Image Steganography Using Inverted LSB Based On 2nd, 3rd and 4th LSB Pattern

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Image Steganography using Inverted LSB based on

2nd, 3rd and 4th LSB pattern


Fauzi Adi Rafrastara Raka Prahasiwi De Rosal Ignatius Moses Setiadi
Department of Informatics Engineering Department of Informatics Engineering Department of Informatics Engineering
Dian Nuswantoro University Dian Nuswantoro University Dian Nuswantoro University
Semarang, Indonesia Semarang, Indonesia Semarang, Indonesia
fauziadi@dsn.dinus.ac.id prahasiwi11@gmail.com moses@dsn.dinus.ac.id

Eko Hari Rachmawanto Christy Atika Sari


Department of Informatics Engineering Department of Informatics Engineering
Dian Nuswantoro University Dian Nuswantoro University
Semarang, Indonesia Semarang, Indonesia
eko.hari@dsn.dinus.ac.id atika.sari@dsn.dinus.ac.id

Abstract—Steganography is one method of concealing data, The inverted bit method is a method that can increase all
research on this topic always aims to improve image quality, three LSB steganography image targets at once. This method
message payload and security. The Inverted Least Significant was originally introduced by Yang [1], wherein the research
Bit (LSB) method is one technique that can be used to optimize introduced an inverted pattern on the LSB substitution
the quality of the steganography method. This study proposes a method. The proposed method which is proven to increase the
modified inverted LSB technique. In the previous study the capacity and imperceptibility of messages, message security
inverted LSB method used a two-bit LSB pattern, on the second also indirectly increases. This is because the insertion of
bit and the third bit LSB. Based on the calculation pattern the
messages is not done sequentially anymore, but with a reverse
calculation of bit changes in each pattern will determine
pattern. The inverted method is then developed by Akhtar [3],
whether or not the inverted LSB process is performed. This
research proposes the use of the LSB three-bit pattern, where
wherein the inverted bit method is based on the combination
the pattern is taken from the second, third and fourth bits of of the 5th and 6th bits of LSB. Inverted bits are done
LSB. The aim is to minimize LSB changes in each pixel of the adaptively based on changes in bit values. Before doing
cover image. Based on the tests that have been conducted, it has inverted bits, the total change is calculated first with the
been proven that the inverted LSB method using three LSB change pattern in the Table as in Table 1. Invert is done if the
patterns produces better results than the two-bit LSB pattern. bit changes are greater than the unchanging bits, thus the
Eight of the ten cover images tested, showed the results of an results of the steganography image change the pixel values
increase in the quality of the stego image. Chaotic map-based smaller.
message encryption methods are also applied and combined
with the LSB inverted method proposed to increase message Bharwaj dan Sharma [12] in his research also analyzed the
security. inverted bit method when compared to simple LSB and
random LSB, the result was that the inverted bit method was
Keywords—Arnold Transform, Data Hiding, Inverted LSB, superior. But the invert bit pattern carried out in Bharwaj and
Image Steganography, LSB Pattern Sharma's research is not based on the 5th and 6th bits but based
on the 2nd and 3rd bits of LSB. The success of the inverted bid
I. INTRODUCTION method has made this method developed again by combining
The steganographic image is a technique for hiding secret it with cryptographic methods such as RC4[13] and Arnold
messages that are done by inserting messages on an image Chaotic Map (ACM) [7]. Chaotic Map is an image
media. One of the classical methods that are widely used today cryptography method that encrypts by randomizing image
is the least significant bit (LSB) method. This method is one pixels, this technique is well known for its superiority and
method in the spatial domain, where the insertion of messages resistance to differential attacks, where ACM is one of the best
is done directly by changing the pixel value of the image. The chaotic map algorithms [14]. Based on these references, this
LSB method is very simple and predicTable, although this study proposes a combination of inverted bit methods with
method remains popular and continues to be developed. The ACM cryptography. But the inverted bit method proposed in
three targets addressed in the development of LSB-based this study was modified, not only based on the 2nd bit and 3rd
image steganography method are improving embedding bit, but also based on the 4th bit. The proposed method was
capacity, imperceptibility and security [1] [2] [3] [4]. There tested for performance and compared with the previous LSB
are various methods combined with the LSB method which inverted method based on the quality of the stego image
aims to improve one, two or three of these targets, such as the produced.
use of edge detection to improve imperceptibility [2] [4] [5],
II. INVERTED LSB BASED ON 2ND, 3RD AND 4TH LSB PATTERN
combine with cryptographic techniques to improve message
security [2] [4] [6] [7] [8], use of modulus and division Inverted LSB has been proposed in several previous
functions to increase message capacity[9], use of color theory studies. Some of the research proves the superiority of this
to increase message imperceptibility [8] [10], use of the method, especially it can give a significant influence on
inverted bit function to increase capacity, imperceptibility and imperceptibility messages. The inverted LSB technique is
message security [1] [3] [7] [11] [12]. based on the LSB 2-bit pattern. There are those that use
patterns on the 5th and 6th bits of LSB, some use the 2nd and
3rd bit patterns of LSB. The inverting process is done from

