Measurement: Sensors 24 (2022) 100520

Contents lists available at ScienceDirect

Measurement: Sensors
journal homepage: www.sciencedirect.com/journal/measurement-sensors

Efficient methods for digital image watermarking and

information embedding
Sunil Gupta a, Kamal Saluja a, *, Vikas Solanki a, Kushwant Kaur b, Parveen Singla c,
Mohammad Shahid d
a
Chitkara University Institute of Engineering and Technology, Chitkara University, Punjab, India
b
Chandigarh University, Mohali, 140413, India
c
Chandigarh Engineering College-CGC Landran, Mohali, 140307, Punjab, India
d
Noida Institute of Engineering and Technology, Greater Noida, UP, India

A R T I C L E I N F O A B S T R A C T

Keywords: Now a day to achieve security of digital data, utilizing digital image watermarking scheme is the option. The
SVD primary purpose to upgrade image watermarking scheme is to fulfill the need of sufficient imperceptibility,
FWHT capacity, and robustness of watermark against several attacks like JPEG compression parameter, salt and pepper
QIM
noise, cropping factor of the image part which contains a watermark and rotating the watermark at one
BPNN
DFIS
particular angle. During embedding stage in digital image watermarking scheme, a secret message or image
Huffman introduced into a host image without destroying the quality of original cover image for security and authenti­
cation of the data. During extraction stage in the watermarking scheme applying a process to that employed in
insertion step but in a reverse way. In this paper different evaluation based on the watermarking methods
described. Many projects have been presented such as DCT, DWT, DFT, SVD using FWHT, QIM, BPNN, DFIS,
EZW, Huffman coding to provide protection, robustness and high embedding capacity. In the digital image
watermarking world, where Peak signal-to-noise ratio (PSNR) is utilized to determine the image quality factor
and Normalized correlation (NC) and robustness of the watermark. Using this learning opportunity an envi­
ronment friendly digital image watermarking scheme can be used with sustainable infrastructure in developing
countries.

1. Introduction picture [3,4].

Invisible watermark images are more resistant to a wider range of
Using digital watermarking, secret information can be hidden in signal processing attacks because they are more resilient. Digital picture
digital multimedia data such as music, video, images, and text via the watermarking’s two-phase idea of watermark insertion and retrieval can
cell phone or the internet [1]. After entering information, the image be summarized as below in Fig. 1:
quality will not be degraded by using this watermarking process. Almost modern transformation technique using for the water­
Nowadays because of the very fast-growing transmission method, it is marking scheme to get the high capacity to insert amount of informa­
very easy to copy or create fake information during transmitting over tion, security and robustness are DCT, DWT, DFT [5]. The great
channel so as solution digital watermarking is very attractive now to decomposition scheme is singular value decomposition (SVD) for
provide authenticity as well as security from copyright [2]. With digital watermarking point of view [6]. So, after applying suitable trans­
image watermarking, the information about the host image is embedded formation technique, the SVD approach is utilized to improve the quality
both securely and undetectably in the picture itself. The frequency and parameter of designing image, but it is not sufficient to resist various
the location This is the easiest method because the watermark image attack. So, researcher uses DCT with SVD approach for protection
pixels are embedded in the LSB of the image pixels used for the cover against several attacks, then the trend to use DWT with SVD [7] is start

Abbreviations: SVD, Singular Value Decomposition; FWHT, Fast Walsh Hadamard Transform; QIM, Quantization Index Modulation; BPNN, Back Propagation
Neural Network; DFIS, Dynamic Fuzzy Interference System.
* Corresponding author.
E-mail address: [email protected] (K. Saluja).

