Devi et al.

SpringerPlus (2016) 5:1091

DOI 10.1186/s40064-016-2733-0

RESEARCH Open Access

A copyright protection scheme

for digital images based on shuffled singular
value decomposition and visual cryptography
B. Pushpa Devi1, Kh. Manglem Singh2* and Sudipta Roy3

*Correspondence:
[email protected] Abstract 
2
National Institute This paper proposes a new watermarking algorithm based on the shuffled singular
of Technology Manipur,
Imphal, India value decomposition and the visual cryptography for copyright protection of digital
Full list of author information images. It generates the ownership and identification shares of the image based on
is available at the end of the visual cryptography. It decomposes the image into low and high frequency sub-bands.
article
The low frequency sub-band is further divided into blocks of same size after shuffling it
and then the singular value decomposition is applied to each randomly selected block.
Shares are generated by comparing one of the elements in the first column of the
left orthogonal matrix with its corresponding element in the right orthogonal matrix
of the singular value decomposition of the block of the low frequency sub-band. The
experimental results show that the proposed scheme clearly verifies the copyright
of the digital images, and is robust to withstand several image processing attacks.
Comparison with the other related visual cryptography-based algorithms reveals that
the proposed method gives better performance. The proposed method is especially
resilient against the rotation attack.
Keywords:  Copyright protection, Singular value decomposition, Orthogonal matrix,
Visual cryptography, Robust

Background
Copyright ownership of multimedia data is vulnerable to the image processing attacks as
it can be copied easily without loss of quality with no limitation on the number of cop-
ies, tempered and redistributed illegally without authorization due to the growth of the
digital multimedia technology tools and omnipresent of the Internet. A good solution
to this problem is to integrate the security information directly to the content of digital
data in inseparable and/or undetectable form during its useful lifespan (Petticolas 1999),
and the digital watermarking is one of such techniques, which embeds the copyright
information such as the watermark to the original digital data to be protected without
degrading it in an imperceptible manner for the ownership proof purpose in such a way
that it is very difficult to detect and remove the information by an unauthorized per-
son. It can be detected or extracted later by the owner to prove his copyright in the case
of legal dispute (Hartung and Kutter 1999). Copyright protection of the digital data is
defined as the process of proving the intellectual property right to a court of law against

© 2016 The Author(s). This article is distributed under the terms of the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium,
provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and
indicate if changes were made.
Devi et al. SpringerPlus (2016) 5:1091 Page 2 of 22

unauthorized reproduction, processing, transformation or broadcasting of digital data

(Ruanaidh and Pun 1998). For copyright-related applications, the watermarked digital
data is expected to be robust to various kinds of geometrical and removal attacks (Cox
and Miller 2002).
Classification of the copyright protection of digital document may be based on types of
data to be watermarked (text, image, audio or video), working domain being used (spa-
tial or transform), information (blind, semi-blind or non-blind), human perception (visi-
ble or invisible), application (source based or destination based), public share (extraction
or detection) etc. (Hwang 2000). Some researchers have proposed detection based on
the visual cryptography (VC) that does not alter the original image in order to preserve
the visual quality of the image, but generates two shares known as the ownership share
and the identification share. The ownership share is generated from the original copy-
righted document and registered to a certified authority (CA). The identification share
is generated from the suspected document. Possessing either one of the shares can not
reveal any information related to the copyright, but stacking of two printed shares on
transparency sheets conveys meaningful details about the copyright information (Chang
et  al. 2002; Hsu and Hou 2005; Singh 2009). Visual cryptography based approach has
attracted for watermarking of sensitive images such as medical images, where alterna-
tion of the pixel values is not permitted (Benyoussef et al. 2015).
Hwang proposed a robust and blind copyright protection scheme based on the visual
cryptography that generates shares comparing the most significant bits (MSB) of the pix-
els in the image with the global mean intensity of the image (Hwang 2000). The probabil-
ity of false alarm is high in his method. MSB based VC method does not work effectively
if the histogram of the grey-level image is either left-skewed or right-skewed (Hassan
and Khalili 2005). To overcome this drawback, a blind and robust watermarking scheme
for copyright protection of the image in spatial domain using visual cryptography is pro-
posed (Hsu and Hou 2005). It generates the ownership share based on the binary secret
message bit, global mean intensity of the image and mean of the neighbouring pixel val-
ues of a randomly selected pixels in the image. Their method is robust to many attacks,
however it is weak to cropping attack. A slightly different approach to the above two
methods is the one that uses the randomly selected pixel value of the image to compare
with the global mean intensity for generation of shares (Singh 2009). A robust and blind
copyright protection scheme based on the visual cryptography is proposed that gener-
ates shares from the product of the normal-distribution random bit and the difference
between the low and middle level wavelet sub-bands (Lou et al. 2007). Their method is
robust to many attacks, but it was proved that the security of their method depends on
the random bit key, but not on the product as the difference between the low wavelet
sub-band and the middle wavelet sub-band is always positive (Chen et al. 2009). Abusitta
proposed a copyright protection scheme of the image based on the relationship between
randomly selected pixels and their 8-neighbours’ pixel and the visual cryptography
(Abusitta 2012). His method is an extension of Hwang’s method (Hwang 2000).
A robust and blind watermarking scheme based on visual cryptography is pro-
posed that generates shares comparing the dc coefficient of the discrete cosine trans-
form (DCT) of the block of size 8  ×  8 of the image with the mean dc coefficients of
blocks from all selected block (Rawat and Raman 2012). Their method is robust to many
Devi et al. SpringerPlus (2016) 5:1091 Page 3 of 22

attacks, but is weak to rotation, cropping, impulse noise, Gaussian noise and sharpen-
ing attacks. Jin and Kim proposed an image watermarking scheme based on the DCT
and the discrete fractional random transform using the visual cryptography (Jin and Kim
2012). A robust and blind watermarking scheme for copyright protection based on the
visual cryptography and the singular values of singular value decomposition (SVD) of
the image is proposed that generates shares comparing the mean of the largest singular
values from each block in the image with the largest singular value of the selected block
(Wang and Chen 2007). The methods mentioned above are robust to many attacks, but
it is possible to reveal the secret message using the unauthorized images. Hossaini et al.
proposed a robust and blind copyright protection scheme based on the visual cryptog-
raphy and the steerable pyramid (Hossaini et al. 2016). Their method is robust against
against different types of attacks.
A robust watermarking scheme is proposed that embeds the principal component of
the watermark of the shuffled SVD (SSVD) of the watermark into the largest singular
value of the image block of the host image (Guo and Prasetyo 2014). Right orthogonal
matrix is kept as the key for the extraction. False alarm of their method is less. It was
reported that the visual quality of the reconstructed image using the SSVD is better than
one that uses the SVD.
Motivated by the above discussion, a robust and blind copyright protection algorithm
based on the SSVD and VC in the DWT domain is proposed. It decomposes the image
into low and high frequency sub-bands and shuffling of pixels is done to the low fre-
quency sub-band. It is then followed by dividing it into similar blocks. Shares are gen-
erated based on the difference between one of the elements in the first column of the
left orthogonal matrix and its corresponding element in the right orthogonal matrix of
the SVD of the blocks in the low frequency sub-band of the image. The experimental
results show that the proposed copyright scheme based on the SSVD and the VC is very
effective.
The rest of the paper is organized as follows. ‘Preliminaries’ section gives brief prelimi-
naries about the cat map transform, discrete wavelet transform, shuffled singular value
decomposition and visual cryptography. ‘Restoration’ section describes the restoration
scheme to restore against the rotation, impulse noise and Gaussian noise attacks. ‘Pro-
posed method’ section describes the proposed method. ‘Experimental results’ section
gives the experimental results, followed by ‘Conclusions’ in the last section.

