See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/276113170

LSB Based Quantum Image Steganography Algorithm

Article  in  International Journal of Theoretical Physics · April 2015

DOI: 10.1007/s10773-015-2640-0

CITATIONS READS

89 905

3 authors, including:

Nan Jiang Luo Wang

Beijing University of Technology Xi'an Jiaotong University
40 PUBLICATIONS   588 CITATIONS    11 PUBLICATIONS   449 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Quantum Image Processing View project

All content following this page was uploaded by Luo Wang on 12 April 2016.

The user has requested enhancement of the downloaded file.

Int J Theor Phys
DOI 10.1007/s10773-015-2640-0

LSB Based Quantum Image Steganography Algorithm

Nan Jiang1 · Na Zhao1 · Luo Wang1

Received: 3 October 2014 / Accepted: 30 March 2015

© Springer Science+Business Media New York 2015

Abstract Quantum steganography is the technique which hides a secret message into quan-
tum covers such as quantum images. In this paper, two blind LSB steganography algorithms
in the form of quantum circuits are proposed based on the novel enhanced quantum repre-
sentation (NEQR) for quantum images. One algorithm is plain LSB which uses the message
bits to substitute for the pixels’ LSB directly. The other is block LSB which embeds a mes-
sage bit into a number of pixels that belong to one image block. The extracting circuits can
regain the secret message only according to the stego cover. Analysis and simulation-based
experimental results demonstrate that the invisibility is good, and the balance between the
capacity and the robustness can be adjusted according to the needs of applications.

Keywords Quantum image processing · Least significant bit · Quantum information

hiding · Quantum steganography

1 Introduction

To satisfy the need for private and personal communication, various techniques are
proposed. One such technique is steganography that is the art of hiding the fact that com-
munication is taking place, by hiding a message in other information (such as image, audio
and video) [1]. The least significant bit (LSB) steganography is the simplest but the most
widely used technique in which least significant bit of the image is replaced with message
bit. In this paper, we give two LSB based quantum image steganography algorithms.

This work is supported by the Beijing Municipal Education Commission Science and Technology
Development Plan under Grants No. KM201310005021, the National Scholarship under Grants No.
201406545034, and the Graduate Technology Fund of BJUT under Grants No. YKJ-2014-10775.

 Nan Jiang
[email protected]

1 College of Computer, Beijing University of Technology, Beijing 100124, China

 Int J Theor Phys

The first task of quantum image processing (QIP) is to give the representation of quantum
images. At present, two representing methods are widely used. The flexible representation
of quantum images (FRQI) [2] uses 2n + 1 qubits to represent a 2n × 2n image. It provides
a fast method to geometrically transform quantum images. The novel enhanced quantum
representation of digital images (NEQR) [3] improves the color information of FRQI from
1 qubit to q qubits, which makes complex and elaborate color operations can be performed
conveniently.
Based on FRQI, a number of researchers give quantum image steganography or water-
marking algorithms. Iliyasu proposes a quantum image watermarking and authentication
algorithm which uses flip and transposition operations to construct a watermark map and
give the watermark circuit based on the map [4]. However, the defects of the algorithm
include: (1) it uses classical computer to gain the watermark map, and (2) it only has
authentication function while the watermark is not really embedded into the cover.
Zhang gives two similar watermark schemes. One embeds watermark into the quantum
Fourier transform (QFT) coefficients, the other embeds watermark into the Taylor series
[5, 6]. Song also advances two similar watermark schemes. One embeds watermark into
the quantum wavelet transform (QWT) coefficients, the other embeds watermark into the
Hadamar coefficients [7, 8]. However, the four schemes are all nonblind.
In this paper, we propose two LSB steganography algorithms based on NEQR. One algo-
rithm is plain LSB which uses the message bits to substitute for the pixels’ LSB directly.
The other is block LSB which embeds a message bit into a number of pixels that belong
to one image block. The extracting circuits can regain the secret message only according to
the stego cover.
The rest of the paper is organized as follows. A brief background on the NEQR repre-
sentation, LSB image steganography, and other related knowledge is presented in Section 2.
The plain LSB network is discussed in Section 3. The block LSB network is discussed in
Section 4. This is followed in Section 5 by the simulations and analysis. Finally, a conclusion
is given in Section 6.