https://doi.org/10.1016/j.measen.2022.100520
Received 30 September 2022; Received in revised form 10 October 2022; Accepted 13 October 2022
Available online 19 October 2022
2665-9174/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
S. Gupta et al. Measurement: Sensors 24 (2022) 100520

to achieve better evaluation factor. Then researcher uses multiple • Imperceptibility: The term imperceptibility is useful to express the
watermarking in the cover image so focus on the combined approach of hiding capability of the inserted watermark. In these terms, the
DCT-DWT-SVD [8]. Now in recent year, the variety of algorithm also quality of a visual image cannot destroy. The comparison between
using in the watermarking scheme. the original host image and the watermarked image is used to
Many watermarking algorithms and approaches of the SVD are determine whether or not this condition exists.
introduced in this paper to increase the quality parameter of the • Computational Cost: For embedding and then recovery methods of
embedded image against the variety of the signal processing attacks to the watermark, it is an important measure to find the better
achieve the solution for copyright protection and authentication of computing resources and for a particular configuration of the com­
hidden content. So, in this paper, several important evolutions of the puter, its processing time is to be helpful to evaluating the compu­
watermark are discussed in section-1. Then in section 2 evaluation and tational costing [11].
in section 3 several transformation techniques are described. In section 4
SVD approach with the latest algorithm presented and section 5 pre­ 3. Transformation techniques
sented the finding. At last, in section 6 we conclude this paper.
The employed transformation is Discrete Cosine Transform (DCT),
2. Evaluation on watermark Discrete Fourier Transform (DFT), Discrete Wavelet Transform (DWT)
[12]. These transform schemes describe below:
Different evaluation based on the watermarking methods is as by this
paper presented [9]: 3.1. Discrete Cosine Transformation (DCT)

• Peak signal-to-noise ratio: The PSNR is not a perfect and required It is useful to remake the set of the coefficient of a spatial domain
metric, but this is useful to evaluate some good comparisons for the waveform into its frequency domain [13]. It is one of the wells receive
performance of its imperceptibility. It is although an enough term to linear transformation and a compression mechanism of the digital
considering the strength of the associated signal, so it doesn’t only signal. Because of its de-correlation as well as energy compression
contain the errors. The equation which is used to define and for feature, it is the most popular procedure in the watermarking scheme.
absolute measurements of the difference between the value obtained With an input as image ‘x,’ the coefficient of the DCT for output
from the two different images. image ‘y’ calculated through the equation as:
( ) √̅̅̅̅̅̅√̅̅̅̅ [ ] [ ]
MAXi2 2 2 ∑
M− 1 ∑
N− 1
(2x + 1)μπ (2y + 1)vπ
PSNR = 10 log (1) Y(μ, ν) = au av f(m, n)cos cos
MSE M N 2N 2N
u=0 v=0

(3)
When the cover image and the watermark enclosing it have a high de­
gree of resemblance, the PSNR estimates will be more accurate. For μ, ν = 0, 1,2 …. N-1

• Capacity: It is showing that the amount of the information inserted • Image ‘x’ having pixels M × N
into the watermarked data of the image. • Pixels intensity with row ‘m’ and column ‘n’ is x (m, n).
• Robustness: After applying several types of attacks, the ability of the • DCT coefficient in the row ‘u’ and column ‘v’ is y (u, v).
host image to withstand against to the non-malicious attacks, so
basically it is showing the resisting power of the watermarked Where,
embedded image under the several image processing.
αu = √1̅̅2 if, u = 0 or αu = 1 if, u = 1, 2, 3, …N-1.
• Mean Square Errors: The function for the MSE written as
and
1∑ n
( )2 αv = √1̅̅2 if, v = 0 or αv = 1 if, v = 1, 2, 3 … N-1.
MSE= Xi − Xi∗ (2)
n i=1 For retrieval of the image, IDCT is applying according to equation as:
√̅̅̅̅̅√̅̅̅̅ M− 1 N − 1 [ ] [ ]
As the MSE function performs the earliest test when the two images are 2 2∑∑ (2m + 1)μπ (2n + 1)vπ
X(m, n) = au av y(u, v)cos cos
very similar [10]. M N u=0 v=0 2N 2N
(4)

Fig. 1. Two phase of digital image watermarking.