Preliminaries
This section gives a brief overview of the cat map for image pixel shuffling, discrete
wavelet transform, singular value decomposition and visual cryptography.

Cat map
The Arnold cat map is a chaotic bijection of the unit square onto itself, which is used to
shuffle coordinates (x, y) of the image of size N × N, realizing the effect of image encryp-
tion (Fu et al. 2013). The encryption is very slow if the conventional methods of the cryp-
tography for text documents are used, because the size of an image is comparatively big
(Wang et al. 2009). The new coordinates (x′, y′) on applying the cat map is given as
Devi et al. SpringerPlus (2016) 5:1091 Page 4 of 22

x′
    
1 a x
y′
=
b ab + 1 y
mod N (1)

where a and b are the positive integers, known as the control parameters, and serve as
permutation keys.
The cat map is a periodic process, which returns the original position after P iterations.
This map is area preserving, because the determinant of the transformation matrix is 1.
Pixels move with periodicity, and P, a, b and the original image’s side length N are corre-
lated; thus, whenever the values a, b and N change, it generates a completely different cat
map. For shuffling of non-square image of size M × N, the image is reshaped to a square
image with side length Ns = ceil M × N , where ceil(x) returns the value of x to the
√ 

nearest integer greater than or equal to x (Fu et al. 2013). The insufficient N2s  − M × N
pixels are padded with pseudo random number either 0 or 1 for binary images and in the
range 0–255 for other images. Keshavarzian and Aghagolzadeh opine that the cat map
provides better security due to the increased number of security keys (Keshavarzian and
Aghagolzadeh 2016)

Discrete wavelet transform

Wavelet is a waveform of limited duration that has an average value of zero, and is used
a basal function for representing signals (Mallat 1989). It gives a multiresolution scheme
for image representation using different frequencies at different resolution. DWT divides
an image of size M1 × M2 into four sub-bands LL, LH, HL and HH, where LL sub-band
represents the low frequency of the image and approximation coefficients of DWT, and
LH, HL and HH indicate the high frequency of the image and are known as the horizon-
tal, vertical and diagonal coefficients respectively. These four sub-bands are approxima-
tion, horizontal details, vertical details and diagonal details of the image. One of the next
sub-bands can be further processed to obtain the next scale of wavelet coefficients until
some final scale is reached.

Singular value decomposition

The singular value decomposition is an important topic in linear algebra to diagonalize
and decompose a matrix into its eigenvectors and eigenvalues (Ranade et al. 2007; Liu
and Tan 2002). It has been applied successfully in variety of applications such as data
compression, signal processing, pattern analysis, regression analysis, image steganogra-
phy, watermarking and noise reduction. From the viewpoint of linear algebra, a digital
image is a matrix with non-negative elements. SVD of a matrix H ∈ RM×N is defined as

H = USV T (2)

where U ∈ RM×M and V ∈ RN ×N are the left and the right orthogonal matrices such
that UUT = UTU = IM and VVT = VTV = IN, S ∈ RM×N is the diagonal matrix consist-
ing of the singular values in a non-increasing order of magnitude and the superscript T
denotes transpose operator.
It was reported that the performance of the SSVD is better than that of SVD in the
reconstructed image quality (Guo and Prasetyo 2014). The SSVD is viewed as a pre-pro-
cessing of the SVD by permuting the original image with data-independent permutation.
Devi et al. SpringerPlus (2016) 5:1091 Page 5 of 22

The shuffled image H

−
of the original image H is then fed into the SVD algorithm. The
SSVD can be defined as

S(H ) = H
−
= USV− T (3)

where S{.} denotes the shuffling operator. The shuffling operator produces an ensemble
image as a low resolution sample of the image.
There are some advantages to employ SVD method in many applications:

1. The size of the block of the image for the SVD transformation is not fixed.
2. The singular value (SV) of the SVD represents the intrinsic algebraic image proper-
ties.
3. A small perturbation in the image does not produce large variation in SVs (Wang and
Chen 2007).
4. All elements in the first column of the left orthogonal matrix U are of same sign, and
differences between them are very small (Su et al. 2013).
5. All elements in the first column of the right orthogonal matrix V are of same sign,
and differences between them are also very small.
6. A small perturbation in the first column element of the either left or right orthogonal
matrix of SVD does not give a large variation in the image.
7. The difference between the corresponding elements in the first column of left and
right orthogonal matrices is small.

Such properties can be explored in many copyright protection schemes.

Visual cryptography
Naor and Shamir introduced visual cryptography in their seminal work in which a secret
message is encrypted in a perfectly secure way in more than one shares such that the
secret can be decrypted directly by the human visual system (Naor and Shamir 1995).
Table 1 illustrates how a binary image of size Ns × Ns is divided into two shares of size
2Ns × 2Ns for a 2-out-of-VC, where each pixel of the secret image is expanded to 2 × 2
subpixels in the shares. If a pixel is white in the secret message, the corresponding sub-
pixels in both two shares are identical, one of six columns under the white pixel in the
second and third rows are selected, and the stacked result contains two white subpixels
and two black subpixels. On the contrary, if a pixel is black in the secret image, the cor-
responding subpixels in the first share are complement to those in the same spatial posi-
tion in the second share, and the stacked result contains four black subpixels. Each block
of sub-pixels of size 2 × 2 of the two shares is selected randomly, and so the scheme is

Table 1  A 2-out-of-2 VC
Pixel White Black
Alternative 0 1 2 3 4 5 0 1 2 3 4 5
Share 1
Share 2
Stacked result
Devi et al. SpringerPlus (2016) 5:1091 Page 6 of 22

secure. Possessing of a single share cannot reveal the secret image. Each block of sub-
pixels has six alternative pairs of blocks for both white and black pixel bits.

Restoration scheme against the rotation, impulse noise and Gaussian noise

attacks
The image is passed through a test for the rotation attack. Once the rotation attack is
confirmed, the image is then corrected by an image restoration stage. There are two
types of rotations—loose and crop. The loose type of the rotation produces an output
image large enough to contain the entire rotated image and the crop type produces an
output image the same size as the input image, cropping the rotated image to fit.
Figure 1 shows the rotation attack and the restoration of the rotated image using the
loose type of rotation. The output image after the rotation becomes bigger than the
input image and padding with 0 is done wherever necessary due to the increase in size.
The rotation is confirmed as shown in Fig. 1b, if w = w′ and h = h′, where w and w′ are
displacements in upper left corner toward right and in bottom right corner towards left,
and h and h′ are displacements in upper left corner toward bottom and in bottom right
corner towards top, in the regions with complete dark color. The image can be restored
by using the following equation.

x′
    
cosθ −sinθ x
y′
=
sinθ cosθ y (4)

where θ is the angle of rotation and is found as θ = tan−1 wh , [xy]T are the coordinates
 
 T
of the pixel value of the rotated image and x′ y′ are coordinates of the pixel value of the
corrected image.
The image is rotated by angle of −θ in the restoration stage. The size of the corrected
image before cropping may be big as shown in Fig.  1c as it is padded with zeros all
around. Cropping by removing the padded portion and resizing are done to obtain the
final corrected image as shown in Fig. 1d.
A watermarked image may be attacked by the impulse noise and Gaussian noise
attacks. The performance of the extraction of the watermark can be improved by
smoothing the image with a median filter prior to the extraction (Chang et al. 2014).