2 Preliminaries

2.1 The Novel Enhanced Quantum Representation (NEQR)

In 2013, NEQR was proposed in [3] which is the latest representation which uses the basis
state of a qubit sequence to store the gray-scale value of every pixel for the first time.
Meanwhile, more quantum image operations related to gray-scale information in the image
can be performed conveniently. According to NEQR, a quantum image can be written as
the form shown below.
22n −1
1 
|I  = n |ci  ⊗ |i,
2
i=0
q−1
|ci  = |ci · · · ci1 ci0 , cik ∈ {0, 1}, k = q − 1, · · · , 1, 0, i = 0, 1, · · · , 22n − 1. (1)
q−1
where the binary sequence ci · · · ci1 ci0 encodes the color value (or gray value), and the
color range is 2q ; |i, for i = 0, 1, · · · , 22n − 1, are 22n dimension computational basis
quantum states; and ⊗ denote the Kronecker product. The two parts in the NEQR: |ci 
and |i encode information about the colors and their corresponding positions in the image,
respectively. The size of the quantum image is 2n × 2n .
Int J Theor Phys

Fig. 1 A simple image and its

NEQR state

The location information |i includes two parts: the vertical and horizontal coordinates.
|i = |y|x = |yn−1 yn−2 · · · y0 |xn−1 xn−2 · · · x0 , |yj |xj  ∈ {0, 1}. (2)
It encodes the first n-qubit yn−1 , yn−2 , · · · , y0 along the vertical location and the second n-
qubit xn−1 , xn−2 , · · · , x0 along the horizontal axis. Therefore, NEQR needs q + 2n qubits
to represent a 2n × 2n image with gray range 2q .
An example of a 2 × 2 image and its NEQR representation are shown in Fig. 1. In this
example, because the gray scale ranges between 0 and 255, 8 qubits are needed in NEQR to
store the gray-scale information for the pixels.

1
|I  = [|10011001 ⊗ |00 + |01100110 ⊗ |01 + |00110011 ⊗ |10 + |11001100 ⊗ |11]
2

2.2 The Classical LSB Image Steganography Method

Steganography is a branch of information hiding (the other branch is digital watermarking)

which hides a message into a cover. It is a kind of subliminal channel and provides secret
communication so that intended hacker or attacker unable to detect the presence of the
message.
The LSB steganography, first proposed by A.Z. Tirkel [9] in 1993, is a fundamental
and simple data hiding method. After that, a number of researchers have proposed many
improved LSB algorithms [10–13]. However, their principles are alike: use message bits to
substitute the least significant bits of the cover.
In general, the cover is a 24-bit or 8-bit image. Taking the latter for example, it has
28 = 256 colors (or gray scales). The least significant bit of one of the pixels is shown in
Fig. 2. If the message bit is “1”, we only need to change the least significant bit from “0” to
“1”, i.e., color 154 is changed into color 155. The receiver can get the message by simply
reading the least significate bit.
The advantages of LSB steganography are that it is simple to understand, easy to imple-
ment, and it produces stego image (the embedded cover image) that is almost similar to
cover image and its visual infidelity cannot be judged by naked eyes.
 Int J Theor Phys

Fig. 2 An example of the LSB of a pixel

2.3 Quantum Counter

Section 2.3–2.5 introduce three quantum circuits that will be used in Section 4.
Ma and etc [14] give a quantum counter circuit shown in Fig. 3. |b is the input qubit and
b ∈ {0, 1}. |an−1 · · · a1 a0  is the counter with initial value |0 · · · 00. If the input qubit |b is
|1, an−1 · · · a1 a0 increase by 1; otherwise, an−1 · · · a1 a0 remain unchanged.

2.4 Quantum Comparator

Wang and etc [15] give a quantum comparator circuit shown in Fig. 4. The comparator
compares a and b, where |a = |an−1 · · · a1 a0  and |b = |bn−1 · · · b1 b0 , ai , bi ∈ {0, 1},
i = n − 1, · · · , 1, 0. Qubits |c1  and |c0  are outputs.

If c1 c0 = 10, then a > b;

if c1 c0 = 01, then a < b;
if c1 c0 = 00, then a = b.

