensional array.
Step 10: pick next 50 pixels and repeat step from 3 to 9.
Encryption process Step 11: if 50 pixels are not available for encryption in the
end, then pick the rest pixels and repeat step from 3 to 9. But
Scrambling process this time key will have the number of bits that will be equal to
number of rest pixels multiplied by 8.
Step 12: convert one-dimensional array into two-dimensional
N times Key matrix having the same dimension as the original image.
Step 13 end of substitution process. The resultant matrix is
encrypted image.
Substitution process
3. DECRYPTION
In order to decrypt the received image, the decoding process
is applied that is reverse procedure of the encryption. In
decryption, substitution takes place, followed by scrambling.
Thus, a decrypted image is generated that is still has four
Encrypted image different images. The last step is to split the four images by
decomposing into four equal parts. To separate the decrypted
Fig 1. Block diagram of proposed technique
image into the four equal pieces, find out the size of this
2.1 Joining Phase image then decompose from middle horizontally. Thus, two
In this phase, take four images of the same dimension and join disconnected parts are obtained; afterward, decompose from
them in the form of a grid. In result, a single image is the middle to both parts vertically. Consequently, receiver will
generated which enters in the encryption phase. obtain four detached images without any loss of information
and confidentiality.
2.2 Encryption phase
Encryption process further works in two stages: scrambling
4. EXPERIMENT AND RESULT
and substitution, which are following as For this experiment, the four different images of same
dimension have been taken that are represented by figure 2,
2.2.1 Scrambling figure 3, figure 4 and figure 5.
Step 1: apply swapping operation among rows with the help
of key.
Step 2: apply swapping operation among columns with the
help of key.
Step 3: apply circular rotation on all rows with the help of
key.
Step 4: apply circular rotation on all columns with the help of
key.
Step 5: end of scrambling process.
Step 6: A scrambled image is produced and this image will Fig 2. First image Fig 3. Second image
enter in substitution phase.

2.2.2 Substitution
This phase has following steps
Step 1: transform image matrix to one-dimensional array.
Step 2: pick 50 pixels in sequence and convert into bits.
Step 3: operate the XOR function on these bits with a key of
400 bits.
Step 4: operate a circular shift process on the result of step 3
with the help of key. Fig 4. Third image Fig 5. Fourth image

31
 International Journal of Computer Applications (0975 – 8887)
Volume 98– No.14, July 2014

Figure 6 represents the concatenated image; figure 7

illustrates the histogram of concatenated image, figure 8
shows the encrypted image of the combined image and figure
9 represents the histogram of this encrypted image.

Fig 8. Encrypted image of concatenated image

Fig 6. Concatenated image

Fig 9. Histogram of encrypted image

After receiving this encrypted image, a user applies
decryption process to decrypt it and afterward decomposes
this concatenated image into two halves. Figure 10 illustrate
decrypted image, figure 11 and figure 12 shows two halves.

Fig 7. Histogram of concatenated image

Fig 10. Decrypted image

32
 International Journal of Computer Applications (0975 – 8887)
Volume 98– No.14, July 2014

Fig 11. First half Fig 13. Decrypted image 1 Fig 14. Decrypted image 2

Fig 15. Decrypted image 3 Fig 16. Decrypted image 4

Fig 12. Second half
Table 1 has the data of this experiment. It includes entropy
After decomposing into two halves, a user further applies and correlation coefficient in individual image and in
decomposition on these halves and obtains four separated concatenated image prior to encryption and after the
image. Figure 13, figure 14, figure 15 and figure 16 shows encryption.
detached images.

