(IJACSA) International Journal of Advanced Computer Science and Applications,

Vol. 10, No. 11, 2019

Image Steganography Using Combined Nearest and

Farthest Neighbors Methods
Farhana Sharmin1 Professor Dr. Muhammad Ibrahim Khan2
Department of Computer Science and Engineering Department of Computer Science and Engineering
Chittagong University of Engineering and Technology Chittagong University of Engineering and Technology
Chattogram, Bangladesh Chattogram, Bangladesh

Abstract—Security is invariably a significant concern A perfect Steganographic system is however an ideal

during communication. With the ease of communication, concept. Research in this field aims at increasing capacity,
there is always a pending threat of intrusion. Steganography security and robustness of the system. Steganography is a
is one such way to achieve security by concealing mechanism that implants a secret message into unspectacular
confidential information within a more innocent looking cover media to prevent arousing the suspicion of an
media like image, audio, video etc. In this paper, a new eavesdropper. The cover media thus serves as a container for
technique is proposed that uses the relationship of a pixel the secret information. The outcome is a stego media that
with its Nearest Neighbor and Farthest Neighbor to hide appears indistinguishable from the original media and becomes
undetectable. In the modern world digital steganography is
secret information into that pixel. The cover image is divided
used for secret communication as well as authentication. A
into2x2 non-overlapping blocks. According to the
steganographic system consists of two major parts: cover
vulnerability of the relationship among the pixels, blocks are media and stego media. The confidential information that
labelled as Stable and Unstable. The Stable block hides ‘k’ needs to be hidden is embedded in seemingly harmless media.
secret while unstable block hides ‘n’ secret bits. 25 types of This is the cover media. The result is a stego media that
different set of ‘k’ and ‘n’ is examined to evaluate the appears identical to the original cover media. Steganography
performance of proposed method. 2k method is applied to hides information by altering certain characteristics of the
improve the quality of stego image. The experimental result cover media that cannot be detected by visual inspection. It
shows that the proposed technique hides a significant makes use of both redundancies that exists in the digital cover
number of secret bits with high PSNR. While compared with media as well as the limitations of the human eye.
other existing methods, the proposed method achieves a
much higher visual quality than that of those methods. II. LITERATURE REVIEW
Different types of steganographic techniques are proposed
Keywords—steganography; cryptography; PVD; image to serve different purposes. Original Pixel Value Differencing
processing; cover image; information security; stego image (PVD) method for gray scale image was proposed by Wu and
Tsai in 2003 [7]. Here the cover image is segmented into non-
I. INTRODUCTION
overlapping blocks consisting two consecutive pixels. The
With the progress made in digital technology, information difference values between the pixels are classified into ranges.
in analog form are being rapidly converted into their digital Based on human vision’s tolerance to gray value fluctuation,
counterparts. Digital technology and the Internet today offer the interval of ‘Range’ is decided. Width of the ranges decides
available cover medias for use in steganography. The the number of secret bits to be hidden in a pixel. Thus, it is an
frequency of digital image on the Internet makes it the most adaptive method. To minimize the exposure of hiding effect of
common option among different cover medias. Internet PVD method, Modulus Function (MF) is employed in [1]. In
technology has bridged the global distance between people and [12], turnover policy and novel adjusting process is employed
organizations by allowing for fast and cheap communication. to improve the performance of method proposed in [1].
In areas such as online intellectual property rights, criminal, Capacity promoting technique is proposed in [2] to increase
political and commercial communications there is always a hiding capacity by finding more edge areas. A new
fear of third-party intruder who is seeking access to our quantization range table, in [11], is designed on the basis of the
communication. It is not enough to rely upon the assumption perfect square number to achieve better image quality and
that communications in today’s high-tech world are safe and higher capacity. By redefining the remainder function to a
not being intercepted. Ensuring secure communication is more general form, in [23], an indeterminate equation is
essential. Different tools ensuring secure communication derived to reduce image distortion. In [13], another
include: hiding the content or nature of communication steganographic technique is offered by using pixel value
(encryption, water-marking, steganography), hiding the parties differencing on colour images which also ensures that no pixel
to a communication (anonymity) and conceal a communication value in stego image exceeds the range 0 to 255. Where pixel
that has been taken place (obscurity). value does not cross the range, the original PVD method is
used and elsewhere proposed method has been used for
embedding data. Instead of focusing on individual color