2.5 Quantum Image Scrambling

The main aim of image scrambling, which is usually used as the preprocessing in many
image processing algorithms, is to transform a meaningful image into a disordered image
by permutating the pixels into new positions. Jiang and etc [16] give a quantum Hilbert
image scrambling algorithm. This quantum algorithm defines three basic modules and uses
them to construct three complex modules which are assembled as the quantum scrambling
algorithm as shown in Figs. 5–7.
Because the circuits are complex, we do not describe them in detail and will give an
example in Section 4.5. If readers want to get more information about it, please reference
[16].

Fig. 3 Quantum counter

Int J Theor Phys

Fig. 4 Quantum comparator

3 Plain LSB Steganography Network

As introduced in Section 2.2, plain LSB steganography uses message bits to substitute the
least significant bits of the cover directly.
Assume the cover image is a 2n × 2n image |I  with gray range 2q and the message is a
binary 2n × 2n image |M. The NEQR representations of |I  and |M are shown in (1) and
(3) respectively.

2 −12n
1 
|M = n |mi  ⊗ |i, mi ∈ {0, 1}, i = 0, 1, · · · , 22n − 1. (3)
2
i=0

The circuit of plain LSB embedding is shown in Fig. 8a. 2n CNOT gates are used to
test whether the position information of |I  and |M is the same. If they are equal, the
position information of |M is changed into |00 · · · 0. Hence, under the control of them,
we swap the least significant bit |ci0  of |I  and the message bit |mi  to get the stego
image |I  .
The extracting circuit is shown in Fig. 8b. 2n Hadamard gates are used to turn
the initial all |0 state to an empty binary image. Then, as in the embedding cir-
cuit, 2n CNOT gates are used. If the position information of |I  and |M is equal,
we swap the least significant bit |ci0  of |I  and the message bit |mi  to get the
message.
Taking the 128 × 128 cover image “Lena” for example. The message needed to be trans-
mitted secretly is a 128 × 128 binary image. Then the stego image is presented in Fig. 9c.
 Int J Theor Phys

Fig. 5 Three basic modules

Fig. 6 Three complex modules

Fig. 7 The integrated Hilbert

image scrambling quantum
circuit
Int J Theor Phys

Fig. 8 Plain LSB

4 Block LSB Steganography Network

Plain LSB steganography network is simple, but its robustness is poor. Many common image
processing methods, such as filtering, noising, compressing and so on, can easily remove
the hidden message. Moreover, malicious people can easily find out and extract the message
from the cover image by reviewing the LSB plane which is a binary image consisted by all
pixels’ LSB.
Therefore, block LSB steganography network is proposed in this paper to improve
the robustness and undetectability of LSB scheme. This method divides the cover image
into blocks. Each block, instead of each pixel, hides one message bit. In fact, plain LSB
steganography can be seen as a special case of block LSB version whose every block only
accommodate one pixel.
For the convenience of description in the following, we first give the concept of control
value and discuss how to divide the cover image into blocks.

Fig. 9 An example for plain LSB steganography

 Int J Theor Phys

Fig. 10 An example for control value

4.1 Control Value of Control Quantum Circuits

Quantum circuits have two control conditions: ◦ and •. If the control condition of a control
qubit is ◦, the target circuit is effective only if the value of the control qubit is 0; otherwise,
the target circuit do nothing. If the control condition is •, the target circuit is effective only
if the control qubit is 1.
Naturally, we can use a 1-bit binary label to represent the two control conditions: ◦ as
0 and • as 1 respectively. Concatenating all the binary labels from top to bottom, it can
be viewed as a binary number which defined as the control value of the control circuit.
Figure 10 gives an example. The circuit has 5 control qubit and the control value is 19.