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


the calculation of changes in pixel values if iverted bits are Where (i, j) is the pixel coordinates of the original image,
done, so before inverted bits the profits must be calculated first (i ', j') is the pixel coordinates that have been shuffle, k and l
as in Table 1. are positive integers.

TABLE I. EXAMPLE INVERTED BIT PATTERN TABEL CALCULATION IV. IMPLEMENTATION AND RESULTS
BASEN ON RESEARCH [3]

Bit Pattern (5th Total changed Total Not Invert


and 6th LSB) pixels changed pixels Decision
00 7000 6500 Yes

01 9000 12000 No

10 8000 13000 No
Cameraman Fishing boat Lena
11 12000 4000 Yes
In this research the calculation Table is modified, where
the profit is calculated not only based on the 2-bit LSB pattern
but using the 3-bit LSB pattern, namely the 2nd, 3rd and 4th
bits. To see a clearer description, you can see the example in
Table 2.
Living room Baboon Pirate
TABLE II. EXAMPLE INVERTED BIT PATTERN TABEL CALCULATION
BASEN ON PROPOSED METHOD

Bit Pattern (2nd, Total changed Total Not Invert


3rd, and 4th LSB) pixels changed pixels Decision
000 10000 9000 Yes

001 6000 80000 No


Tank Truck Woman blonde
010 5000 70000 No

011 10000 5000 Yes

100 8000 6500 Yes

101 11000 3000 Yes

110 5500 8000 No


Woman dark hair Message Image
111 6500 7000 No
Fig. 1. Image datasets used
Based on Table 2 it appears that the calculation of the
pattern becomes more complex because in the previous As explained in the previous section, this research
method there were only four types of patterns, while in the proposes a combination of Inverted LSB and ACM methods
proposed method there were eight kinds of patterns. to secure messages. The message is embedded in the form of
Nevertheless, this method will have a positive impact if the a digital image with a depth of eight bits or grayscale type.
quality of the stego image produced is better. The size of the message image embedded is the maximum
size, where one pixel of the image of the cover is pinned to
III. ARNOLD CHAOTIC MAPS (ACM) one bit of the message image. Because the cover image used
ACM is a two-dimensional cryptographic algorithm in the is a standard test image with a size of 512 pixels * 512 pixels.
image found by Vladimir Arnold [7] [14]. ACM is often Then the size of the message image used is 128 pixels * 256
referred to as Arnold's transformation [7]. The type of pixels. The cover image used can be downloaded from the
cryptography that is done is to shuffle the pixel of the image website page [15]. The image of the cover and message image
with formula (1) without changing the pixel value of the used is shown in Fig. 1.
image. Image pixel shuffle process is based on a certain The first process is to shuffle the message image using the
number of iterations, where the number of iterations greatly ACM algorithm. This process is useful for giving additional
influences the visual results of image pixel shuffle. The security to messages. Because ACM input requires the same
number of iterations also greatly affects the length of time it size and width of the image, the message image is split into
takes to encrypt the image. Also note that the image shuffling two parts, each part measuring 128 pixels * 128 pixels. ACM
process is performed on images that have the same width and is then carried out with the same iterations in each section.
height (𝑁 ∗ 𝑁). While formula (2) is used to decrypt shuffled After scrambling using ACM, then the two parts are combined
images. again into one message image. For more details, see the stages
𝑖′ 1 𝑘 𝑖 of ACM-based encryption proposed in Fig. 2. After the
[ ]=[ ] [ ] 𝑚𝑜𝑑 𝑁 (1) message encryption process is done, the message inserted in
𝑗′ 𝑙 𝑘𝑙 + 1 𝑗
the cover image uses a reverse bit process based on the pattern
𝑖 1 𝑘 −1 𝑖′ (2) to the 2nd, 3rd and 4th LSB bits. This process will produce a
[ ]=[ ] [ ] 𝑚𝑜𝑑 𝑁 reverse bit decision Table consisting of eight kinds of pattern
𝑗 𝑙 𝑘𝑙 + 1 𝑗′
bits. Each pattern of bits will determine the bit process
decision. The decision to inverted the bit is based on the total
change in the bit of the cover image if the message bit is
inserted. If the total change in bits is greater than the number Bit Pattern Total Total Not
Invert
Final
of bits that have not changed, then inverted bits are performed. (2nd and 3rd changed changed changed
Decision
LSB) pixels pixels pixels
For example, in Table 3 shows the decision Table in the
Living Room image based on the proposed method. Total changed pixels 130332
In Table 3 and Table 4, there is a statistical result of
inverted LSB calculation on the living room image. In Table
3, it appears that there are 129360 pixels of cover images
whose LSB changes due to message insertion, whereas in the
previous method presented in Table 4 produces a change in
the total pixel of the cover image by 130332, meaning that
there are 972 pixels which do not change LSB values. Also
Split image note that the testing process carried out in Tables 3 and 4 is
done with the same message process and data, so in the
previous method the image message encryption process using
ACM was also performed.
The Living room image used as an example does not
produce a significant difference. Nevertheless, it is necessary
to see the quality comparison of imperceptibility of stego
Shuffled images based on measurements of MSE and PSNR. MSE
with ACM stands for a mean square error, while PSNR stands for peak
signal to noise ratio. These two measuring instruments are
standard measuring instruments that are widely used in
steganography image research. From the measurement of the
Joined two stego image produced the proposed method is obtained MSE
shuffled images value is 0.493469 and PSNR is 51.198203 dB, whereas for the
previous method the value of MSE is 0.497177 and PSNR is
51.165692 dB. The formula used to calculate MSE and PSNR
is shown in formula (1) and formula (2). Whereas the results
of MSE and PSNR produced in all cover images used are
shown in Table 5.