2
S. Gupta et al. Measurement: Sensors 24 (2022) 100520

• The wavelet-based functions are also known as mother wavelet

3.2. Discrete wavelet transform (DWT) signal (gѱ)

DWT is useful to subdivide a host image into sub-band of different So, these both orthogonal signals are used as the basis set help to start
resolution as one is low-pass sub-band and three high-pass sub-band the wavelet transformation and a full suite can be achieved by scaling
namely LL, LH, HL, and HH. It can be shown by Fig. 2 below as: and conversion of these both orthogonal functions.
Among this sub-band because of the Human Visual System (HVS), The wavelet transformation equation can be defined by
the best choice is to select HL region to insert the watermark. The ∫∞
discrete wavelet transform calculated as [14]: F(a. b) = f (x)ψ ∗(a,b) (x)dx (10)
− ∞
− j ( )
ψ j,k (t) = a0 ψ
2
a−0 j .t − k.bo (5) Where the functions are.
( −j ) Ѱ* = The complex conjugate.
(6) Ѱ = Wavelet function or known as mother wavelet.
− j
ψ j,k (t) = 2 2 ψ 2 2 .t − k .j
For the discrete time, digital signal and periodic signal f(x), discrete
Fourier transform is utilized. The DFT of the signal with period ‘M’ is:
3.3. Discrete Fourier Transform (DFT) ∑
M− 1
(11)
j2π ux
F(u) = f (x)e M

Required robustness against signal processing attacks provided by x=0

discrete Fourier transform. It decomposes the host image in sine and

And the inverse discrete Fourier transform (IDFT) is
cosine form [15]. The watermarking technique of discrete Fourier
transform into two types, one is directly embedding technology and the 1 ∑
M− 1
(12)
− j2πux
second one is template-based embedding. f (x) =
M x=0
F(u)e M
The global distribution of frequency for the time signal x(t) denoted
as XFT(t) by the Fourier transformation and the original form of this
Where μ, x = 0, 1, 2, ….M-1.
period signal after from this frequency domain signal reconstructed by
using this inverse equation of Fourier transformation.
4. SVD and recent algorithms
∫∞
XFT (f ) = x(t)e− j2πft dt (7)
− ∞ Watermarking digital images with Singular Value Decomposition
∫ (SVD) has become common in the modern day [16].
∞
X(t) = XFT (f)ej2πft df (8)
− ∞
4.1. Singular value decomposition (SVD)
A similar transformation concept known as Short Time Fourier
Transformation (STFT) is to use because it almost overcomes the all
The singular value of the image contains very proficient stability.
limitation come during use of Fourier transformation method. This short
The slight modification does not influence much effect on the quality of
time Fourier transform can reconstruct the time information signal as
the cover image. Because of the most frequent geometric attack, the
well as frequency domain information also from a signal. Along this
largest value of singular value decomposition presents smallest change
attractive feature of this short time Fourier transform, a concept based
[17].
on windowing is used with this Fourier transformation.
For image ‘A,’ with M × N dimension the singular value decompo­
∫∞
sition is defined as [18].
XST FT(τ,f ) = x(t)g∗ (t − τ)e− j2πft dt (9)
(13)
− ∞
A = UΣV T or A = λ1 U1V1T + λ2 U2 V2T + ………λr UrVrT
The Fourier transformation (FT) method was useful for this recon­
struction of time signal x(t) till the wavelet transformation process Where.
developed. It is a local analysis to differentiate all the advantages offered U, V are orthogonal matrix and U, V are left and right singular vec­
by wavelet transformation method over the Fourier transformation tors respectively of image ‘A’.
method it’s mean when analyzing the benefits of wavelet transformation The analysis of the new algorithms can describe many algorithms
method for this reconstruction process of time signal x(t). The wavelet related to the image watermarking. In the literature part of this paper is
transforms signals reveal signals aspects, the breakdown point, as well as showing the achieved advantages during the use of wavelet transform in
discontinuities, arrived but which are the more carefully comparison to the working system with the concept of SVD. A review for some tech­
the acknowledged signal during Fourier transformation. nique used now a day for embedding watermark is explaining signifi­
The two orthogonal functions: cantly, and these are useful for protecting the watermarked host image
from the copyright purpose, and the concept of cryptography method for
• The scaling-based function is also known as father wavelet signal (t) the watermark and its uses for some different coding for that also
exposed as the future work in this paper approached. So, by the addition
of the techniques based on the coding can be used to increase the
robustness of the scheme.