4.2 Image Partition

The quantum image representations NEQR and FRQI provide a easy way to partition a
image into blocks. If part of the location information is restricted as a specific value, then
some of pixels will be selected. Take the simple image shown in Fig. 1 as an example, if
|y0  is restricted as |1, the bottom two pixels will be selected.
In our scheme, the 2n × 2n cover image is needed to be divided into 2n−p1 × 2n−p2
blocks sized 2p1 × 2p2 , where p1 , p2 ∈ {0, 1, · · · , n} and p1 + p2 = p. We split the
location information |y and |x of cover image into two parts: |yn−1 yn−2 · · · yp1  and
|yp1 −1 · · · y1 y0 , and |xn−1 xn−2 · · · xp2  and |xp2 −1 · · · x1 x0 , where |yn−1 yn−2 · · · yp1  and
|xn−1 xn−2 · · · xp2  are called inter-block coordinates because if they are restricted as a spe-
cific value, a number of pixels, i.e., a image block, will be selected. We define the block as
Bk,l , where
|k = |yn−1 yn−2 · · · yp1 , |l = |xn−1 xn−2 · · · xp2 .
The remainder location information |yp1 −1 · · · y1 y0  and |xp2 −1 · · · x1 x0  are called intra-
block coordinates.

4.3 Block Embedding Procedure

Block embedding procedure divides the cover image into blocks. Each block, hides one
message bit. Assume that the message is a 2n−p1 × 2n−p2 binary image mk,l , where mk,l ∈
{0, 1}, k = yn−1 yn−2 · · · yp1 , l = xn−1 xn−2 · · · xp2 . The block embedding circuit is shown
in Fig. 11.
The concrete procedure of embedding is as following.
Int J Theor Phys

Fig. 11 Block LSB embedding circuit

Step 1 Scrambling. The cover image is scrambled in order to enhance the undetectability
in our scheme. Figure 12 shows the effect of scrambling. Without it, the LSB plane
of the stego image is the message. Malicious people can find out the message
effortlessly. On the other hand, scrambling makes the LSB plane like random that
will not attract malicious people’s attention.
Because scrambling is the procedure of permutating pixels into new positions,
the circuit about this step only works at position qubits |y and |x.

Fig. 12 Scrambling enhance the undetectability

 Int J Theor Phys

Fig. 13 Block LSB extracting circuit

Step 2 Embedding. If the position information |yn−1 yn−2 · · · yp1 xn−1 xn−2 · · · xp2  of |I 
and is equal to the position information of |M, embedding swaps the least
significant bit |cyx
0  and the message bit |m .
kl
Step 3 Inverse scrambling. This step uses the inverse scrambling circuit to recover from
the scrambled image.

4.4 Block Extracting Procedure

In general steganography scheme, the extracting procedure is only available to the particular
receiver, which ensures the security of the embedded message. Our strategy is no exception.
In our scheme, the receiver only uses the stego image to extract the message. It is a blind
strategy.
After embedding every message bit 2p times, the stego image may be attacked mali-
ciously or unawares, which may change some LSBs’ value. This will cause that the sum
of all the pixels’ LSB belonging to one block is not 0 or 2p , but a value between the two
numbers. The solution to determine the extracted message bit is 0 or 1 according to the sum
value is setting a threshold. If the sum is greater than or equal to the threshold, the message
bit is 1; otherwise, the message bit is 0. The block extracting circuit is shown in Fig. 13.
The concrete extracting procedure is as following.
Step 1 Scrambling. It is the same as Step 1 of embedding procedure.
Step 2 Dividing. Unlike embedding that has two inputs, extracting use the stego image
as the only input. It can not take advantage of subscript of message to partition
the cover. In order to solve the problem, we use a control circuit — Dividing, to
partition stego image into blocks.
Dividing uses all the inter-block coordinates as control qubit. Because it divides
the cover into 2n−p1 × 2n−p2 = 22n−p1 −p2 blocks, it has 22n−p1 −p2 control layers
(one control layer corresponding to one block). If we list the inter-block coordi-
nates from top to bottom in order of yn−1 yn−2 · · · yp1 xn−1 xn−2 · · · xp2 , the control
values of all the control layers form left to right are 0, 1, · · · , 22n−p1 −p2 − 1.
Step 3 Counting. This step use quantum counting circuits to sum all the pixels’ LSB that
belong to one block. We use the counter circuit proposed in [14] to achieve the
aim. It contains 22n−p1 −p2 counters denoted as ayn−1 ···yp1 xn−1 ···xp2 and the counting
number are also represented as ayn−1 ···yp1 xn−1 ···xp2 .
Int J Theor Phys

Fig. 14 The example cover and

message. The number in the
pixels is their gray value. The
cover image is divided into 4 × 2
blocks sized 1 × 2 represented by
white thick lines

Step 4 Comparing. Because each block has 2p pixels, the counting number gained in Step
3 should be compared with 2p−1 which is the threshold T that we set. Note that
if p > 0, each block has even pixels. If half of the pixels’ LSBs are 0 and the
other half are 1, our scheme set the extracted message bit as 1 because in general,
the proportion of white pixels is greater than that of black pixels in binary image.
The quantum comparator circuit our scheme used is proposed in [15]. That is to
say, if ayn−1 ···yp1 xn−1 ···xp2 ≥ T , the extracted message is 1; otherwise, the extracted
message is 0.