Fig. 1  Procedure for the image restoration for the rotation attack a Original image, b after rotation attack,
c after restoration and d after cropping and scaling
Devi et al. SpringerPlus (2016) 5:1091 Page 7 of 22

Proposed technique
In this section, the proposed copyright protection scheme is proposed. In order to
enhance the security and improve robustness of the proposed watermarking scheme,
Arnold cat map is applied to both the image and the watermark (Keshavarzian and Agh-
agolzadeh 2016). The scheme is divided into two phases: ownership share construction
and identification share construction. During the ownership share construction, one of
the channels of the color image is used for generation of the share. Padding with the
pseudo random number in the appropriate range is done prior to the further processing
if the original host image is not square. The selected channel is decomposed first using
the DWT into four sub-bands, and the sub-band LL is used for generation of the shares.
The sub-band LL is least effected by any kind of noise suffered by the image (Rani et al.
2015). The ownership share is generated from the LL image block of the selected channel
of the image by comparing one of the elements in the first column of the left orthogo-
nal matrix with the corresponding element in the right orthogonal matrix of the SSVD
of the image block. The share generations based on SSVD–VC in DWT domain are as
follows.

Ownership share generation scheme

Let H be a greyscale image or the selected channel of the host color image of size
M1  ×  M2, W be the binary watermark of size Ns  ×  Ns, a andb be the control param-
eters of the cat map for shuffling of pixel coordinates of sub-band LL and encrypting
the watermark, P2 and P1 be the periods of the cat map for the low sub-band LL of the
image and the watermark respectively, K be a private key for selecting the block Bi, and
C1 be the codebook as shown in Table 2. Figure 2 shows the schematic diagram of the
proposed ownership share generation and identification share generation scheme given.
Steps for the ownership share generation are given below.

O1. Perform 1-level DWT on the image H of size M1 × M2 to obtain four sub-bands
LL, LH, HL and HH.
If the image is not square, padding is done using the pseudo random numbers prior to
DWT operation to make a square image. Let the size of each sub-band be Ms × Ms.
O2. Apply the cat map on the watermark W of size Ns × Ns N1 times using the control
parameters a, b and period P1 (where N1 < P1 ). Store the encrypted watermark in an
array.
O3. Generate a list of random numbers { i|such that total number of random numbers =
Ns × Ns } using pseudo random number generator (PRNG) with the private key K.
O4. Apply the cat map to the sub-band LL, N2 times using the control parameters a, b
and period P2 (where N2 < P2 ). Then divide the encrypted sub-band into several non-
overlapping blocks of size 4 × 4.

Table 2  Codebook C 1 for generation of ownership share

Feature | 4,1 |<| 4,1 | | 4,1 |≥| 4,1 |
Watermark bit 0 1 0 1
mod( , 6) = 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Ownership share
Devi et al. SpringerPlus (2016) 5:1091 Page 8 of 22

, , ,

Image H Perform LL LH Encrypt- cat Block Extracted

DWT map on LL paron watermark
HL HH

Watermark Encrypt- cat Covert to Generate Perform Decrypt-cat

W map on W 1D array Ownership SVD on map
share ,
,
, , ,
Ownership Perform
share Reducon

Stack
, , ,

Watermark Perform LL ‘ LH’ Encrypt- cat Generate Idenficaon

ed image H’ DWT map on LL’ identification share
HL’ HH’ share

Block Perform
paron SVD on

Fig. 2  Schematic block diagram of the proposed embedding and extraction scheme

O5. Perform the SVD on a randomly selected block B −

(i denotes the block number)
i
generating the following left orthogonal, singular and right orthogonal matrices.

B
−
=U S VT
− − − (5)
i i i i

where

u1,1 u1,2 u1,3 u1,4

 
 u2,1 u2,2 u2,3 u2,4 
U = ,
−
i u3,1 u3,2 u3,3 u3,4 
u4,1 u4,2 u4,3 u4,4

s1,1 0 0 0
 
 0 s2,2 0 0 
S− = 
i 0 0 s3,3 0 
0 0 0 s4,4

and

v1,1 v1,2 v1,3 v1,4

 
 v2,1 v2,2 v2,3 v2,4 
V− = 
i v3,1 v3,2 v3,3 v3,4 
v4,1 v4,2 v4,3 v4,4

O6. Construct the ownership share block oi based on the feature value
u4,1  < v4,1  or u4,1  ≥ |v4,1 | , shuffled watermark bit and mod(i, 6), as shown in
      
Devi et al. SpringerPlus (2016) 5:1091 Page 9 of 22

the codebook C1 of Table 2. mod(i, 6) is used to select one of the alternative sub-pixel
blocks of the binary bits from the codebook.
O7. Repeat Steps O5–O6 until all the encrypted watermark bits are exhausted. Finally,
all the ownership share blocks are combined to form the ownership share O.

After the construction of the ownership share, the watermark, the private key K, the
control parameters a, b, the periods P1, P2 and the numbers of times of shuffling N1, N2
must be kept secretly by the copyright owner, and the ownership share O should be reg-
istered to a CA for further authentication.

Identification share generation scheme

Steps of the identification share generation are described below.

I1. Perform 1-level DWT on the possibly attacked image H′ of size M1 × M2 to obtain
four sub-bands LL′ , LH ′ , HL′ and HH′. If the image is not square, padding is done using
the pseudo random numbers prior to the DWT operation to make a square image. The
size of each sub-band is Ms × Ms.
I2. Generate a list of random numbers { i|such that total number of random numbers
Ns × Ns } using pseudo random number generator (PRNG) with the private key K.
I3. Apply the cat map on the sub-band LL′, N2 times using the control parameters a, b
and period P2 (whereN2 < P2 ). Then divide the encrypted sub-band into several non-
overlapping blocks of size 4 × 4.
I4. Perform the SVD on a randomly selected block B −
′
(i denotes the block number) gen-
i
erating the left orthogonal, singular and right orthogonal matrices in Eq. 6.

′ ′ ′ ′T
B
−
=U SV
− − − (6)
i i i i

where
 ′ ′ ′ ′ 
u1,1 u1,2 u1,3 u1,4
 u′ ′
u2,2
′
u2,3
′
u2,4 
′
U =  2,1 ,
 
− ′ ′ ′ ′
i  u3,1 u3,2 u3,3 u3,4 
′ ′ ′ ′
u4,1 u4,2 u4,3 u4,4

 ′ 
s1,1 0 0 0
 0 ′
s2,2  0 0
S− ′ = 
 
′
 0 0 s3,3 0

i 
′
0 0 0 s4,4
and
 ′ ′ ′ ′ 
v1,1 v1,2 v1,3 v1,4
 v′ ′
v2,2
′
v2,3
′
v2,4 
V− ′ =  2,1
 
′ ′ ′ ′
 v3,1 v3,2 v3,3 v3,4 

i
′ ′ ′ ′
v4,1 v4,2 v4,3 v4,4
Devi et al. SpringerPlus (2016) 5:1091 Page 10 of 22

Table 3  Codebook C 2 for generation of identification share

Feature | ′
4,1 | <| ′
4,1 | | 4,1 |
′ ≥| 4,1 |
′

mod( , 6) = 0 1 2 3 4 5 0 1 2 3 4 5
Identification share

I5. Construct the identification share block di based on the feature value
u  < v  or u  ≥ v  and mod(i, 6), as shown in the codebook C of Table 3.
 ′   ′   ′   ′ 
4,1 4,1 4,1 4,1 2
I6. Repeat Steps I4–I5 until all the Ns × Ns blocks are used up from the host image H′.
Finally, all the identification share blocks are combined to form the identification share
D.
I7. Retrieve the secret image W′ of size 2Ns × 2Ns by stacking the ownership share O
and the identification share D.
I8. Divide the retrieved secret image W′ into non-overlapping 2  ×  2 blocks s′j,k
(1 ≤ j, k ≤ 2).
I9. Perform the reduction process to obtain a reduced secret image W′′ of size Ns × Ns
by the following rules:

 �� ′
 1,
 if sj,k ≥ 2
j k
w= ′
(7)
� �
 0,
 if sj,k < 2
j k

where w is a binary pixel in W′′.

I10. Scramble the watermark W′′ by the cat map (P1  −  N1) times using the control
parameters a and b to obtain the descrambled watermark W′′′.