Table I. Experimental data

Entropy Correlation coefficient

Image (size) In original image In encrypted image In original image In encrypted image

First image (235*235) 7.4738 7.992 .7080 .0042

Second image (235*235) 7.1468 7.992 .4762 .0045

Third image (235*235) 7.1301 7.991 .7083 .0042

Fourth image (235*235) 7.6207 7.992 .6464 .0047

Concatenated image (470*470) 7.8701 7.9994 .4767 .0019

Figure 17 shows the correlation coefficient in encrypted image, number 4 shows the fourth image and number 5
image. In this figure, horizontal or x-axis represents the represents the concatenated image. In this figure, vertical or
images. At x-axis number 1 represents the first image, number y-axis represents the correlation coefficients in encrypted
2 illustrates the second image, number 3 shows the third

33
 International Journal of Computer Applications (0975 – 8887)
Volume 98– No.14, July 2014

image. This figure confirms that image of large area has lower As, low correlation creates difficulty to guess the value of
correlation coefficient than image of small area. neighbor pixels so encrypted image is secured. Consequently,
this research concludes that a user should encrypt image after
concatenating them for higher security, provided that the user
want to send multiple images of the same size.

6. FUTURE WORK
As the condition is that dimensions of images should be same
to concatenate them, if images are not of the same dimension
then a sender can add extra bits in the smaller image to make
this image equal to other images.
A video is made up of multiple frames or images with little
differences and these frames have same dimensions. So, a
user can use this technique effectively in video encryption. To
make this technique suitable for videos, there is a need to
improve performance of this technique in terms of encryption
time. So that it will become usable for real time encryption
system.

7. ACKNOWLEDGEMENT
Author is thankful to his guides Mr. Akshat Agrawal and Mr.
Ankit Garg who give him an opportunity to carry out this
work. Author is also thankful to NPTEL which provides E-
learning through online Web and Video courses in
Fig 17. Correlation coefficient in encrypted image Engineering, Science and humanities streams.

Figure 18 represents the entropy in individual encrypted 8. REFERENCES

image and in the concatenated image. In this figure, horizontal [1] W. Stallings, Cryptography and Network Security
or x-axis represents the encrypted image of the first image, principles and practices, 3rd ed., Pearson Education,
second image, third image, fourth image and attached image. 2003.
Moreover, vertical or y-axis represents the entropy in the [2] H. EI-din H. Ahmed, M. K Hamdy, and O. S. Farag
corresponding image. This figure shows that the first image, Allah, "Encryption quality analysis of the RC5 block
second image, third image and the fourth image has low cipher algorithm for digital images," Optical
entropy due to lower size, the fifth image that is concatenated Engineering, Vol. 45, Issue 10107003,2006
image has higher entropy. In results, this technique provides
high uncertainty in system, which provides high security. [3] Amitava Nag, Jyoti Prakash Singh, Srabani Khan,
Sushanta Biswas, D. Sarkar, Partha Pratim Sarkar
“Image Encryption Using Affine Transform and XOR
Operation” 2011 International Conference on Signal
Processing, Communication, Computing and
Networking Technologies (ICSCCN 2011), 21-22 July
2011, pages : 309-312.
[4] Shujiang Xu, Yinglong Wang, Jizhi Wang, Yucui Guo,
“A Fast Image Encryption Scheme Based on a Nonlinear
Chaotic Map”, 2010 2nd International Conference on
Signal Processing Systems (ICSPS), 5-7 July 2010,
pages: v2-326-v2-330.
[5] Mohit Kumar, Akshat Aggarwal and Ankit Garg, “A
Review on Various Digital Image Encryption Techniques
and Security Criteria”, International Journal of Computer
Applications, Volume-96, no-13, 17 June, 2014,
pages:19-26.
[6] Khaled Loukhaoukha, Jean-Yves Chouinard, and
Abdellah Berdai, “A Secure Image Encryption
Algorithm Based on Rubik's Cube Principle”, Journal of
Electrical and Computer Engineering, Volume 2012
Fig 18. Entropy in encrypted image (2012), Article ID 173931, 13 pages.
[7] Shujiang Xu, Yinglong Wang, Jizhi Wang, Yucui Guo,
5. CONCLUSION “A Fast Image Encryption Scheme Based on a Nonlinear
This work reveals that an image of large dimension has a Chaotic Map”, 2010 2nd International Conference on
more entropy compared to an image of a small dimension. Signal Processing Systems (ICSPS), 5-7 July 2010,
High entropy provides high randomness and uncertainty in the pages: v2-326-v2-330.
system. Moreover, after encryption, a large image has low
correlation coefficient compared to an image of small size.

IJCATM : www.ijcaonline.org 34