4.5 An Example

An example suffices to demonstrate the practicable features and to illustrate the effective-
ness of the proposed algorithm. To accomplish this, let us consider a simple 4 × 4 cover
image and a 8 bit message 00110110 which is the ASCII code of character “6”. We divide
the cover image into 8 blocks sized 1 × 2 and rearrange the message into a 4 × 2 binary
image as shown in Fig. 14. In this example, n = 2, p1 = 0, p2 = 1.
The block LSB embedding circuit is given in Fig. 15. It has 3 parts which correspond
to the 3 steps in Section 4.3. The first part and the last part accomplish Hilbert image
scrambling and its inverse operation which is referenced from [16].
The block LSB extracting circuit is given in Fig. 16. It has 4 parts which correspond to
the 4 steps in Section 4.4. The Hilbert scrambling part is the same as the one of embedding.
The Dividing part is a control circuit that determines cy01 y0 x1 x0 enters which counter. For
0
example, if the control value is 000, c000x 0
swaps with the first auxiliary qubit |0, i.e., it
enters the first counter.

Fig. 15 Block LSB embedding circuit of the example

 Int J Theor Phys

Fig. 16 Block LSB extracting circuit of the example

The Counting part is consisted of 2n−p1 × 2n−p2 = 8 counters which correspond to

the 8 blocks. Each counter ay1 y0 x1 sums the LSB of pixels of block By1 y0 x1 . Because each
block has 2 pixels, the maximum value of ay1 y0 x1 is 2 and 2 qubits is enough, |ay1 y0 x1  =
|ay11 y0 x1 ay01 y0 x1 .
The Comparing part has 2n−p1 × 2n−p2 = 8 comparators which respectively compare
ay1 y0 x1 with the threshold 2p−1 = (01)2 . As stated in Section 2.4,

if ay1 y0 x1 > 01, the bottom two qubits of each comparator is 10;
if ay1 y0 x1 = 01, the bottom two qubits of each comparator is 00;
if ay1 y0 x1 < 01, the bottom two qubits of each comparator is 01.

Hence, we only need to invert the bottom qubit to get message qubit my1 y0 x1 .

5 Simulations and Analysis

Due to the absence of the physical quantum hardware to implement our proposed protocol,
we just make the simulations of the quantum circuits on a classical computer equipped with
software Matlab.
Information hiding schemes including steganography and watermarking, have three
key properties: invisibility, capacity, and robustness. Invisibility represents the similarity
between the original covers and the stego/watermarked covers. Capacity measures how
many message bits one cover pixel can accommodate. Robustness shows the message’s abil-
ity to survive in attacks. We analyze the proposed scheme from the three properties. The
cover images and the message images used in our simulations are shown in Fig. 17.
Int J Theor Phys

Fig. 17 The cover images and the message images used in our simulations. (a)-(f) are cover images. (g)-(h)
are message images

5.1 Invisibility

The peak-signal-to-noise ratio (PSNR) is one of the most used quantity for comparing the
fidelity of a stego image with its original version. It is most easily defined via the mean
squared error (MSE), which for two m × n images I and J is defined as

1   
m−1 n−1
MSE = (I (i, j ) − J (i, j ))2 ,
mn
i=0 j =0

PSNR is defined as
 
MAXI
PSNR = 20log10 √ ,
MSE
where, MAXI is the maximum pixel value of image I . In our scheme, I and J corresponds
to the original image and the stego image respectively.
Because it is a LSB based algorithm, if a pixels’ LSB is the same as the mes-
sage bit it accommodates, (I (i, j ) − J (i, j ))2 = 0; otherwise (I (i, j ) − J (i, j ))2 =
1. That is to say, MSE can be seen as a ratio of the number of pixels that
changed in embedding to the total number of pixels. Therefore, MSE is indepen-
dent with the size of the cover images and the size of blocks. Assume MSE =
0.5, i.e., half of the pixels is changed, and MAXI ≈ 255 for 8-bit natural image,
then
 