4.2. Fast walsh Hadamard Transform (FWHT)

The matrix of this Hadamard Transform is an orthogonal square

matrix which contains its element value only as 1 or -1, so this type of
transformation known as Walsh-Hadamard Transformation [19].
According to A. Marjuni et al. [20] and Aung et al. [21]. The H1
Fig. 2. DWT decomposition. Hadamard matrix is smallest express as:

3
S. Gupta et al. Measurement: Sensors 24 (2022) 100520

[ ]
1 1 1 second one is feedback system. The feed forward has similarity with
H1 = √̅̅̅ (14)
2 1 − 1 network back propagation, at the input level all the error is to be back
propagating in the BPNN with depicting the input layer, hidden layer,
[ ]
HN− HN− 1 and output layer the whole back propagation network designed as
H2 = H1 ⊗ H2 = 1
(15)
HN− 1 − HN− 1 shown in Fig. 3:
Where at the hidden layer and output layer, the bias has been
When N = 2n, n is an integral no. applying, the value of the bias is always ‘1’. The primary motive to
For example, matrix of H2 for 4 × 4 Hadamard matrix is: design this neural network is to maintain a necessary balance in between
the capability to produce a good a good response to an input layer and to
⎡1 1 1 1 ⎤0
the layer of entry capability for responding correctly. The equation of
1⎢ 1 − 1 1 − 1 ⎥3 BPNN for weight updating are:
H2 = H1 ⊗ H2 = ⎣ ⎦ (16)
2 1 1 − 1 − 1 1 [ ]
Wjk (t + 1) = Wjk (t) + αδkyj + μ Wjk (t) − Wjk (t − 1) (19)
1 − 1 − 1 1 2
[ ]
For retrieval of the watermark the inverse WHT [22] is express as: Vij (t + 1) = Vij (t) + αδjxi + μ Vij (t) − Vij (t − 1) (20)

WHT (x) = X = Hw x (17) Where.

Wjk showing the weight parameter in between hidden layer to output
IWHT (X) = x = Hw X (18) layer.
Vij showing the weight parameter in between input layer to hidden
4.3. Quantization Index Modulation (QIM) layer.
α is the constant of learning rate.
The distortion-compensated QIM is a new type of Quantization Index μ is the parameter of momentum.
Modulation thanks to the embedding method (DC-QIM). This QIM t showing the required time in sec.
approach is better than the additive spread spectrum. The reason to Sk denoting the differential error signal in between hidden unit to
employ DC-QIM method is that DC-QIM gives optimal results for external unit.
Gaussian channel whereas against the limited and bounded perturba­ Sj denoting the error signal in between the hidden unit to input unit
tion, the generalized form of LBM, and in the case of apparent attacks are The formula used for calculating the total no. of the hidden nodes is n
only nearly optimal and the type of dither modulation is used to develop = (om + cm + d)1/2.
realization conveniently [23]. By employing the deterministic model to Where,
illustrate the digital image watermarking method and because of this is n = Total no. of hidden nodes possible.
proved that against arbitrary limitation and attacks, the QIM is good m = Total no. of input nodes in use.
because the dither modulation is practically comparatively the imple­ c and d are determining parameter.
mentation of this QIM method, which has many no. of the attractive In the proposed methodology the value of these parameter taken are
properties of its performance. If this deterministic model used in several n ~ 4.8 (or 5 taken as round off), m = 8, O = 1 and there are 3 layers in
copyright applications, but if in particular style this deterministic model this neural network, the input layer has 8 nodes and hidden layer have 5
used. nodes, whereas in output layer there are 1 node.
Thus, the QIM method gains the ability to conclude self as the uni­
versally good strategy for information embedding in front to a wide 4.5. Dynamic fuzzy interference system (DFIS)
variety of attacker creates distortion intentionally or unintentionally
[24]. The data rate in analog signal near about 1/3/b/s/HZ for each In digital image watermarking the non-zero coefficient can be
drop of decibel achieved by using DC-QIM in an application such as determined and the formula for determining the texture sensitivity is as
broadcasting of digital audio in the category of over analog trans­ [25]:
mission. In the broadcast application of digital over digital communi­ ∑ [( )/ ]
cation as any digital signal transmission like a superposition of Sj = round Tj+key Q (21)
alternative code and decoding approach of successive cancellation, the j

DC-QIM is comparatively safe. In future, if stronger models of attacks For inserting the watermark only on the lower-level frequency
generated then some insights into the embedding information structure
as well as in the fundamental principal and digital watermarking system
will be developed. Thus, at the ending of the spectrum, the data from the
theoretic formulation games and also from the arbitrarily varying type
of channel emerged and seemed to be a universally good starting point
in this embedding respect.