255 255

Fig. 2. Encryption steps based on ACM 𝑀𝑆𝐸 = ∑ ∑‖𝑐(𝑖, 𝑗) − 𝑠(𝑖, 𝑗)‖ (1)
𝑚=0 𝑛=0
TABLE III. INVERTED BIT PATTERN TABEL CALCULATION ON LIVING
ROOM IMAGE BASEN ON PROPOSED METHOD
2552
Bit Pattern Total Total Not Final 𝑃𝑆𝑁𝑅 = 10 𝑙𝑜𝑔10 ( ) (2)
(2nd, 3rd, and changed changed
Invert
changed √𝑀𝑆𝐸
Decision
4th LSB) pixels pixels pixels Where 𝑖, 𝑗 are the size of the image, 𝑐 is the cover image,
000 16570 17109 No 16570 𝑠 is the stego image.
001 16281 15826 Yes 15826
TABLE V. MSE AND PSNR RESULTS FROM PROPOSED METHOD AND
010 15980 15964 Yes 15964 PREVIOUS METHOD

011 16939 16554 Yes 16554 Bhardwaj and


Proposed Method
Image Sharma Method [12]
100 16624 16101 Yes 16101 MSE PSNR(dB) MSE PSNR(dB)

101 15902 16351 No 15902 Lena 0.4993 51.1476 0.4988 51.1516

110 17506 17074 Yes 17074 Woman dark hair 0.4969 51.1681 0.4977 51.1610

111 15994 15369 Yes 15369 Fishing boat 0.4921 51.2099 0.4930 51.2020

Total changed pixels 129360 Baboon 0.4983 51.1558 0.4983 51.1556

Tank 0.4856 51.2679 0.4903 51.2258


TABLE IV. INVERTED BIT PATTERN TABEL CALCULATION ON LIVING
ROOM IMAGE BASEN ON THE PREVIOUS METHOD [12] Cameraman 0.4980 51.1584 0.4979 51.1592
Bit Pattern Total Total Not Final Truck 0.4825 51.2954 0.4834 51.2876
Invert
(2nd and 3rd changed changed changed
Decision Living room 0.4935 51.1982 0.4972 51.1657
LSB) pixels pixels pixels
00 33194 33210 No 33194 Pirate 0.4967 51.1699 0.4976 51.1621
01 32183 32177 Yes 32177 Woman blonde 0.4952 51.1833 0.4982 51.1567
10 33486 33038 Yes 33038 Based on Table 5, it appears that the proposed method
appears superior, where out of the ten sample images used
11 32933 31923 Yes 31923
there are eight images that produce better stego image quality.
Although the difference is not significant, it appears that the TABLE IX. INVERTED BIT PATTERN TABEL CALCULATION ON
BABOON IMAGE BASEN ON PROPOSED METHOD
proposed method contributes positively to improving the
quality of the imperceptibility of stego images. This method Bit Pattern Total Total Not
Invert
Final
also keeps changing LSB values at each image pixel not (2nd, 3rd, and changed changed changed
Decision
reaching 50% from 262144 or equivalent 131072. To see the 4th LSB) pixels pixels pixels
total change in the number of LSB pixels in each cover image 000 16201 16401 No 16201