1|Page
www.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 11, 2019

component, three color components are constituted into two its Nearest Neighbor and the pixel that has the highest distance
overlapping blocks in [19], where PVD technique is used from P1 is called its Farthest Neighbor. If Di is the lowest
separately on each block to conceal secret data. In [10], two distance then Pi+1 is the nearest neighbor and If Di is the highest
steganographic technique is proposed based on adaptive distance then Pi+1 is the farthest neighbor, where i = 1, 2, 3.
quantization range. In first technique, image is divided into 2x2
non overlapping pixel blocks and second technique divides the Step 4: Determine the Range of P1 according to the
image into 3x3 overlapping pixel blocks. First technique offers following four relationship conditions:
higher capacity where higher PSNR is achieved by using the Condition 1: If NN > P1 < FN then
second technique which in consequence offers higher security.
Smart pixel-adjustment technique is proposed in [5] to reduce Range = [0, min (NN - 1, FN - 1)]
visual distortion. Simple LSB substitution method [4] also Condition 2: If NN < P1 > FN then
achieves great image quality by applying Optimal Pixel
Adjustment Process (OPAP). To provide greater embedding Range = [max (NN + 1, FN + 1), 255]
capacity and less manipulation of image, the side match
method [3] exploits correlation between neighboring pixels. It Condition 3: If NN ≥ P1 ≥ FN then
hides secret data depending on the degree of smoothness. Pixel Range = [FN + 1, NN - 1]
Value Modification (PVM) method [22] uses modulus function
to embed secret information in a color image. Combined Pixel Condition 4: If NN ≤ P1 ≤ FN then
Value Differencing and Pixel Indicator Technique [15] is Range = [NN + 1, FN - 1]
proposed to ensure higher security. Here one channel acts as an
indicator where other two channels hold the secret information. Step 5: Calculate,
Just Noticeable Difference (JND) technique and method of |Range| = upper limit – lower limit + 1
Contrast Sensitivity Function (CSF) is used in [14]. It is an
edge detection method that utilizes partial information (3 bits Step 6: Calculate the number of secret bits, n, to be hidden
from MSB) of each pixel value. The authors proposed a into P1.
mathematical method, named 2k correction, for better
imperceptibility. n = min(floor(log2(|Range|)), k)
where ‘k’ may vary from 1 to 5. During experiment it is
III. PROPOSED METHOD found that the method reaches the maximum capacity at k = 5.
The proposed method can be applied both on gray image
Step 7: Take ‘n’ secret bits from the secret bit stream and
and color image. In case of color image, the Red, Green and
convert into its decimal value ‘b’.
Blue channels are separated first. Then embedding and
extraction algorithm is repeated on all three channels Step 8: Modify the value of pixel P1:
separately. Gray image does not need any pre-processing.
P1′ = P1 – (P1 mod 2n) + b
A. Embedding Process
Step 9: To keep the new pixel value P1′ nearest to P1, 2k
Step 1: Divide the cover image into 2X2 non-overlapping Correction method is applied. The steps are as follows:
blocks consisting four pixels: P1, P2, P3 and P4 as shown in
Figure 1. Step 9.1: If |P1′ - P1| > 2n-1 then go to Step 9.2. Otherwise
P1′ remains unchanged.
Step 9.2: If (P1′ - P1) > 0, then P1″ = P1′ - 2n else P1″ = P1′ +
P1 P2 2n
Step 10: Check the stability of the relationship among the
pixels in the block using Step 2 to Step 4. For correct message
P3 P4 extraction, it is required to preserve the same relationship
condition it has among the pixels even after hiding secret data.
If the relationship among the pixels remains unchanged then
Fig. 1. Cover image block the block is labelled as Stable Block and P1″ remains
unchanged. Otherwise the block is labelled as Unstable Block.
Step 2: Calculate the distance of P1 with its other neighbor In Unstable Block, all four pixels hide ‘n’ secret data. ‘n’ may
pixels P2, P3 and P4 by taking the differences as follows: vary from 1 to 5. For n>5, though the hiding capacity increases
significantly but PSNR also drops drastically, which leads to
Difference between P1 and P2 = D1 = |P1 – P2| vulnerability of the method. For this reason, the value of ‘n’ is
Difference between P1 and P3 = D2 = |P1 – P3| limited from 1 to 5. After hiding n bits, P1, P2, P3 and P4 will be
changed to P1′, P2′, P3′ and P4′ as follows:
Difference between P1 and P4 = D3 = |P1 – P4|
P1′ = P1 – (P1 mod 2n) + b1
Step 3: Determine the Nearest Neighbor (NN) and Farthest
Neighbor (FN) of P1. Compare D1, D2 and D3 calculated in P2′ = P2 – (P2 mod 2n) + b2
Step 2. The pixel that has the lowest distance from P1 is called