255
PSNR = 20log10 √ = 51.1411.
0.5
Even if every pixel is changed, i.e., MSE = 1, PSNR still can reach 48.1308.
It indicates that the invisibility of our scheme is perfect. Experimental results prove
the conclusion. Figure 18 gives the visual effects and Table 1 provides the PSNR
values.
 Int J Theor Phys

Fig. 18 The visual effects. The first row is original covers. The second row is stego covers

5.2 Capacity

Steganography capacity is defined as ratio between the number of message bits and the
number of cover pixels, which has directly relation with the number of blocks in our scheme
because one block embeds one message bits. Hence, the proposed algorithm’s capacity is

The number of message bits 22n−p1 −p2 1

C= 
= 2n
= p (bit/pixel)
The number of cover image s pixels 2 2
It can be presented in Fig. 19.

5.3 Robustness

The Bit Error Rate (BER) and the Normalized Correlation (NC) are two of the most used
quantity for robustness. Assume that the original message is M and the extracted message
is M  , and use N (A) denote the number of elements in A, then BER defined as
2n−p1 −1 2n−p2 −1
 k=0 l=0 (mk,l − mk,l )2 N (mk,l
= mk,l )
BER(M, M ) = = ,
22n−p1 −p2 N (M)
and NC defined as
2n−p1 −1 2n−p2 −1
 k=0 l=0 (2mk,l − 1) × (2mk,l − 1)
NC(M, M ) =
22n−p1 −p2
N (mk,l = mk,l ) − N (mk,l
= mk,l )
= = 1 − 2BER(M, M  ).
N (M)

R = Pr{BER(M, M  ) < α}, α ∈ [0, 1].

where, Pr denotes probability, α is the acceptable error that is given artificially.

Table 1 PSNR (in dB) values for different message images

Message Lena Baboon Barbara Cameraman Goldhill Peppers

Message 1 50.8426 50.3786 50.8852 51.6169 50.9598 50.6278

Message 2 50.3511 49.9285 50.4111 51.1554 50.4888 50.1073
Int J Theor Phys

Fig. 19 Variation of capacity with p

In order to get R, we use b, b ∈ [0, 1], to denote the attack strength, which is the proba-
bility that a pixel’s LSB of a cover image will be changed after a kind of attack. The higher
the strength of the attack, b is greater. Then, according to the proposed extracting procedure,
if mk,l = 0 and the counting number ak,l ≤ 2p−1 − 1, i.e., less than or equal to 2p−1 − 1
pixels’ LSB in block Bk,l are inversed, mk,l = mk,l ; otherwise, mk,l
= mk,l .
if mk,l = 1 and the counting number ak,l ≥ 2p−1 , i.e., less than or equal to 2p−1 pixels’
LSB in block Bk,l are inversed, mk,l = mk,l ; otherwise, mk,l
= mk,l .
Hence, the probability of a message bit is extracted correctly is
⎧
⎪1−b
⎪
⎪   for plain LSB procedure
⎪
⎨ 2p−1 −1 2p p −r
r=0 b (1 − b)
r 2 for block LSB procedure and mk,l = 0
 p r
⎪
⎪ 2p−1 2
⎪
⎪ br (1 − b)2 −r
p
⎩ r=0 for block LSB procedure and mk,l = 1
r
 p
2
where, is the combination operation.
r
Assume the probability distribution of message is
 
0 1
, t ∈ [0, 1],
t 1−t
the average probability for block LSB procedure is
−1  2p 
2p−1 p−1 
2 
2p
br (1 − b)2 −r + (1 − t) br (1 − b)2 −r
p p
P =t
r r
r=0 r=0

  2p 
2p−1
p −r

2p

p−1 p−1
= br (1 − b)2 −t b2 (1 − b)2
r 2p−1
r=0
Then,
R = Pr{BER(M, M  ) < α} = Pr{N (mk,l
= mk,l ) < α × 22n−p }.
 Int J Theor Phys

Fig. 20 Robustness

Denote α × 22n−p as , where · is the floor operation, then

Θ  2n−p 
2
R = Pr{N (mk,l
= mk,l ) ≤ Θ} =
2n−p −r
(1 − P )r P 2 .
r
r=0
It can be presented in Fig. 20. In this figure, n = 4, p = 3 and α = 0.3.
From Figs. 19 and 20b, we can see that the capacity and the robustness are both concern
with p, i.e., the size of the block. With the increase of p, the capacity is smaller and the
robustness is stronger. That is to say, the balance between the capacity and the robustness
can be adjusted by given a suitable p.