Experimental results
A set of experiments was performed to verify the robustness of the proposed copy-
right protection algorithm using several images and a binary watermark. Representative
images are shown in Fig.  3. The images are Lena, Mandrill, Building, Aptus, Goldhill,
Zelda, Airplane, Barbara, Tiffany, Girl and Brain of size 512 × 512 (Sipi, Imagecompres-
sion, Cipr, Hlevkin). The original binary watermark of size 64 × 64 is shown in Fig. 4a.
The proposed method (PM) is compared with other popular methods in transform
domain such as Lou et  al. method (LM) (Lou et  al. 2007), Wang et  al. method (WM)
(Wang and Chen 2007) and Rawat et  al. method (RM) (Rawat and Raman 2012) that
use VC. To evaluate the robustness of the proposed method, the proposed method was
tested using ten different types of attacks: JPEG compression (JP), rotation (RO), median
filtering (MF), cropping (CR), scaling (SC), impulse noise (IN), blurring (BL), Gaussian
noise (GN), sharpening (SH) and Gamma correction (GC). The normalized correlation
(NC) is used to measure the similarity between the original watermark and the revealed
watermark. It is defined as

Ns Ns ⌢
m=1 n=1 W (m, n) ⊕ W (m, n) (8)
NC =
Ns × Ns
Devi et al. SpringerPlus (2016) 5:1091 Page 11 of 22

Fig. 3  Representative images: a Lena, b Mandrill, c Building, d Aptus, e Goldhill, f Zelda, g Airplane, h Bar-
bara, i Tiffany, j Girl and k Brain

⌢
where W(m, n) and W (m, n) represent the original secret image and the detected secret
image respectively, ⊕ denotes the exclusive-or (XOR) operation and Ns × Ns is the size
of the secret image.
PSNR is used to measure the quality of the attacked image. It is given by

2552
PSNR = 10log10 (9)
MSE

where MSE stands for mean squared error between the original image and the attacked
image.

Example of cat map encryption

An example of cat map encryption of an image of size 8 × 8 is shown in Fig. 4. Original
pixel values are shown in Fig. 4a. Values of control parameters a = 3 and b = 2 are con-
sidered in this example. Figure 4b–e are the encrypted pixel values for N2 = 1, 2, 3 and
4, where N2 is the number of iterations. The period P2 is found to be 4 for these param-
eters. Figure 4d is the encrypted pixel values after 3 iterations and it should be further
encrypted 1 time to get the decrypted image as shown in Fig. 4e. Periods for different
combinations of a and b such as 1 & 1, 1 & 2, 1 & 3, 1 & 4, 2 & 1, 2 & 2, 2 & 3, 2 & 4, 3 &
1, 3 & 2, 3 & 3, 3 & 4, 4 & 1, 4 & 2, 4 & 3 and 4 & 4 are 6, 8, 6,16, 8, 4, 4, 8, 6, 4, 6, 16, 16,
8, 16 and 8 respectively.
The coordinate (0, 0)′ having the pixel value of 150 will follow the path shown below
for control parameters a = 3 and b = 2 and image size of 8 × 8 to return to the original
position. It is shown below.
         
0 4 3 7 0
→ → → →
0 1 0 7 0

The original coordinate returns to initial position after 4 iterations. In general, it is

claimed that as the value of image size increases, the period tends to increase, but it is
not always true (Pages.physics).

Fig. 4  Example of cat map encryption: a Pixel values of an image of size 8 × 8, b Encrypted pixels with
N2 = 1, c Encrypted pixels with N2 = 2, d Encrypted pixels with N2 = 3 and e Encrypted pixels with N2 = 4
Devi et al. SpringerPlus (2016) 5:1091 Page 12 of 22

Assessment of robustness
Table 4 shows the robustness test of PM on different types of attacks such as JP attack
for quality (Q) from 40 to 90, RO attack for angle (A) of 1°, 2°, 3°, 4°, 5° and 6°, MF
attack for window size (ws) of 2 × 2, 3 × 3, 4 × 4, 5 × 5, 6 × 6 and 7 × 7, CR attack for
cropping percentage (C) of 10, 20, 30, 40, 50 and 60  %, SC for scaling factor (F) from
2 × 2, 3 × 3, 4 × 4, 5 × 5, 6 × 6 and 7 × 7, IN attack for impulse noise ratio (R) of 0.05,
0.10, 0.15, 0.20, 0.25 and 0.30, GN attack for zero mean and variance (V) from 0.01 to
0.10, BL for sigma (ζ) of 0.1, 0.2, 0.3, 0.4, 0.5 and 0.6, SH attack for alpha (α) from 0.1 to
1.0 and GC attack for gamma (G) from 0.6 to 1.5. It was found that the performance of
PM is very good for JP, RO, MF, SC, IN, BL, GN, SH and GC attacks on different types of
images for various ranges and NCs are above 90 % on different values of attacks. The NC
values are between 70 and 90 % for CR attack. This shows that PM is robust.

Comparison with other methods

Table 5 shows comparison of the proposed method with LM, WM and RM on six differ-
ent images for ten different attacks. The unweighted average (UA) is also shown in the
table. The table clearly shows the superiority of PM to LM, WM and RM in term of NC
for JP, RO, MF, SC, IN, BL and GN attacks on different images. LM gives slightly better
results for CR, SH and GC attacks. Figure 5 is the graphical comparison of PM, LM, WM
and RM for ten attacks on Lena image. The unweighted average in Table 5 and Figs. 5, 6
show the superior performance of PM on different images for different parameters.

No attack
Figure 7 shows the various steps to establish the ownership of the copyright. The original
binary watermark is encrypted by using the cat map and is shown in Fig. 7b. The owner-
ship share is prepared based on the encrypted watermark and the original image and it is
shown in Fig. 7c. The identification share is prepared from the watermarked image and
is shown in Fig. 6d. The superimposed image of the ownership share and the identifica-
tion share is shown in Fig. 7e. It is blurred and not recognizable. The reduction superim-
posed image before the decryption is shown in Fig. 7f. It is blurred and not recognizable.
Figure 7g shows the decrypted watermark, which is exactly similar to the original water-
mark. NC value of the extracted is 100 and PSNR value of the watermarked image is
infinite for no attack.

JPEG compression attack

Figure 8 shows the quality of the extracted watermark. NC values of LM, WM, RM and
PM for the JPEG compression attack for Q of 90 are 97.72, 97.09, 97.72 and 99.31 respec-
tively with PSNR value of 39.48  dB on Lena image. NC values are 93.75, 95.50, 96.11
and 98.04 respectively with PSNR value of 33.93 dB for LM, WM, RM and PM on Lena
image for Q of 50. It clearly shows the superior performance for the JPEG compression
attack.

Rotation attack
Figure  8 shows the quality of the extracted watermark. NC values of LM, WM, RM
and PM for the rotation attack for an angle of an 1° are 79.78, 84.93, 85.49 and 93.82
Devi et al. SpringerPlus (2016) 5:1091 Page 13 of 22

Table 4  Robustness test on different images (in NC)

Attack Lena Mandrill Building Aptus Goldhill Zelda Airplane Barbara Tiffany Girl Brain