used can see Table 6 to Table 14. 001 16371 16420 No 16371

TABLE VI. INVERTED BIT PATTERN TABEL CALCULATION ON 010 16449 16562 No 16449
CAMERAMAN IMAGE BASEN ON PROPOSED METHOD
011 16720 16623 Yes 16623
Bit Pattern Total Total Not Final
Invert 100 16538 16581 No 16538
(2nd, 3rd, and changed changed changed
Decision
4th LSB) pixels pixels pixels
101 16288 16564 No 16288
000 15250 15445 No 15250
110 16159 16152 Yes 16152
001 15267 15234 Yes 15234
111 16108 16007 Yes 16007
010 15244 15173 Yes 15173
Total changed pixels 130629
011 15136 15281 No 15136

100 17888 17869 Yes 17869 TABLE X. INVERTED BIT PATTERN TABEL CALCULATION ON PIRATE
IMAGE BASEN ON PROPOSED METHOD
101 17143 16986 Yes 16986
Bit Pattern Total Total Not Final
Invert
110 17749 17863 No 17749 (2nd, 3rd, and changed changed changed
Decision
4th LSB) pixels pixels pixels
111 17039 17577 No 17039
000 14720 14811 No 14720
Total changed pixels 130436
001 14820 14877 No 14820

TABLE VII. INVERTED BIT PATTERN TABEL CALCULATION ON 010 14249 14023 Yes 14023
FISHING BOAT IMAGE BASEN ON PROPOSED METHOD
011 18906 19470 No 18906
Bit Pattern Total Total Not Final
Invert 100 19590 19215 Yes 19215
(2nd, 3rd, and changed changed changed
Decision
4th LSB) pixels pixels pixels
101 16620 16894 No 16620
000 13668 13546 Yes 13546
110 15894 15974 No 15894
001 15272 13938 Yes 13938
111 16072 16009 Yes 16009
010 16333 17131 No 16333
Total changed pixels 130436
011 16803 16182 Yes 16182

100 18951 19127 No 18951 TABLE XI. INVERTED BIT PATTERN TABEL CALCULATION ON TANK
IMAGE BASEN ON PROPOSED METHOD
101 16177 15309 Yes 15309
Bit Pattern Total Total Not Final
Invert
110 18064 17953 Yes 17953 (2nd, 3rd, and changed changed changed
Decision
4th LSB) pixels pixels pixels
111 16889 16801 Yes 16801
000 13706 13667 Yes 13667
Total changed pixels 129013
001 19880 17885 Yes 17885

TABLE VIII. INVERTED BIT PATTERN TABEL CALCULATION ON LENA 010 12187 13691 No 12187
IMAGE BASEN ON PROPOSED METHOD
011 21186 20377 Yes 20377
Bit Pattern Total Total Not Final
Invert 100 14704 14099 Yes 14099
(2nd, 3rd, and changed changed changed
Decision
4th LSB) pixels pixels pixels
101 16809 15460 Yes 15460
000 17163 17315 No 17163
110 15298 14745 Yes 14745
001 16756 16687 Yes 16687
111 18881 19569 No 18881
010 16040 16115 No 16040
Total changed pixels 127301
011 15501 15619 No 15501