2|Page
www.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 11, 2019

P3′ = P3 – (P3 mod 2n) + b3 Step 6: Calculate the number of secret bits ‘n’ to be
extracted from P1′.
P4′ = P4 – (P4 mod 2n) + b4
n = min(floor(log2 (|Range|)), k)
Where, n is the secret binary bits and b1, b2, b3 and b4 are
Step 7: Extract n secret bits from the stego pixel P1′.
corresponding decimal values of those secret bits.
b = P1′ mod 2n
To increase security, secret information is embedded in
random fashion. Secret message is embedded into stable blocks Step 8: Convert the decimal value b into its corresponding
first then into unstable blocks. In that way, message is hidden binary value. Obtained binary bits are the secret bits.
in random order instead of sequential order.
In case of Unstable Block, secret message is extracted
from all four pixels using following equations:

b1 = P1′ mod 2n

b2 = P2′ mod 2n

b3 = P3′ mod 2n

b4 = P4′ mod 2n

By converting b1, b2, b3 and b4 into their corresponding

binary value, we get the secret bits hidden into the pixels.
Fig. 2. Embedding secret message in random order IV. EXPERIMENTAL RESULTS
Most widely used 512X512X3 color images (Lena,
For example, in Figure 2, if ‘abcdefghijklmnop’ is the Baboon, Pepper, Jet, Sailboat, House, Splash, Tiffany etc.) and
secret message, then it is embedded as ‘jabkcdlemfgnohpi’. 512X512 gray images (Lena, Boat, Baboon, Tank, Pepper,
After embedding into all blocks of the cover image, we get the Tiffany, Jet, Elaine, Barbara etc.) are used to conceal the secret
stego image. information. In order to assess the efficiency of proposed
B. Extraction Process method, two parameters are employed: Capacity and Peak
Signal to Noise Ratio (PSNR). Capacity is the maximum
Step 1: Divide the stego image into 2X2 non overlapping number of secret bits an image can hide. Peak Signal to Noise
blocks and identify stable and unstable blocks. Ratio (PSNR) quantifies the quality of stego image.

m-1 n-1
1
MSE = c m n Σ Σ
i=0 j=0
[ I(i , j) – K(i , j)]2

Fig. 3. Stable block Fig. 4. Unstable block Here, MSE calculates the Mean Squared Error. I and K are
In case of stable blocks: cover and stego image respectively where both have m rows
and n columns, c is the number of channels of the image.
Step 2: Calculate the distance of P1′ with P2, P3 and P4. MAXI is the maximum pixel value of cover image. Higher
PSNR indicates better image quality. Usually, when the PSNR
Difference between P1′ and P2 = D1′ = |P1′ – P2|
is greater than 30 dB, Human Visual System (HVS) is unable
Difference between P1′ and P3 = D2′ = |P1′ – P3| to detect the changes made in an image [21].
Difference between P1′ and P4 = D3′ = |P1′ – P4| In steganography, capacity and PSNR walk in reverse way.
The more the capacity, usually the lower the PSNR. And lower
Step 3: Determine the Nearest Neighbor (NN) and Farthest PSNR leads to vulnerability of a method. In our proposed
Neighbor (FN) of P1′ same as described in Step 3 of embedding method, ‘k’ and ‘n’ play a major role to control the capacity.
process. As for example, Table 1 (color image) and Table 2 (gray
Step 4: Determine the Range of P1′ according to the four image) depict how capacity (in bits) as well as PSNR (in db)
relationship conditions specified in Step 4 of embedding varies for different values of ‘k’ and ‘n’ for image Lena. Users
process. can choose any set of (k,n) to meet their priority requirement.
Step 5: Calculate, |Range| = upper limit – lower limit + 1