4.4. Back Propagation Neural Network (BPNN)

After applying DWT on the blue plane only of the original host cover
image and on the component of the middle and higher frequency band
watermark should be generated. To get the security about the capacity of
remembering the relational characteristics in the watermark using in
this process and the image where this watermark is to inserted, some
addition information should need for this neural network. A sigmoid
activation function was used in the hidden layer, but a linear activation
function was used in the received output layer. As a result, the BPNN
watermarking concept is an excellent one [25].
There are two classes of the neural network first feed forward and the Fig. 3. Back propagation neural network.

4
S. Gupta et al. Measurement: Sensors 24 (2022) 100520

coefficients of the DWT, which are related to the host signal, a formula 5. Results
can be used as:
( ) The quality of the image is crucial when trying to verify the
∑
′
Tj+key = DFIS round
[(
Tj+key
)/ ]
Q + Xj (22) authenticity of a photograph. Various performance measures and
j benchmarking techniques, such as PSNR, MSE, Euclidean Distance (ED),
SSIM, and the Feature Similarity Indexing Method (FSIM), among
Where. others, are used to assess and compare the effectiveness of watermarked
X j denoting sequence of the related watermark. images. The four measurements that are taken into account are the de­
Q denoting a quantized value. viation from the mean, the root mean square error, the normalized mean
T’ j þkey denoting a coefficient of image having a watermark. cross-correlation (NCC), and the image fidelity (IF) (NMSE), the Corre­
lation Quality (CQ) Index, and some additional comparable metrics. The
4.6. Embedded zero tree wavelet (EZW) ratio of the noisiest peaks to the loudest is known as the peak-to-noise
ratio (PSNR). a comparison of the signal’s intensity and the noise it is
An encoder of EZW is designed to utilize wavelet transformation being contaminated by. The system’s dependability is impacted by the
[26]. The progressive coding is its base to compress host image into few PSNR. The best summary of it can be found in the MSE. The mean
bits stream. The EZW technique can be used to calculate the peak signal squared error (MSE) formula is used in statistics to calculate the average
to noise ratio and the squared error of the mean. There are two steps in squared intensity differences. incorporating the reference watermark as
the encoding process [27]. well as the extracted watermark in its composition. A higher PSNR value
represents a more stable signal. effectiveness of the method; a flawless
• Dominant Pass: It is employed to provide any one of four possible image and an undistorted one appear to be identical. image. The SSIM
combinations as significant negative (SN), significant positive (SP), and FSIM are used to analyse various PSNR and MSE possibilities. By
zero tree root (ZR) and isolated zero (IZ). contrasting the structures and attributes of the recovered images with
• Sub-ordinate Pass: In it the all-relative coefficient encodes by ‘0’ or those of the originals, one can determine the degree of resemblance
‘1’ according to its current threshold value. between the recovered images and the originals. perception. The SSIM is
employed to forecast how well the chromatic (HSV) or colour (RGB)
The concept of EZW can be express by Fig. 4 as: variables will appear to the human eye. Calculating the degree to which
The decoding process also has two passes for redesigning, the decode an ideal image has been altered or deteriorated allows one to determine
value identifying the symbol as negative, positive, isolated or zero trees the RGB or chromatic (YCbCr) values.
[28]. Under encoding stage, the host image data retrieved from the
higher level to lower-level decomposition [29–41]. 6. Conclusion