100 15639 15592 Yes 15592 TABLE XII. INVERTED BIT PATTERN TABEL CALCULATION ON TRUCK
IMAGE BASEN ON PROPOSED METHOD
101 16173 16074 Yes 16074
Bit Pattern Total Total Not Final
Invert
110 16493 16765 No 16493 (2nd, 3rd, and changed changed changed
Decision
4th LSB) pixels pixels pixels
111 17158 17054 Yes 17054
000 18000 19875 No 18000
Total changed pixels 130604
001 15425 16188 No 15425
Bit Pattern Total Total Not
Invert
Final algorithm. The measurement of computational time needed
(2nd, 3rd, and changed changed changed uses the tic toc function in Matlab where the computational
Decision
4th LSB) pixels pixels pixels
process is also performed on the same computer. Table 15
010 15114 14993 Yes 14993 shows the results of the measurement of computational time
011 15873 19168 No 15873 needed for each algorithm.
100 17978 19104 No 17978

101 14392 14519 No 14392


Comparison Changed Pixels and
110 17862 16245 Yes 16245
Benefit
111 13817 13591 Yes 13591 133000 6000
132000 5000
Total changed pixels 126497
131000
130000 4000
TABLE XIII. INVERTED BIT PATTERN TABEL CALCULATION ON
WOMAN BLONDE IMAGE BASEN ON PROPOSED METHOD 129000 3000
128000 2000
Bit Pattern Total Total Not Final
(2nd, 3rd, and changed changed
Invert
changed
127000
Decision 126000 1000
4th LSB) pixels pixels pixels
000 18598 19106 No 18598 125000 0

001 16616 16289 Yes 16289

010 15621 15931 No 15621

011 16695 16372 Yes 16372

100 15900 15898 Yes 15898


Benefit Without Inverted LSB
101 15974 16505 No 15974
With Inverted LSB
110 16021 15645 Yes 15645

111 15408 15565 No 15408 Fig. 3. A comparative graph of the number of pixel changes with and
without the LSB inverted method and its benefits
Total changed pixels 129805
The graph shown in Fig. 3 is an illustration of the
TABLE XIV. INVERTED BIT PATTERN TABEL CALCULATION ON conclusions of the ten cover images used. Based on the graph
WOMAN DARK HAIR IMAGE BASEN ON PROPOSED METHOD it appears that all inverted bit processes can reduce the number
Bit Pattern Total Total Not Final of changes in the pixel value of the cover image. By reducing
Invert
(2nd, 3rd, and changed changed
Decision
changed the number of changes in pixel values, the quality of the stego
4th LSB) pixels pixels pixels image will be better. This is evidenced by the red line which
000 16573 16678 No 16573 always has a greater number of changes in bits than the blue
line. While the benefit bar that is green is the difference in
001 15771 15914 No 15771
changes in bits with and without the proposed method. Seen
010 14392 14362 Yes 14362 in the graph, the image of the tank has the greatest benefits
compared to the other nine images.
011 14436 14366 Yes 14366

100 13651 13823 No 13651 TABLE XV. COMPUTATION TIME FROM PROPOSED METHOD AND
PREVIOUS METHOD (IN SECONDS)
101 16554 16029 Yes 16029
Bhardwaj and
Image Proposed Method
110 20851 20422 Yes 20422 Sharma Method [12]
Lena 12.0368 12.9145
111 19086 19236 No 19086
Woman dark hair 13.6017 13.3959
Total changed pixels 130260
From the results shown in Table 6 to Table 14 it appears Fishing boat 15.3636 12.9030
that in each image there is an inverted LSB process, but each Baboon 13.4290 12.0615
image has an inverted LSB process that varies depending on
the decision on each pattern. Fig. 3 presents the difference in Tank 14.3830 13.1133
the number of image pixel changes after inserting the message Cameraman 13.8492 12.8612
using the inverted LSB method and without using the LSB
inverted method. Truck 12.5232 12.6537