3|Page
www.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 11, 2019

TABLE I CAPACITY AND PSNR OF COLOR Taking 30 db as the threshold level PSNR, Table 3 and
IMAGE LENA FOR DIFFERENT VALUES OF ‘K’ AND ‘N’ Table 4 depict the highest capacity with threshold level PSNR
Type of (k,n) Capacity PSNR and capacity with highest PSNR for color images and gray
Type 1 (1,1) 742866 56.20
images respectively. Capacity is measured in bits and PSNR is
Type 2 (1,2) 1471202 51.45
measured in db. From the experimental results, we can see that
the proposed method can hide a large amount of secret
Type 3 (1,3) 2199538 45.83
information yet keeping a high PSNR.
Type 4 (1,4) 2927874 39.85
Type 5 (1,5) 3470530 34.13
Type 6 (2,1) 712671 56.40
TABLE III. CAPACITY AND PSNR FOR PROPOSED
Type 7 (2,2) 1380255 51.76 METHOD (COLOR IMAGE)
Type 8 (2,3) 2047839 46.18
Type 9 (2,4) 2715423 40.22 Cover Image Highest Capacity Capacity with
TABLE IV. CAPACITY AND PSNR OF PROPOSED
(512 x 512 x 3) METHOD
with(C
threshold
OLOR IMAGE)
highest PSNR
Type 10 (2,5) 3381967 34.28 level PSNR
Type 11 (3,1) 702121 56.23 Capacity PSNR Capacity PSNR
Type 12 (3,2) 1326941 51.88 Lena 3470530 34.13 712671 56.40
Type 13 (3,3) 1951761 46.42 Baboon
TABLE V. 3490302 34.14 768667 56.03
Type 14 (3,4) 2576581 40.50 Pepper 3474527 34.28 730659 56.25
Sailboat 3481380 34.18 745340 56.17
Type 15 (3,5) 3201401 34.57
Car house 3475411 34.19 708470 56.46
Type 16 (4,1) 702444 50.43 Splash 3321788 34.21 669662 56.75
Type 17 (4,2) 1058920 49.44 Jet 3424467 34.33 684031 56.64
Type 18 (4,3) 1415396 46.83 Tiffany 3471557 34.64 710055 56.40
Type 19 (4,4) 1771872 42.29
Type 20 (4,5) 2128348 36.83 TABLE IV CAPACITY AND PSNR FOR
PROPOSED METHOD (GRAY IMAGE)
Type 21 (5,1) 702612 55.46
Type 22 (5,2) 1303484 51.69 Cover Image Highest Capacity Capacity with
TABLE
(512 I.
x 512) CAPACITY AND PSNR OF PROPOSED
with threshold highest PSNR
Type 23 (5,3) 1904356 46.48 METHOD (COLOR IMAGE)
level PSNR
Type 24 (5,4) 2505228 40.63 Capacity PSNR Capacity PSNR
Type 25 (5,5) 3106100 34.73 Lena 948378 35.22 232365 51.74
Baboon 1013763 34.97 249900 51.36
TABLE II.
Pepper 1038439 30.01 237555 51.62
TABLE II CAPACITY AND PSNR OF GRAY IMAGE
LENA FOR DIFFERENT VALUES OF ‘K’ AND ‘N’ Jet 1038254 30.00 226950 51.91
Airplane 910753 30.74 191549 52.74
Type of (k,n) Capacity PSNR Boat 1002498 35.02 244560 51.47
Type 1 (1,1) 242106 51.55 Lake 998313 35.04 244987 51.48
Type 2 (1,2) 477530 46.82 Tiffany 1017169 30.11 228397 51.83
Type 3 (1,3) 712954 41.19 Elaine 1010298 34.96 247009 51.43
Goldhill 994878 35.08 242765 51.52
Type 4 (1,4) 948378 35.22
Cameraman 1018789 30.05 217845 52.16
Type 5 (1,5) 1158682 29.39
Carhouse 932328 35.30 233172 51.75
Type 6 (2,1) 232365 51.74 Tank 1047990 30.00 242889 51.49
Type 7 (2,2) 448009 47.11 Truck 980733 35.16 238737 51.60
Type 8 (2,3) 663653 41.55 Couple 985443 35.08 240799 51.53
Type 9 (2,4) 879297 35.60 Zelda 1027092 30.00 230426 51.77
Type 10 (2,5) 1094941 29.68 Barbara 979443 35.09 241677 51.54
Type 11 (3,1) 229963 51.65
Type 12 (3,2) 435735 47.22 Figure 5 shows some color images and their corresponding
Type 13 (3,3) 641507 41.73 stego images where some gray images and their respective
Type 14 (3,4) 847279 35.79 stego images are shown in Figure 6. However, it is also clear
Type 15 (3,5) 1053051 29.89
from the figures that it is evident that Human Visual System
Type 16 (4,1) 230056 45.95
(HVS) does not perceive distortion.
Type 17 (4,2) 351460 44.87 V. COMPARATIVE ANALYSIS AND DISCUSSION
Type 18 (4,3) 472864 42.13
To analyse how well the proposed method is working, it is
Type 19 (4,4) 594268 37.51
compared to some relevant methods proposed before like Wu
Type 20 (4,5) 715672 32.03
and Tsai’s method [7], Mandal and Das’s method [13], Tseng
Type 21 (5,1) 229831 51.26 and Leng’s method [11], Yang et al.’s method [2], Joo et al’s
Type 22 (5,2) 430735 47.12 method [12]. Swain’s method [10], Yang and Wang’s method
Type 23 (5,3) 631639 41.78 [5], and Prasad and Pal’s method [19]. In order to have a
Type 24 (5,4) 832543 35.89 comparative analysis of the outcomes of proposed method,
Type 25 (5,5) 1033447 29.98 Table 5 and Table 6 is constructed to show the comparison