JPEG Co-
mpres-
sion
Q = 40 97.99 98.58 98.36 97.97 99.34 98.63 97.90 98.24 98.26 98.58 98.73
Q = 50 98.04 98.60 98.77 97.90 99.46 98.70 97.97 98.80 98.43 98.63 98.90
Q = 60 98.24 99.02 98.92 98.55 99.24 98.99 98.33 98.73 98.38 98.80 99.07
Q = 70 98.73 99.36 99.09 98.95 99.48 99.09 98.87 98.95 98.65 98.90 99.12
Q = 80 98.97 99.48 99.26 99.09 99.68 99.38 98.99 99.34 99.02 98.95 99.53
Q = 90 99.31 99.87 99.63 99.63 99.73 99.41 99.34 99.60 99.38 99.43 99.63
Rotation
A = 1° 93.82 93.75 94.36 92.50 98.26 96.28 93.31 95.45 92.43 92.55 96.46
A = 2° 93.92 93.62 93.23 90.84 97.87 96.31 93.33 95.50 92.40 92.77 96.43
A = 3° 95.92 95.33 95.58 93.85 98.21 97.80 95.80 96.24 95.67 96.92 96.41
A = 4° 95.80 95.45 95.72 93.55 97.99 97.70 95.77 96.24 95.67 96.99 96.89
A = 5° 95.89 95.50 95.62 93.75 97.97 97.66 95.87 96.02 95.77 97.02 96.85
A = 60 96.04 95.45 95.50 93.65 98.26 97.65 95.87 96.31 95.55 97.07 96.45
Median
filter
ws = 2 × 2 96.11 95.89 95.97 93.72 98.31 98.02 96.36 95.99 96.48 96.43 96.58
ws = 3 × 3 97.90 96.60 97.77 95.89 99.12 98.87 98.73 97.11 97.43 99.14 98.09
ws = 4 × 4 95.62 94.72 95.23 91.72 98.07 97.58 95.89 95.33 95.31 96.14 96.09
ws = 5 × 5 96.06 94.11 95.23 91.35 98.51 98.07 97.24 95.23 96.31 98.02 96.14
ws = 6 × 6 94.72 93.23 93.62 88.91 97.70 97.16 95.14 94.70 94.84 95.60 95.96
ws = 7 × 7 94.84 92.84 93.89 88.96 98.07 97.48 95.77 94.92 95.62 97.14 94.79
Cropping
%C = 10 86.10 93.03 88.84 89.72 93.79 82.95 80.85 89.86 85.98 79.17 93.77
%C = 20 82.61 89.52 83.76 84.44 89.69 79.68 73.33 83.32 79.10 72.63 87.79
%C = 30 81.56 88.20 79.80 80.15 85.72 77.97 71.48 77.09 77.07 70.77 85.62
%C = 40 78.97 80.17 74.68 78.24 80.54 76.09 71.92 72.46 77.26 70.16 81.32
%C = 50 76.09 77.97 72.36 76.24 76.39 75.46 72.11 71.24 73.48 69.67 78.19
%C = 60 73.33 73.99 69.14 75.46 72.80 74.21 72.85 69.50 74.90 69.67 75.48
Scaling
F = 2 × 2 97.85 96.89 97.80 95.99 99.09 98.99 98.41 97.46 97.92 99.16 98.07
F = 3 × 3 96.16 94.60 95.12 91.82 98.19 98.04 96.67 96.04 96.67 97.63 98.31
F = 4 × 4 94.89 93.43 93.79 89.99 97.58 97.14 95.28 95.45 95.89 96.43 95.14
F = 5 × 5 94.16 92.50 93.28 88.74 97.11 97.04 94.14 94.97 95.04 95.92 93.84
F = 6 × 6 93.55 91.99 92.18 87.35 96.58 96.63 93.28 94.60 94.75 95.28 93.26
F = 7 × 7 92.91 91.87 91.11 86.71 96.14 96.28 92.79 94.21 94.33 95.14 92.57
Impulse
Noise
R = .05 97.72 96.41 97.75 95.55 99.09 98.75 98.51 96.53 97.38 99.07 97.80
R = .10 97.41 96.04 97.60 95.84 98.77 98.92 98.16 96.43 97.14 99.02 97.82
R = .15 97.07 95.89 96.85 94.45 98.87 98.33 97.65 96.38 96.77 98.19 97.16
R = .20 96.43 95.21 96.04 93.67 97.99 97.77 96.82 95.31 95.65 97.14 97.11
R = .25 95.43 94.36 95.23 92.26 98.07 96.75 95.89 94.87 94.84 95.99 96.24
R = .30 93.89 93.96 93.82 91.72 97.11 95.87 94.01 94.01 92.89 93.89 95.48
Blurring
ζ = 0.1 100 100 100 100 100 100 100 100 100 100 100
ζ = 0.2 100 100 100 100 100 100 100 100 100 100 100
ζ = 0.3 100 100 99.97 99.97 100 100 100 99.95 99.97 100 99.92
Devi et al. SpringerPlus (2016) 5:1091 Page 14 of 22

Table 4  continued
Attack Lena Mandrill Building Aptus Goldhill Zelda Airplane Barbara Tiffany Girl Brain

ζ = 0.4 99.56 99.53 99.65 99.43 99.82 99.82 99.78 99.65 99.87 99.92 99.65
ζ = 0.5 99.29 98.87 99.16 98.60 99.58 99.48 99.34 99.02 99.48 99.63 99.26
ζ = 0.6 98.85 98.21 98.70 97.68 99.41 99.38 99.09 98.41 99.36 99.36 98.73
Gaussian
noise
V = .01 93.18 95.16 94.77 91.16 97.55 95.77 92.67 95.04 91.94 91.94 96.26
V = .02 91.18 93.89 93.21 89.37 96.82 93.56 90.94 93.75 89.33 88.96 95.31
V = .03 89.81 93.45 91.91 86.93 95.99 92.18 88.59 92.30 87.13 87.40 94.77
V = .04 88.47 92.40 91.82 86.79 95.45 91.11 87.54 91.91 86.27 87.40 93.67
V = .05 87.40 91.89 90.50 86.54 94.75 90.25 86.91 91.77 84.57 85.83 93.01
V = .06 87.08 90.79 90.01 85.67 93.87 89.96 86.32 91.13 84.49 84.93 92.67
V = .07 85.67 91.47 88.74 84.27 93.89 88.84 85.32 90.06 83.05 82.73 91.50
V = .08 85.13 90.33 87.91 84.54 93.96 87.45 84.27 90.11 82.91 82.15 91.33
V = .09 84.93 90.08 87.15 82.44 93.35 87.62 84.03 89.64 80.90 82.27 91.67
V = .10 84.30 89.47 86.86 81.86 92.72 87.81 84.15 89.16 81.20 82.22 90.99
Sharpening
α = .1 92.08 92.33 92.79 91.33 96.58 95.70 93.57 92.48 93.11 95.48 94.79
α = .2 92.28 92.40 93.04 91.62 96.53 95.72 93.70 92.74 93.11 95.41 94.72
α = .3 92.28 92.60 93.23 91.74 96.65 95.80 93.87 92.94 93.13 95.48 94.77
α = .4 92.40 92.79 93.21 91.79 96.60 95.70 93.94 93.13 93.16 95.53 94.94
α = .5 92.40 92.96 93.28 91.89 96.67 95.70 93.92 93.26 93.23 95.55 95.04
α = .6 92.40 92.99 93.31 92.08 96.72 95.80 93.92 93.40 93.31 95.65 95.09
α = .7 92.45 93.13 93.35 92.21 96.75 95.75 94.04 93.60 93.45 95.60 95.11
α = .8 92.48 93.21 93.38 92.16 96.67 95.82 94.09 93.65 93.43 95.65 95.09
α = .9 92.57 93.21 93.45 92.23 96.70 95.80 94.14 93.65 93.57 95.67 95.14
α = 1 92.67 93.33 93.55 92.26 96.75 95.77 94.16 93.65 93.53 95.70 95.10
Gamma
correction
G = .6 97.60 96.80 97.24 97.46 97.29 95.99 99.16 97.02 99.12 98.75 93.33
G = .7 98.29 97.68 97.92 98.02 98.09 97.21 99.31 98.02 99.24 99.04 97.48
G = .8 98.87 98.36 98.51 98.75 98.75 98.02 99.48 98.65 99.53 99.19 98.36
G = .9 99.41 99.21 99.36 99.56 99.48 98.99 99.75 99.38 99.68 99.60 99.24
G = 1 100 100 100 100 100 100 100 100 100 100 100
G = 1.1 99.07 99.31 99.21 99.41 99.43 98.99 99.68 99.36 99.78 99.63 99.29
G = 1.2 98.46 98.73 98.63 98.87 98.87 98.02 99.60 98.85 99.73 99.56 98.75
G = 1.3 98.02 97.97 97.99 98.29 98.43 97.38 99.31 97.94 99.43 99.34 98.09
G = 1.4 97.33 97.41 97.53 97.92 98.09 96.65 99.21 97.50 99.16 99.12 97.43
G = 1.5 96.94 96.99 96.89 97.11 97.87 96.02 98.82 96.94 98.92 99.07 96.87

respectively with PSNR values of 21.01  dB for LM, WM and RM respectively and
25.09 dB for PM on Lena image. NC values are 77.29, 73.19, 73.53 and 95.92 respectively
with PSNR value of 16.38 dB for LM, WM and RM and 29.29 dB for PM on Lena image
for an angle of 3°. It clearly shows the superior performance for the rotation attack.