4.7. Huffman coding Some novel algorithm has been reviewed in this paper to improve the
required PSNR for imperceptibility and NC to get robustness of the
It encodes the secret image data, utilizes for compression of infor­ image. In this paper, we also present some basic evolution of the
mation in a lossless way. It can be express as: watermarking technique which is building the block to get better image
Input: watermarking technique. Many watermarking algorithms and ap­
proaches of the SVD introduce in this paper to increase the quality
•An array of alphabet is the input as A = {a1, a2, a3, ….an}, a symbol parameter of the embedded image against the variety of the signal
alphabet with size ‘n’. processing attacks to achieve the solution for copyright protection and
•Now take w = {w1, w2, w3 … wn}, it is the set of positive symbol authentication of hidden content.
weight 1 ≤i ≤ n.
CRediT authorship contribution statement
Output:
Sunil Gupta: Conception and Design of study, Drafting the manu­
•The relative encode is C (A, w) = {C1, C2, C3 … Cn}, it is the set of script, Approval of the version of the manuscript. Kamal Saluja:
binary bits, here Ci is code word for data ai, 1 ≤ i ≤ n. Conception Design of study, Revising the manuscript critically for
• In the Huffman encode the secret key data and rebuilt the embedded important intellectual content, Approval of the version of the manu­
data into the host image. script. Vikas Solanki: Acquisition of data, Drafting the manuscript,
Approval of the version of the manuscript. Kushwant Kaur: Acquisition
of data, Drafting the manuscript, Approval of the version of the

Fig. 4. Ezw algorithm.