4|Page
www.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 11, 2019

a b c a b c

d e f d e f

g h i g h i

j k l j k l
Fig. 5. Color cover images: (a) Lena (b) Baboon (c) Peppers (d) Jet Fig. 6. Gray cover images: (a) Lena (b) Baboon (c) Peppers (d) Jet
(e) Splash (f) Carhouse and (g) – (l) are their respective stego (e) Splash (f) Carhouse and (g) – (l) are their respective stego
images for Type 18 (k = 4, n = 3) images for Type 18 (k = 4, n = 3)

between proposed method and other existing methods for color Mandal and Das’s method, Yang and Wang’s method, Swain’s
image and gray image respectively. For each existing method, method and Prasad & Pal’s method respectively. Applying the
PSNR at the highest capacity is compared with the PSNR of proposed method on gray images also produces better result
proposed method at the same capacity. From Table 5 and Table than other methods compared. Here also, for same capacity,
6, it is evident that, proposed method outperformed the existing proposed method’s average PSNR is 16.29%, 3.50%, 16.94%
methods each time. For same capacity, proposed method’s and 12.81% higher than the PSNR gained in Wu & Tsai’s
PSNR is significantly higher than other methods compared. In method, Tseng et al.’s method, Yang et al.’s method and Joo et
case of color image, proposed method’s average PSNR is al.’s methods respectively.
26.43%, 49.21%, 11.02% and 41.32% higher than that of

TABLE V COMPARISON BETWEEN PREVIOUS METHODS AND PROPOSED METHOD FOR COLOR IMAGES
Cover Image Mandal and Proposed Yang and Proposed Swain’s method Proposed Prasad and Pal’s Proposed
TABLE VI. CAPACITY AND PSNR OF PROPOSED METHOD (COLOR IMAGE)
(512 x 512 x 3) Das’s method method Wang’s method method method method method