Median filter attack

Figure 8 shows the quality of the extracted watermark. NC values of LM, WM, RM and
PM for the median filter attack for window size of 3 × 3 are 94.60, 94.99, 95.14 and 97.90
respectively with PSNR value of 36.88  dB on Lena image. NC values are 92.62, 93.11,
Devi et al. SpringerPlus (2016) 5:1091 Page 15 of 22

Table 5  Comparison with other methods

Attack LM WM
Lena Build- Gol- Air- Tif- Brain Lena Build- Gol- Air- Tif- Brain
ing dhill plane fany ing dhill plane fany

JP
Q = 50 93.75 95.84 94.43 92.18 91.94 95.04 95.50 95.45 95.87 96.28 93.50 97.38
Q = 70 94.79 97.41 96.60 93.65 94.53 95.87 96.41 96.36 96.33 97.09 94.70 97.87
Q = 90 97.72 98.75 98.38 96.60 97.14 97.70 97.09 98.24 98.14 98.26 96.77 98.73
RO
A = 1° 79.78 79.19 80.46 79.98 81.49 83.96 84.93 79.29 84.79 79.54 79.12 84.61
A = 3° 77.29 78.32 78.83 78.54 77.95 78.63 77.39 73.19 73.16 73.31 69.75 75.10
A = 5° 79.37 77.73 78.63 78.34 77.46 78.41 73.19 67.04 68.60 70.50 67.43 70.41
MF
ws = 3 × 3 94.60 95.94 95.58 95.89 94.23 97.11 94.99 94.06 94.82 96.09 91.25 94.75
ws = 5 × 5 92.62 92.57 93.11 93.43 91.57 94.11 93.11 90.28 92.99 94.04 88.54 91.77
ws = 7 × 7 89.99 87.93 89.69 90.01 88.35 90.52 92.01 88.35 92.13 91.87 87.54 89.99
CR
%C = 20 95.53 94.45 94.84 95.31 95.62 95.01 87.52 80.88 79.49 72.72 63.62 83.66
%C = 40 90.77 90.62 90.16 91.08 91.45 91.18 83.44 78.44 77.90 72.70 61.23 82.15
%C = 60 87.32 86.32 86.10 86.74 86.69 86.96 80.00 73.53 72.72 70.06 72.97 77.41
SC
F = 2 × 2 96.70 97.75 96.92 95.99 96.33 97.29 94.79 92.94 94.65 94.43 91.16 94.31
F = 4 × 4 90.45 90.40 89.52 88.76 88.89 90.42 92.08 88.59 91.94 91.28 88.08 90.84
F = 6 × 6 83.52 84.35 83.66 83.08 83.49 84.98 90.50 85.76 90.03 88.84 87.06 87.86
IN
R = .05 84.91 86.49 85.79 84.35 83.32 96.70 80.22 79.49 79.32 79.10 76.90 94.53
R = .15 81.71 83.10 83.17 82.00 80.78 94.79 74.04 73.77 75.19 74.38 72.14 93.31
R = .25 80.27 82.34 81.25 81.03 80.46 91.74 71.14 70.26 70.99 71.24 70.31 90.89
BL
ζ = 0.2 100 100 100 100 100 100 100 100 100 100 100 100
ζ = 0.4 99.34 99.56 99.31 99.14 99.21 99.80 98.82 99.16 99.19 99.29 98.68 99.19
ζ = 0.6 98.14 98.75 98.43 98.14 97.97 99.12 96.77 96.46 96.67 97.26 95.33 97.19
GN
V = .01 84.57 87.59 85.96 84.25 83.66 89.28 84.42 84.91 85.15 82.71 78.93 92.60
V = .03 83.25 84.13 83.22 83.78 80.68 87.57 78.73 79.51 79.29 77.49 73.46 89.18
V = .05 82.93 83.83 81.90 82.03 80.90 86.76 74.53 76.34 75.75 75.09 72.09 87.89
SH
α = .1 95.19 95.67 95.99 95.33 94.99 96.41 86.05 87.15 86.81 89.74 84.35 89.52
α = .3 95.50 95.72 96.26 95.26 95.38 96.58 87.23 87.81 87.52 90.16 85.64 90.06
α = .5 95.28 95.87 96.41 95.53 95.58 96.75 87.84 88.01 87.67 90.60 86.27 90.23
GC
G = .8 99.56 99.29 99.12 99.02 99.16 99.38 98.92 97.53 98.16 99.21 98.92 97.65
G = 1 100 100 100 100 100 100 100 100 100 100 100 100
G = 1.2 99.48 99.16 99.29 99.29 99.21 98.73 98.85 97.87 98.02 99.34 99.07 97.75
UA 90.81 91.30 91.10 90.62 90.28 93.02 88.35 86.68 87.44 87.08 84.16 90.89
RM PM

JP
Q = 50 96.11 95.31 95.92 96.31 94.14 96.63 98.04 98.77 99.46 97.97 98.43 98.90
Q = 70 96.48 96.75 96.67 96.97 94.89 97.43 98.73 99.09 99.48 98.87 98.65 99.12
Q = 90 97.72 97.92 97.65 98.14 96.80 98.58 99.31 99.63 99.73 99.34 99.38 99.63
Devi et al. SpringerPlus (2016) 5:1091 Page 16 of 22