Algorithms in the proposed method appear to be more Living room 14.3858 13.2595
complex. This is caused by differences in the number of bits Pirate 14.0238 12.4162
that are made into a pattern. Then testing the performance of
program calculations also needs to be measured. This is Woman blonde 14.3550 12.6670
necessary to ensure that these positive contributions do not Average 13.7951 12.8246
overload the computer's performance when executing the
Based on Table 15, the results of the measurement of the [5] D. R. I. M. Setiadi and J. Jumanto, "An Enhanced LSB-Image
computation time needed in the proposed method prove to be Steganography Using the Hybrid Canny-Sobel Edge Detection,"
a little longer. This is logical because the number of bit Cybernetics and Information Technologies, vol. 18, no. 2, pp. 74-88,
2018.
patterns used is twice as many, but the difference in computing
time is not significant because the difference is less than one [6] G. Ardiansyah, C. A. Sari, D. R. I. M. Setiadi and E. H. Rachmawanto,
"Hybrid Method using 3-DES, DWT and LSB for Secure Image
second[16]. This should not be a problem because the increase Steganography Algorithm," in International conferences on
in imperceptibility quality has paid off the time difference. Information Technology, Information Systems and Electrical
Engineering (ICITISEE), Yogyakarta, 2017.
V. CONCLUSIONS [7] E. J. Kusuma, C. A. Sari, E. H. Rachmawanto and D. R. I. M. Setiadi,
This research proposes modifying the inverted LSB "A Combination of Inverted LSB, RSA, and Arnold Transformation
method using a pattern made based on the second, third and to get Secure and Imperceptible Image Steganography," Journal of
ICT Research and Applications, vol. 12, no. 2, pp. 103-122, 2018.
fourth LSB bits. The process of testing the method uses the
maximum message size assuming one pixel of the cover [8] M. Juneja and P. S. Sandhu, "An Improved LSB Based Steganography
Technique for RGB Color Images," International Journal of
image is pinned to one message bit. Based on the results of Computer and Communication Engineering, vol. 2, no. 4, pp. 513-
testing the proposed method was successful in improving the 517, 2013.
quality of the stego image produced. The quality of stego [9] D. R. I. M. Setiadi, H. A. Santoso, E. H. Rachmawanto and C. A. Sari,
images can increase because the proposed method can reduce "An improved message capacity and security using divide and
the number of changes in pixel values. With increasing modulus function in spatial domain steganography," in International
quality, it can later be used to add message payload. More Conference on Information and Communications Technology,
inverted LSB patterns also indirectly increase message Yogyakarta, 2018.
security because the pattern of message embedding varies. But [10] A. Singh and H. Singh, "An improved LSB based image
steganography technique for RGB images," in IEEE International
more patterns make the computing process a little longer. Conference on Electrical, Computer and Communication
However, this is paid off by increasing the quality of the stego Technologies, Coimbatore, 2015.
image produced. To more guarantee message security, the [11] N. Akhtar, S. Khan and P. Johri, "An improved inverted LSB image
proposed method is also combined with ACM encryption. steganography," in International Conference on Issues and
Challenges in Intelligent Computing Techniques, Ghaziabad, 2014.
REFERENCES [12] R. Bhardwaj and V. Sharma, "Image Steganography Based on
[1] C.-H. Yang, "Inverted pattern approach to improve image quality of Complemented Message and Inverted bit LSB Substitution," in
information hiding by LSB substitution," Pattern Recognition, vol. 41, International Conference On Advances In Computing &
no. 8, pp. 2674-2683, 2008. Communications, Cochin, 2016.
[2] C. Irawan, D. R. I. M. Setiadi, C. A. Sari and E. H. Rachmawanto, [13] N. Akhtar, P. Johri and S. Khan, "Enhancing the Security and Quality
"Hiding and securing message on edge areas of image using LSB of LSB based Image Steganography," in Enhancing the Security and
steganography and OTP encryption," in International Conference on Quality of LSB Based Image Steganography, Mathura, 2013.
Informatics and Computational Sciences (ICICoS), Semarang, 2017. [14] K. K. Kabi, C. Pradhan, B. J. Saha and A. K. Bisoi, "Comparative
[3] N. Akhtar, "An LSB Substitution with Bit Inversion Steganography study of image encryption using 2D chaotic map," in International
Method," in International Conference on Advanced Computing, Conference on Information Systems and Computer Networks,
Networking and Informatics. Smart Innovation, Systems and Mathura, 2014.
Technologies, New Delhi, 2015. [15] Ming Hsieh Department of Electrical Engineering USC Viterbi
[4] E. J. Kusuma, O. R. Indriani, C. A. Sari, E. H. Rachmawanto and D. School of Engineering, "SIPI Image Database," [Online]. Available:
R. I. M. Setiadi, "An Imperceptible LSB image Hiding on Edge http://sipi.usc.edu/database/. [Accessed 2018].
Region using DES Encryption," in International Conference on [16] J. Nielsen, "Response Time," in Usability Engineering, California,
Innovative and Creative Information Technology (ICITech), Salatiga, Morgan Kaufmann, 1993, p. 135.
2017.

You might also like