Capacity PSNR PSNR Capacity PSNR PSNR Capacity PSNR PSNR Capacity PSNR PSNR
Lena 1166296 42.26 52.45 196608 41.58 62.14 1341192 46.17 51.89 1976671 31.01 46.34
Baboon TABLE VII.38.44
1159328 52.43 196608 33.29 61.96 1489945 48.49 51.38 2219715 32.29 45.79
Peppers 1167960 42.28 52.42 196608 39.43 62.01 1350251 47.06 51.81 1783210 30.10 46.81
Jet 1165184 42.60 52.51 196608 43.73 62.35 1267690 46.18 52.14 1753707 35.66 46.88
Sailboat 1146224 40.66 52.48 196608 47.41 62.04 1424967 47.29 51.59 2130772 33.11 46.16
Car-House 1162992 41.41 52.49 196608 41.34 62.26 1339985 44.73 51.90 2079088 34.59 46.08
Splash 1173856 42.86 52.52 196608 44.86 62.39 - - - - - -

5|Page
www.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 11, 2019

TABLE VI COMPARISON BETWEEN PREVIOUS METHODS AND PROPOSED METHOD FOR GRAY IMAGES
Cover Image Wu and Tsai’s method Proposed Tseng et al. Proposed Yang et al. Proposed Joo et al. Proposed
TABLE VIII. CAPACITY AND PSNR OF PROPOSED METHOD (COLOR IMAGE)
(512 x 512) Method method Method method Method Method Method
Capacity PSNR PSNR Capacity PSNR PSNR Capacity PSNR PSNR Capacity PSNR PSNR
Lena 407680 41.79 47.51 215740 50.70 52.07 528966 36.75 42.56 407152 43.4 47.52
Baboon 450328 37.90 47.04 241719 48.57 51.50 559222 34.30 42.22 456344 39.2 46.98
TABLE IX.
Jet 409944 40.97 47.52 204682 50.89 52.37 - - - 409768 42.8 47.52
Peppers 405480 41.73 47.50 217290 50.57 52.01 528791 36.83 42.54 406520 42.5 47.49
Elaine 407128 42.1 47.44 - - - - - - 407144 43.5 47.44
Tiffany 403764 41.47 47.53 210935 50.86 52.18 528678 36.35 42.49 - - -
Boat 418560 39.6 47.29 - - - - - - 419920 41.0 47.27
Tank 403990 42.3 47.33 - - - - - - - - -
Lake 420912 40.0 47.34 224915 49.86 51.85 535984 36.51 42.41 421296 41.5 47.34
Gold hill 405634 42.20 47.48 - - - 529319 37.15 42.51 - - -
Zelda 398584 42.66 47.63 - - - 526145 37.78 42.60 - - -
Barbara 442529 36.24 47.12 - - - 548206 34.80 42.33 - - -