6 Conclusions

In this paper, we present two LSB based steganography algorithms for quantum images,
which at least have two advantages: (1) they are absolutely blind. The extracting procedure
does not need the original cover or the original message. (2) the whole procedure can be
accomplished in quantum computers and it does not need the help of classical computers
or human beings. Analysis and simulation-based experimental results demonstrate that the
invisibility is good, and the balance between the capacity and the robustness can be adjusted
according to the needs of applications.

References

1. Gupta, S., Goyal, A., Bhushan, B.: Information hiding using least significant bit steganography and
cryptography. Int. J. Educ. Comput. Sci. 6, 27–34 (2012)
2. Le, P.Q., Dong, F., Hirota, K.: A flexible representation of quantum images for polynomial preparation,
image compression and processing operations. Quantum Inf. Process. 10(1), 63–84 (2011)
3. Zhang, Y., Lu, K., Gao, Y.H., et al.: NEQR: a novel enhanced quantum representation of digital images.
Quantum Inf. Process. 12(12), 2833–2860 (2013)
4. Iliyasu, A.M., Le, P.Q., Dong, F., Hirota, K.: Watermarking and authentication of quantum images based
on restricted geometric transformations. Inf. Sci. 186, 126–149 (2012)
5. Zhang, W.W., Gao, F., Liu, B., et al.: A watermark strategy for quantum images based on quantum
Fourier transform. Quantum Inf. Process. 12(4), 793–803 (2013)
Int J Theor Phys

6. Zhang, W.W., Gao, F., Liu, B., et al.: A quantum watermark protocol. Int. J. Theor. Phys. 52, 504–513
(2013)
7. Song, X.H., Wang, S., Liu, S., Abd El-Latif, A.A., Niu X.M.: A dynamic watermarking scheme for
quantum images using quantum wavelet transform. Quantum Inf. Process. 12(12), 3689–3706 (2013)
8. Song, X.H., Wang, S., Liu, S., Abd El-Latif, A.A., Niu, X.M.: Dynamic watermarking scheme for
quantum images based on Hadamard transform (2014). Multimedia Systems, published online
9. Tirkel, A.Z., Rankin, G.A., VanSchyndel, R.M., et al.: Electronic watermark. Proceedings of digital
image computing: techniques and applications, pp. 666–672. Macquarie University (1993)
10. Chan, C.K., Cheng, L.M.: Hiding data in image by simple LSB substitution. Pattern Recogn. 37(3), 469–
474 (2004)
11. Dey, S., Abraham, A., Sanyal, S.: An LSB data hiding technique using prime numbers. In: Third
international symposium on information assurance and security, pp. 101–108 (2007)
12. Ghazanfari, K., Ghaemmaghami, S., Khosravi, S.R.: LSB++: an improvement to LSB+ steganography.
In: IEEE Region 10 Conference TENCON, pp. 364–368 (2011)
13. Akhtar, N., Johri, P., Khan, S.: Enhancing the security and quality of LSB based image steganography. In:
5th International Conference on Computational Intelligence and Communication Networks, pp. 385–390
(2013)
14. Ma, L., Lu, J.: Construction of controlled quantum counter. Chinese J. Quantum Electron. 20(1), 47–50
(2003)
15. Wang, D., Liu, Z.H., Zhu, W.N., Li, S.Z.: Design of quantum comparator based on extended general
Toffoli gates with multiple targets. Comput. Sci. 39(9), 302–306 (2012)
16. Jiang, N., Wang, L., Wu, W.Y.: Quantum Hilbert image scrambling. International Journal of Theoretical
Physics, accepted. doi:10.1007/s10773-014-2046-4 (2014)

View publication stats