5
S. Gupta et al. Measurement: Sensors 24 (2022) 100520

manuscript. Parveen Singla: Formal analysis, Revising the manuscript [20] A. Marjuni, R. Logeswaran, M.F. Ahmad Fauzi, An image watermarking scheme
based on Flash Walsh Hadamard Transformation and discrete cosine
critically for important intellectual content, Approval of the version of
transformation, in: Proc. of International Conference on Networking and
the manuscript. Mohammad Shahid: Formal analysis, Revising the Information Technology, June 2010, pp. 289–293.
manuscript critically for important intellectual content, Approval of the [21] A. aung, P.N. Boon, S. Rahardja, A robust watermarking scheme using sequence-
version of the manuscript. ordered complex Hadamard transform, J. signal proces. sys. 64 (3) (2010)
319–333.
[22] Aris Marjuni, Mohammad Faizal Ahmad Fauzi, Rajasvaran Logeswaran, Swee-
Declaration of competing interest Huay Heng, in: An Improved DCT-Based Image Watermarking Scheme Using Fast
Walsh Hadamard Transform 5, June 2013, pp. 271–278, 3.
[23] Mohsen Zareian, Hamid Reza Tohidypour, Robust quantization index modulation-
The authors declare that they have no known competing financial based approach for image watermarking, IEEE Trans. Image Process. 7 (5) (2013)
interests or personal relationships that could have appeared to influence 432–441.
the work reported in this paper. [24] Shikha Mittal, Ankit Bansal, Deepali Gupta, Sapna Juneja, Hamza Turabieh,
Mahmoud M. Elarabawy, Ashish Sharma, Zelalem Kiros Bitsue, Using identity-
based cryptography as a foundation for an effective and secure cloud model for E-
Data availability health, 2022, Comput. Intell. Neurosci. (2022), https://doi.org/10.1155/2022/
7016554. Article ID 7016554, 8 pages.
[25] Chin-Ta Yen, Yi-Jie Huang, Frequency domain digital watermark recognition using
No data was used for the research described in the article. image code sequence with a back propagation neural network and dynamic fuzzy
interference system, 16, in: Springer, 75, August 2016, pp. 9745–9755.
References [26] j. Shapiro, Embedded image coding using zero tree of wavelet coefficient, IEEE
Trans. Signal Process. 41 (1993), 34453462.
[27] M. Shravani, B. Krishna Reddy, Image compression using EZW and WDR
[1] Roxana Yahya Abadi, Payman Moallem, Robust and optimum color image
techniques with different wavelet codes, Int. J. Math. Archit. Arvh. 6 (2) (2015).
watermarking method based on a combination of DWT and DCT, Optik 261 (2022),
Feb. 2015.
169146, https://doi.org/10.1016/j.ijleo.2022.169146. ISSN 0030-4026.
[28] S.P. Raja, A. Suruliandi, Performance Evaluation on EZW & WDR Image
[2] Asha Rani, Balasubramanian Raman, An image copyright protection scheme by
Compression Technique, ICCCCT, 2010, pp. 661–664.
encrypting secret data with the host image, 2, in: Springer, Multimedia Tools and
[29] S.H. Wang, Y.P. Lin, Wavelet Tree quantization for copyright protection for
Applications, 75, 2016, pp. 1027–1042.
watermarking, IEEE Trans. Image Process. (2002) 154–165.
[3] Amit Kumar Singh, Mayank Dave, Anand Mohan, Hybrid technique for robust and
[30] J.S. Vitter, Design and analysis of dynamic Huffman codes, J. ACM 34 (4) (1978)
imperceptible multiple watermarking scheme, 14, in: Springer Multimedia Tools
825–845.
and Applications, 75, July 2016, pp. 8381–8401.
[31] Bartosz Boguslawski, Vincent Gripon, Fabrice Seguin, Huffman coding for storing
[4] Kulwinder Singh Mann, Ankit Bansal, HIS integration systems using modality
non-uniformly distribution message in networks of neural cliques, in: Proc. of the
worklist and DICOM”, Procedia Comput. Sci. 37 (2014) 16–23, https://doi.org/
Twenty- Eight AAAI Conference in Artificial Intelligence, 2015, pp. 262–268.
10.1016/j.procs.2014.08.007. ISSN 1877-0509.
[32] K. Saluja, A. Bansal, A. Vajpaye, S. Gupta, A. Anand, Efficient bag of deep visual
[5] Matthieu Urvoy, Dalila Goudia, Florent Autrusseau, Perceptual DFT watermarking
words based features to classify CRC images for colorectal tumor diagnosis, in:
with improved detection and robustness to geometrical distortions, IEEE Trans. Inf.
2022 2nd International Conference on Advance Computing and Innovative
Forensics Secur. 9 (7) (Jul. 2014) 1108–1119.
Technologies in Engineering (ICACITE), Apr. 2022, pp. 1814–1818, https://doi.
[6] R. Sun, H. Sun, T. Yao, A SVD and quantization based semi-fragile watermarking
org/10.1109/ICACITE53722.2022.9823727.
technique for image authentication, in: Proc. Of IEEE International Conference on
[33] S.C. Anita, S. Gupta, An effective model for anomaly IDS to improve the efficiency,
Signal Processing, 2, 2002, pp. 1592–1595.
in: Proceedings of the 2015 International Conference on Green Computing and
[7] Nasrin M Makbol, Khoo Bee Ee, Robust blind image watermarking scheme based
Internet of Things, ICGCIoT 2015, 2016, pp. 190–194, https://doi.org/10.1109/
on redundant discrete wavelet transform and singular value decomposition,
ICGCIoT.2015.7380455.