Table 5  continued
RM PM
RO
A = 1° 85.49 78.88 84.93 79.27 77.49 84.83 93.82 94.36 98.26 93.31 92.43 96.46
A = 3° 77.34 72.26 73.77 72.46 68.48 76.41 95.92 95.58 98.21 95.80 95.67 96.41
A = 5° 73.53 66.62 69.92 70.21 66.47 70.31 95.89 95.62 97.97 95.87 95.77 96.85
MF
ws = 3 × 3 95.14 93.75 94.99 96.38 91.62 94.97 97.90 97.77 99.12 98.73 97.43 98.09
ws = 5 × 5 93.79 90.57 92.99 94.21 88.79 92.30 96.06 95.23 98.51 97.24 96.31 96.14
ws = 7 × 7 92.79 88.89 91.87 92.50 87.25 90.42 94.84 93.89 98.07 95.77 95.62 94.70
CR
%C = 20 84.81 79.41 78.14 72.43 64.74 83.10 82.61 83.76 89.69 73.33 79.10 87.79
%C = 40 84.20 79.24 80.49 74.58 62.67 85.18 78.97 74.68 80.54 71.92 76.26 81.32
%C = 60 78.63 72.80 71.77 68.60 70.09 75.34 73.33 69.14 72.80 72.85 74.90 75.48
SC
F = 2 × 2 95.04 92.55 94.67 94.48 91.74 94.26 97.85 97.80 99.09 98.41 97.92 98.07
F = 4 × 4 92.77 88.23 92.11 90.91 88.79 90.60 94.89 93.79 97.58 95.28 95.89 95.14
F = 6 × 6 91.08 85.40 90.42 88.67 86.96 88.57 93.55 92.18 96.58 93.28 94.75 93.26
IN
R = .05 82.25 81.05 82.78 79.00 76.53 94.92 97.72 97.75 95.55 98.75 97.38 97.80
R = .15 75.36 76.00 76.17 73.65 72.90 93.67 97.07 96.85 94.45 98.33 96.77 97.16
R = .25 72.09 72.38 72.41 70.89 69.77 91.47 95.43 95.23 92.26 96.75 94.84 96.24
BL
ζ = 0.2 100 100 100 100 100 100 100 100 100 100 100 100
ζ = 0.4 99.19 99.19 99.26 99.41 98.97 99.02 99.56 99.65 99.82 99.78 99.87 99.65
ζ = 0.6 96.89 96.36 96.92 97.43 95.72 96.80 98.85 98.70 99.41 99.09 99.36 98.73
GN
V = .01 83.74 84.91 85.64 83.20 77.90 92.30 93.18 94.77 97.55 92.67 91.94 96.26
V = .03 77.61 78.78 79.02 76.63 73.02 89.91 89.81 91.91 95.99 88.59 87.13 94.77
V = .05 75.56 76.12 77.51 76.02 71.87 87.40 87.40 90.50 94.75 86.91 84.57 93.01
SH
α = .1 86.88 88.03 87.37 89.06 84.17 90.45 92.08 92.79 96.58 93.57 93.11 94.79
α = .3 87.74 88.40 87.79 89.79 85.74 91.04 92.28 93.23 96.65 93.87 93.13 94.77
α = .5 88.69 88.62 88.13 90.35 86.79 90.89 92.40 93.28 96.67 93.92 93.23 95.04
GC
G = .8 99.21 98.21 98.46 99.26 98.90 97.80 98.87 98.51 98.75 99.48 99.53 98.36
G = 1 100 100 100 100 100 100 100 100 100 100 100 100
G = 1.2 99.29 98.36 98.60 99.36 99.09 98.16 98.46 98.63 98.87 99.60 99.73 98.75
UA 88.64 86.83 87.87 87.00 84.07 91.09 94.16 94.10 96.07 93.97 93.97 95.42

93.79 and 96.06 respectively with PSNR values of 34.34 for LM, WM, RM and PM on
Lena image for window size of 5 × 5. It clearly shows the superior performance for the
median filter attack.

Cropping attack
The performance of LM is better than the other methods for the cropping attack. How-
ever WM, RM and PM also give good and recognizable extracted watermark.
Devi et al. SpringerPlus (2016) 5:1091 Page 17 of 22

Fig. 5  Comparison of different methods: a JPEG compression, b Rotation, c Median filter, d Cropping, e Scal-
ing, f Impulse noise

Scaling attack
NC values of LM, WM, RM and PM for the scaling attack for scale factors of 2  ×  2
and 6 × 6 are 96.70, 94.79, 95.04, 98.07 respectively, and 83.52, 90.50, 91.08 and 93.26
respectively with PSNR values 32.99 dB and 26.65 dB. It clearly shows the superior per-
formance for the scaling attack.

Impulse noise attack

Figure 8 shows the quality of the extracted watermark. NC values of LM, WM, RM and
PM for impulse noise attack for impulse noise ratio of 0.05 are 84.91, 80.22, 82.25 and
Devi et al. SpringerPlus (2016) 5:1091 Page 18 of 22

Fig. 6  Comparison of different methods: a Gaussian noise, b Blurring, c Sharpening and d Gamma correc-
tion attacks

Fig. 7  No attack: a Original watermark, b Encrypted watermark, c Ownership share, d Identification share, e
Superimposed share, f Superimposed share after reduction and g Extracted watermark

97.80 respectively with PSNR values of 23.34 dB for LM, WM and RM and 36.42 dB for
PM, and for impulse noise ratio of 0.15 are 81.71, 74.04, 75.36 and 97.16 respectively with
PSNR values of 18.60 dB for LM, WM and RM and 34.81 dB for PM. It clearly shows the
better performance of the proposed method for impulse noise attack. The better perfor-
mance of PM is due to inbuilt restoration scheme against the impulse noise attack.

Blurring attack
Figure  8 shows the quality of the extracted watermark. LM, WM, RM and PM give
very good performance for blurring attack, and the detected secret images are also not
blurred.
Devi et al. SpringerPlus (2016) 5:1091 Page 19 of 22

Attack LM WM RM PM
JP for 93.75 97.72 95.50 97.09 96.11 97.72 98.04 99.31
Q=50
and
Q=90

RO for 79.78 77.29 84.93 73.19 85.49 73.53 93.82 95.92

=1
and
=3

MF for 94.60 92.62 94.99 93.11 95.14 93.79 97.90 96.06

ws =
3×3
and
5×5
CR 95.53 87.32 87.52 80.00 84.81 78.63 82.61 73.33
for C=
10%
and
60%
SC 96.70 83.52 94.79 90.50 95.04 91.08 97.85 93.55
for F=
2×2
and
6×6
IN for 84.91 81.71 80.22 74.04 82.25 75.36 97.80 97.16
R=.05
and
.15

BL for 99.34 98.14 98.82 96.77 99.19 96.89 99.56 98.85

= 0.4
and
0.6

GN for 84.57 82.93 84.42 74.53 83.74 75.56 96.26 94.77

V=
0.01
and
0.05
SH for 95.19 95.28 86.05 87.84 86.88 88.69 92.08 92.40
= 0.1
and
0.5

GC for 99.56 99.48 98.92 98.85 99.21 99.29 98.51 98.63

G=
0.8 and
1.2

Fig. 8  Detected watermarks by various methods

Gaussian noise attack

Figure 8 shows the quality of the extracted watermark. NC values of LM, WM, RM and
PM for the Guassian noise attack for zero mean and variance of 0.01 are 84.57, 84.42,
83.74 and 96.26 respectively with PSNR values of 24.77  dB for LM, WM and RM and
30.89 dB for PM, and for variance of 0.03 are 82.93, 74.53, 75.56 and 94.77 respectively
with PSNR values of 20.21  dB for LM, WM and RM and 27.08  dB for PM. It clearly
shows the better performance of the proposed method for Gaussian noise attack. The
better performance of PM is due to inbuilt restoration scheme against the Gaussian
noise attack.
 Devi et al. SpringerPlus (2016) 5:1091

Table 6  NC values and  % errors of detection of angles for the rotation attack

Lena
Rotation 5° 10° 15° 20° 25° 30° 35° 40° 45° 50° 55° 60° 65° 70° 75° 80° 85°
Detection 5.15° 10.17° 15.09° 20.10° 25.13° 30.06° 35.11° 39.99° 45.00° 50.00° 54.88° 59.93° 64.86° 69.86° 74.90° 79.82° 84.84°
Error  % −3.02 −1.78 −0.62 −0.50 −0.53 −0.22 −0.32 0.01 0 0 0.20 0.11 0.20 0.19 0.12 0.22 0.17
NC 95.89 95.84 95.89 95.94 93.67 95.97 95.99 95.80 95.80 95.94 95.89 95.99 95.06 95.92 95.94 95.89 95.84
Mandrill
Rotation 5° 10° 15° 20° 25° 30° 35° 40° 45° 50° 55° 60° 65° 70° 75° 80° 85°
Detection 5.11° 10.17° 15.09° 20.10° 25.13° 30.06° 35.11° 39.99° 45° 50° 54.88° 59.93° 64.86° 69.69° 74.90° 79.82° 84.84°
Error  % −3.02 −1.78 −0.62 −0.50 −0.53 −0.22 −0.32 0.01 0 0 0.20 0.11 0.20 0.14 0.12 0.22 0.17
NC 95.50 95.50 95.55 95.48 93.70 93.82 95.43 95.38 93.82 93.67 95.45 95.41 93.79 95.45 95.50 95.48 95.48
Page 20 of 22
Devi et al. SpringerPlus (2016) 5:1091 Page 21 of 22

Sharpening attack
Figure 8 shows the quality of the extracted watermark. LM shows better performance in
comparison with WM, RM and PM for sharpening attack.