VI. CONCLUSION
An efficient technique is introduced to hide a significant [10] G. Swain, "Adaptive pixel value differencing steganography using both
amount of secret information keeping the distortion vertical and horizontal edges", Multimedia Tools and Applications, vol.
unnoticeable. Capacity and PSNR is quite higher than other 75, no. 21, pp. 13541-13556, 2015.
existing methods discussed. The greatest advantage of our [11] H.-W. Tseng and H.-S. Leng, “A Steganographic Method Based on
proposed method is, users can easily tailor this method Pixel-Value Differencing and the Perfect Square Number,” Journal of
Applied Mathematics, vol. 2013, pp. 1–8, 2013.
according to their need by taking the desired Type of (k,n).
[12] J. Joo, H. Lee and H. Lee, "Improved Steganographic Method
Either to achieve higher capacity or to maintain a higher Preserving Pixel-Value Differencing Histogram with Modulus
PSNR, whatever the target may be, just tuning the values of ‘k’ Function", EURASIP Journal on Advances in Signal Processing, vol.
and ‘n’ is sufficient. Random embedding order is followed 2010, no. 1, 2010.
instead of traditional sequential order by hiding secret [13] J. K. Mandal and D. Das, “Colour Image Steganography based on Pixel
information into Stable Blocks first than into Unstable Blocks. Value Differencing in Spatial Domain,” IJIST International Journal of
This step is working as an extra layer of security if any Information Sciences and Techniques, vol. 2, no. 4, pp. 83–93, 2012.
unauthorized attack happens. Since only the sender and [14] J.-G. Yu, E.-J. Yoon, S.-H. Shin, and K.-Y. Yoo, “A New Image
recipient know the correct order of extraction, the intruder is Steganography Based on 2k Correction and Edge-Detection,” Fifth
International Conference on Information Technology: New Generations
supposed to get a meaningless message in that case. (itng 2008), 2008.
[15] M. G. Gouthamanaath and A. Kangaiammal, “Color Image
REFERENCES Steganography using Combined Pixel Value Differencing and Pixel
[1] C. Wang, N. Wu, C. Tsai and M. Hwang, “A high quality Indicator Technique in Spatial Domain,” IJCA Proceedings on National
steganographic method with pixel-value differencing and modulus Conference on Research Issues in Image Analysis and Mining
function”, Journal of Systems and Software, vol. 81, no. 1, pp. 150-158, Intelligence, pp. 20–23, Jun. 2015.
2008. [16] M. Hussain and M. Hussain, “A Survey of Image Steganography
[2] C. Yang, C. Weng, H. Tso and S. Wang, “A data hiding scheme using Techniques,” International Journal of Advanced Science and
the varieties of pixel-value differencing in multimedia images”, Journal Technology, vol. Vol. 54, pp. 113–124, May 2013.
of Systems and Software, vol. 84, no. 4, pp. 669-678, 2011. [17] M. Sabokdast and M. Mohammadi, “A steganographic method for
[3] C.-C. Chang and H.-W. Tseng, “A steganographic method for digital images with modulus function and modified LSB replacement based on
images using side match,” Pattern Recognition Letters, vol. 25, no. 12, PVD,” The 5th Conference on Information and Knowledge
pp. 1431–1437, 2004. Technology, 2013.
[4] C.-K. Chan and L. Cheng, “Hiding data in images by simple LSB [18] S. Gupta, A. Goyal, and B. Bhushan, “Information Hiding Using Least
substitution,” Pattern Recognition, vol. 37, no. 3, pp. 469–474, 2004. Significant Bit Steganography and Cryptography,” IJMECS
[5] C.-Y. Yang and W.-F. Wang, “Block-Based Colour Image International Journal of Modern Education and Computer Science, vol.
Steganography Using Smart Pixel-Adjustment,” Advances in Intelligent 4, no. 6, pp. 27–34, 2012.
Systems and Computing Genetic and Evolutionary Computing, pp. 145– [19] S. Prasad and A. Pal, "An RGB colour image steganography scheme
154, 2015. using overlapping block-based pixel-value differencing", Royal Society
[6] Chadha A., Satam N., Sood R., and Bade D., “An Efficient Method for Open Science, vol. 4, no. 4, p. 161066, 2017.
Image and Audio Steganography using Least Significant Bit [20] T. Denemark and J. Fridrich, “Side-informed steganography with
Substitution,” International Journal of Computer Applications, vol. 77, additive distortion,” 2015 IEEE International Workshop on Information
no. 13, pp. 37-45, 2013. Forensics and Security (WIFS), 2015.
[7] D.-C. Wu and W.-H. Tsai, “A steganographic method for images by [21] T.-S. Chen, C.-C. Chang, and M.-S. Hwang, “A virtual image
pixel-value differencing,” Pattern Recognition Letters, vol. 24, no. 9-10, cryptosystem based upon vector quantization,” IEEE Transactions on
pp. 1613–1626, 2003. Image Processing, vol. 7, no. 10, pp. 1485–1488, Oct. 1998.
[8] Dange and C. G. Desai, “Secured data communication system using [22] V. Nagaraj, V. Vijayalakshmi, and G. Zayaraz, “Color Image
RSA with mersenne primes and Steganography,” 2015 2nd International Steganography based on Pixel Value Modification Method Using
Conference on Computing for Sustainable Global Development Modulus Function,” IERI Procedia, vol. 4, pp. 17–24, 2013.
(INDIACom), pp. 1306–1310, 2015. [23] W. Zhao, Z. Jie, L. Xin, and W. Qiaoyan, “Data embedding based on
[9] G. Prabakaran. and R. Bhavani., “A modified secure digital image pixel value differencing and modulus function using indeterminate
steganography based on Discrete Wavelet Transform,” 2012 equation,” The Journal of China Universities of Posts and
International Conference on Computing, Electronics and Electrical Telecommunications, vol. 22, no. 1, pp. 95–100, 2015.
Technologies (ICCEET), 2012.

6|Page
www.thesai.org