Elsevier, Int. J. Electr. Commun. , AEU 67 (2013) 102–112.
[34] B.K. Sahoo, A. Sardana, V. Solanki, S. Gupta, K. Saluja, Novel approach of
[8] Sivavenkateswara Rao, V. Rajendra, S. Shekhawat, V.K. Srivastava, A DWT-DCT-
diagnosing significant metrics of load balancing using CloudSim, in: 2022 10th
SVD based digital image watermarking scheme using particle swarm optimization,
International Conference on Emerging Trends in Engineering and Technology -
in: IEEE Students’ Conference on Electrical, Electronics and Computer Science
Signal and Information Processing (ICETET-SIP-22), 2022, pp. 1–6, https://doi.
(SCEECS), Mar. 2012. I-4, 1-2.
org/10.1109/ICETET-SIP-2254415.2022.9791688.
[9] V.M. Potdar, S. Han, E. Chang, A survey on digital watermarking technique, in:
[35] S. Gupta, M. Shahid, A. Goyal, R.K. Saxena, K. Saluja, Black hole detection and
Proc. Of the 3rd IEEE International Conference on Industrial Information (INDIN
prevention using digital signature and SEP in MANET, in: International Conference
2005), 2005, pp. 709–716.
on Emerging Trends in Engineering and Technology, ICETET, 2022, 2022-April,
[10] Gilbert Strang, Truong Nguyen, Wavelets and Filter Banks, Wellesley-Cambridge
https://doi.org/10.1109/ICETET-SIP-2254415.2022.9791738.
Press, Wellesley, MA, 1996.
[36] V. Kukreja, D. Kumar, A. Bansal, V. Solanki, Recognizing wheat aphid disease using
[11] P. Kumasawat, K. Attakitmongcol, A. Srilkaew, A new approach for optimization in
a novel parallel real-time technique based on mask scoring RCNN, in: 2022 2nd
image watermarking by using genetic algorithms, IEEE Trans. Signal Process. 53
International Conference on Advance Computing and Innovative Technologies in
(12) (2005) 4707–4719.
Engineering (ICACITE), Apr. 2022, pp. 1372–1377, https://doi.org/10.1109/
[12] Hina Saxena, Praful Saxena, Shubham Rastogi, DWT-DCT-SVD based semi-blind
ICACITE53722.2022.9823459.
reference image watermarking scheme using trigonometric function, Int. J.
[37] Sunil Gupta, Kamal Saluja, Ankur Goyal, Vajpayee Amit, Vipin Tiwari, Comparing
Concept. Comput. Inform. Tech. 2 (2) (April 2014) 14|29–18|29.
the performance of machine learning algorithms using estimated accuracy,
[13] Saied AmirgholipourKasmani, Ahmadreza Naghsh-Nilchi, A New Robust Digital
Measurement: Sensors 24 (2022), 100432, https://doi.org/10.1016/j.
Image Watermarking Technique Based on Joint DWT-DC Transformation, IEEE
measen.2022.100432. ISSN 2665-9174.
Third International Conference on Convergence and Hybrid Information
[38] Wadhwa, Shivani and Saluja, Kamal and Gupta, Sunil and Gupta, Divya,
Technology, 2008, pp. 539–544.
Blockchain Based Federated Learning Approach for Detection of COVID- 19 Using
[14] Chih-Chin Lai, Cheng-Chih Tsai, Digital image watermarking using discrete
Io MT (July 14, 2022). Available at SSRN: https://ssrn.com/abstract=4159195 or
wavelet transform and singular value decomposition, IEEE Trans. Instrum. Meas.
https://doi.org/10.2139/ssrn.4159195.
59 (II) (Nov. 2010) 3060–3063.
[39] K. Yadav, S. Gupta, N. Gupta, S.L. Gupta, G. Khandelwal, Hybridization of K-means
[15] Chuntao Wang, Jiangqun Ni, Jiwu Huang, An informed watermarking scheme
clustering using different distance function to find the distance among dataset, in:
using HMM in the wavelet domain, IEEE Trans. Inf. Forensics Secur. 7 (3) (June
T. Senjyu, P.N. Mahalle, T. Perumal, A. Joshi (Eds.), Information and
2012).
Communication Technology for Intelligent Systems. ICTIS 2020. Smart Innovation,
[16] Samira Lagzian, Mohsen Soryani, Mahmood Fathy, Robust Watermarking Scheme
Systems and Technologies, 195, Springer, Singapore, 2021, https://doi.org/
Based on RDWT-SVD: Embedding Data in All Sub Bands, IEEE Transaction, 2011,
10.1007/978-981-15-7078-0_29.
pp. 48–52.
[40] D. Hemanand, Nilamadhab Mishra, G. Premalatha, Dinesh Mavaluru,
[17] C.I. Podilchuk, E.J. Delp, Digital Watermarking Algorithm and Applications”, IEEE
Amit Vajpayee, Sumit Kushwaha, Kibebe Sahile, Applications of intelligent model
Signal Processing Magazine, July 2001, pp. 33–46.
to analyze the green finance for environmental development in the context of
[18] M. Nasrin, Makbol, EeKhoo Bee, Robust blind image watermarking scheme based
artificial intelligence, 2022, Comput. Intell. Neurosci. (2022), https://doi.org/
on redundant discrete wavelet transform and singular value decomposition,
10.1155/2022/2977824. Article ID 2977824, 8 pages.
Elsevier, Int. J. Electr. Commun. , AEU 67 (2013) 102–112.
[41] Shyam Nandan Kumar, Vajpayee Amit, ASP: advanced security protocol for
[19] Lusia Rakhmawati, Wirawan Wirawan, Suwadi Suwadi, Delpha Claude,
security and privacy in cloud computing, Am. J. Inform. Sys. 4 (2) (2016) 17–31.
Pierre Duhamel, Blind robust image watermarking based on adaptive embedding
http://pubs.sciepub.com/ajis/4/2/1.
strength and distribution of quantified coefficients, Expert Syst. Appl. 187 (2022),
115906, https://doi.org/10.1016/j.eswa.2021.115906. ISSN 0957-4174.