Gamma correction attack

For gamma correction attack, all LM, WM, RM and PM give good performance.
False positive detection problem arises in most of the SVD and VC-based algorithms.
Methods proposed by Lou et al. (2007), Rawat and Raman (2012) and Wang and Chen
(2007) suffer from this problem. An unauthorized image can be used to extract or detect
the watermark producing the watermark, though the quality is not good. This means
that anyone who can detect watermark can claim ownership. Our method solves this
false claim by encrypting the watermark prior to the ownership share generation, and it
decrypts at the time of detection.

Robustness against the rotation attack

Table 6 gives the NCs and errors in detection for the rotation angles from 5° to 85° on Lena
and Mandrill images. It shows that the NC values of the proposed algorithm for both images
lie above 93.67 and below 95. Errors in detection for the rotation angles are comparatively
low. The minimum error in magnitude is 0.20 % and the maximum error in magnitude is
3.02 %. It shows that the proposed method is very effective to handle the rotation attack.

Conclusions
The paper describes a new watermarking algorithm based on the shuffled singular value
decomposition and the visual cryptography for copyright protection of digital images
in the DWT domain. The robustness of the proposed method was verified on different
types of images for different attacks. Comparison with the other related VC-based algo-
rithms reveals that the proposed method gives better performance.
Authors’ contributions
BPD is the first author of this paper. She is working as an assistant professor, NIT Meghalaya, India. She is pursuing PhD
at Assam University, Silchar. KMS is the corresponding author of this paper. He is working as an associate professor at NIT
Manipur. SR is the third author of this paper. He is working as a professor at Assam University, Silchar. BPD and KMS have
equal contribution in planning, designing, implementation and testing of this work. Both authors are involved in writing
and revision process of this manuscript. Both authors read and approved the final manuscript.

Author details
1
 National Institute of Technology Meghalaya, Shillong, India. 2 National Institute of Technology Manipur, Imphal, India.
3
 Assam University, Silchar, India.

Acknowledgements
Authors would like to thank to NIT Manipur, Imphal, NIT Meghalaya, Shillong and Assam University, Silchar, India for
technical support during this research work.

Competing interests
The authors declare that they have no competing interests.

Received: 7 January 2016 Accepted: 30 June 2016

References
Abusitta AH (2012) A visual cryptography based digital image copyright protection. J Inf Secur 3:96–104
Benyoussef M, Mabtoul S, Marraki ME, Aboutajdine D (2015) Robust ROI watermarking scheme based on visual cryptog-
raphy: applications on mammograms. J Inf Process Syst 11(4):495–508
Chang CC, Hsiao JY, Yeh JC (2002) A colour image copyright protection scheme based on visual cryptography and
discrete cosine transform. Imaging Sci J 50(3):133–140
Devi et al. SpringerPlus (2016) 5:1091 Page 22 of 22

Chang SS, Lin CH, Chen TH, Lin KS (2014) Robust watermarking for multiple images and users based on visual cryptogra-
phy. In: Pan JS et al (eds) Genetic and evolutionary computing, advances in intelligent systems and computing, vol
238. Springer International Publishing, Switzerland, pp 175–181
Chen TH, Chang CC, Wu CS, Lou DC (2009) On the security of a copyright protection scheme based on visual cryptogra-
phy. Comput Stand Interfaces 31(1):1–5
“Cipr”. www.cipr.rpi.edu/resource/stills/misc1.html. Last accessed 12 Jan 2016
Cox IJ, Miller ML (2002) The first 50 years electronic watermarking. Eurasip J Appl Signal Process 2:126–132
Fu C, Meng WH, Zhan YF, Zhu ZL, Lau FCM, Tse CK, Ma HF (2013) An efficient and secure medical image protection
scheme based on chaotic map. Comput Biol Med 43(8):1000–1010
Guo JM, Prasetyo H (2014) False-positive-free SVD-based image watermarking. J Vis Commun Image Represent
25(5):1149–1163
Hartung F, Kutter M (1999) Multimedia watermarking techniques. Proc IEEE 87(7):1079–1107
Hassan MA, Khalili MA (2005) Self watermarking based on visual cryptography. World Acad Sci Eng Technol 8(1):159–162
“Hlevkin”. www.hlevkin.com/TestImages. Last accessed 12 Jan 2016
Hossaini AEAE, Aroussi ME, Jamali K (2016) A new robust blind copyright protection scheme based on visual cryptogra-
phy and steerable pyramid. Int J Netw Secur 18(2):250–262
Hsu CS, Hou YC (2005) Copyright protection scheme for digital images using visual cryptography and sampling methods.
Opt Eng 44(7):077003-10
Hwang RJ (2000) A digital image copyright protection scheme based on visual cryptography. Tamkang J Sci Eng
3(2):97–106
“Imagecompression”. http://imagecompression.info/test_images. Last accessed 12 Jan 2016
Jin X, Kim JW (2012) A secure image watermarking using visual cryptography. In: Yeo S-S et al (eds) Lecture notes in
electrical engineering, computer science and its applications, vol 203, pp 179–187
Keshavarzian R, Aghagolzadeh A (2016) ROI based robust and secure image watermarking using DWT and Arnold map.
Int J Electron Commun 70(3):278–288
Liu RZ, Tan TN (2002) An SVD-based watermarking scheme for protecting rightful ownership. IEEE Trans Multimed
4(1):121–126
Lou DC, Tso HK, Liu JL (2007) A copyright protection scheme for digital images using visual cryptography technique.
Comput Stand Interfaces 29(1):125–131
Mallat SG (1989) A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans Pattern Anal
Mach Intell 11(7):674–693
Naor M, Shamir A (1995) Visual cryptography. In: Proceedings of advances cryptology. EUROCRYPT94, LNCS950, 1-12,
Springer, Berlin
“Pages.physics”. http://pages.physics.cornell.edu/~sethna/teaching/562_S03/HW/pset02_dir/catmap.pdf. Last accessed
28 April 2016
Petticolas F (1999) Information hiding techniques for steganography and digital watermarking. Artech House Inc, Stefen
Katzenbeisser
Ranade A, Mahabalarao SS, Kale S (2007) A variation on SVD based compression. Image Vis Comput 25(6):771–777
Rani A, Bhullar AK, Dangal D, Kumar S (2015) A zero-watermarking scheme using discrete wavelet transform. Proc Com-
put Sci 70:603–609
Rawat S, Raman B (2012) A publicly verifiable lossless watermarking scheme for copyright protection and ownership
assertion. AEU 66(11):955–962
Ruanaidh JJKÒ, Pun T (1998) Rotation, scale and translation invariant spread spectrum digital image watermarking. Signal
Process 66(3):303–317
Singh KM (2009) Dual watermarking scheme for copyright protection. Int J Comput Sci Eng Syst 3(2):99–106
“Sipi”. http://sipi.usc.edu/database. Last accessed 12 Jan 2016
Su Q, Niu Y, Zhao Y, Pang S, Liu X (2013) A dual color images watermarking scheme based on the optimized compensa-
tion of singular value decomposition. AEU 67(8):652–664
Wang MS, Chen WC (2007) Digital image copyright protection scheme based on visual cryptography and singular value
decomposition. Opt Imaging 46(6):067006-8
Wang Y, Wong KW, Liao X, Xiang T, Chen G (2009) A chaos-based image encryption algorithm with variable control
parameters. Chaos Solitons Fractals 41(4):1773–1783